Glossary

A doji star whose doji is cut off from both neighbors by full-range gaps.

The doji's whole range sits beyond the first candle's range, and the third candle's whole range sits beyond the doji's, with the third close passing the first body's midpoint. Every abandoned baby is also a doji star. Full-range gaps generally need a break in trading, which is why the pattern is rare on continuously traded charts.

See also: Doji star, Gap, Morning star / evening star.

A label for heavy trading accompanied by unusually little net price progress.

The observable evidence is high volume with little movement in the direction being tested. Traders often call that absorption, but ordinary price-and-volume data cannot identify one large participant, separate initiating from passive interest perfectly, or reveal anyone's future intention. Treat the label as a description of price response to traded quantity, not as proof of who acted or what happens next.

See also: Churn, Climax, Volume.

A cumulative volume line weighted by where the close sits inside each period's range.

The usual money-flow multiplier places the close within the high–low range, multiplies that value by the period's volume, and adds the result cumulatively. A close near the high contributes a positive value, one near the low a negative value, and one near the midpoint about zero. These formula labels do not measure net cash flow, distinguish buyer from seller volume, or identify accumulation by a holder group.

See also: On-balance volume, Chaikin Money Flow, Volume.

The share of predictions a model got right, which is not the same as profit.

Profit depends on payoff size, timing, position sizing, and costs, which classification accuracy omits. A 40% hit rate with gains three times losses may have positive gross expectancy; a 90% hit rate can still have negative expectancy if rare losses are large. Neither is enough to judge a model without a representative sample, calibration, dependence, and net costs.

See also: Class imbalance, Expectancy, Risk-to-reward ratio, Profit factor.

Historical prices restated for splits and dividends.

Splits change the per-share price and count without changing proportional ownership at the effective moment; cash dividends transfer value out of the company. Adjusted data restates earlier prices under a provider's method so returns and chart comparisons are more consistent. The adjustment method must be checked because price-only and total-return series are not the same.

See also: Stock split, Dividend.

A fill being more likely when subsequent price movement is unfavorable to the liquidity provider.

A resting order is filled only when opposing flow arrives, so fill events can be correlated with later adverse moves. The counterparty need not possess private information; urgency, inventory, latency, news, and random flow can create the same observation. Adverse selection is one cost of liquidity provision, while spreads also compensate for order processing, inventory, risk, competition, and fees.

See also: Passive order, Queue position, Market maker, Spread.

An order that crosses the spread to trade immediately.

A market order or marketable limit order seeks available opposing liquidity and commonly incurs spread, fees, and possible market impact. A limit supplies a worst acceptable price but not complete execution; a market order prioritizes execution but supplies no price bound, and either can be delayed by market conditions. The trade-off is execution probability and speed versus price control and cost.

See also: Passive order, Spread, Slippage, Market impact.

Gathering more evidence past the point where it changes the decision.

Beyond a certain point extra information stops improving a judgment and starts providing material for whichever answer you were already inclined toward. In trading it usually looks like adding indicators until one of them agrees. The practical fix is not more confidence, it is a written checklist with a fixed number of conditions, so the decision is made when the list is satisfied rather than when the feeling arrives.

See also: Trade checklist, Strategy hopping, Confirmation bias.

A VWAP measured from a chosen event rather than from the session open.

Anchoring to an earnings release, gap, breakout, or selected high measures the volume-weighted transaction price since that chosen point. It does not measure the cost basis of current buyers or holders. Different anchors produce different lines, and selecting an anchor after seeing the result creates data-snooping risk; any predictive rule still needs an exact definition and out-of-sample evidence.

See also: VWAP, Overfitting, Volume.

Letting an arbitrary reference price govern a decision it has no bearing on.

Your entry price, the year's high, and a round number are all anchors, and the market has no knowledge of any of them. "I will sell when it gets back to what I paid" is anchoring in its purest form: the price you paid is a fact about your history rather than about the instrument, and it cannot make a level more likely to be reached. Every anchor is defeated the same way, by deciding exits from structure before entry.

See also: Invalidation, Sunk cost, Support and resistance.

The audited yearly account of a company's finances and business.

An annual report commonly contains audited financial statements, management discussion, business information, risks, and governance disclosures under applicable rules. It is a primary source but remains historical, uses estimates, and includes unaudited narrative sections. Read the auditor's opinion, footnotes, later filings, and amendments rather than treating the report as a guarantee.

See also: Regulatory filing, Auditor, Free cash flow.

Increasing size after wins and reducing it after losses.

Fixed-fraction sizing has an anti-martingale shape because the same percentage of lower equity produces a smaller position and of higher equity a larger one. Other risk frameworks do not necessarily increase after wins. Reducing size can slow recovery and limit subsequent loss, but it does not create a guaranteed drawdown floor because gaps, leverage, and correlated exposures remain.

See also: Martingale, Pyramiding, Risk per trade.

The compression at the tip of a converging pattern, where the range has narrowed to little.

The apex is the intersection of the lines used to define a triangle and marks where that drawn range converges. Price may break before reaching it, drift through it, or make the drawing invalid. The geometry can define candidate boundaries but supplies no directional forecast or uniquely correct invalidation.

See also: Triangle, Consolidation, Volatility.

The simple average of period returns, ignoring compounding.

Add the period returns and divide by how many there were. It is the right number for answering "what does a typical period look like", and the wrong number for answering "what did I end up with", because money compounds and averages do not. It is always greater than or equal to the geometric return, and the gap widens with volatility, which means quoting it for a volatile strategy systematically overstates what an investor actually received.

See also: Geometric return, Compounding, Volatility drag.

The lowest price sellers are currently accepting.

The best displayed ask is the lowest current sell order in the quoted market. A marketable buy may execute there if enough size remains, but quotes can change and a larger order may fill across several prices.

See also: Bid, Spread.

A single moment where orders are pooled and matched at one price.

Opening and closing auctions collect eligible orders and determine a clearing price under venue-specific rules, often seeking to maximize executable volume subject to tie-breakers. Closing auctions can concentrate substantial volume because many funds and benchmarks reference the close, but they are not one ordinary trade and their share of daily volume varies by market and day.

See also: Marking the close, Matching engine, Limit order book, TWAP.

An append-only record of what the system decided and why.

Every signal, order, fill, and rejection stored with its inputs and timestamp, written in a form nothing later can alter. Its value appears on the day something goes wrong: without it, diagnosing a bad session is guesswork and reconstruction, and the temptation is to explain the loss rather than find its cause. It is also frequently a regulatory requirement rather than a nicety.

See also: Reconciliation, Data pipeline, Kill switch, Trading journal.

The outside firm that checks a company's published accounts.

An audit is an opinion that the statements are fairly presented under the accounting rules, given by a firm the company pays. It is the strongest assurance in the set of company publications and it is not a guarantee: audits have missed frauds, and figures a company labels adjusted sit outside the audited numbers entirely.

See also: Annual report, Regulatory filing.

How much a series' current value relates to its own recent past.

Positive linear autocorrelation means high observations tend to be followed by high ones at the measured lag; negative means the reverse. Its size varies by asset, horizon, sample, and data treatment. Absolute or squared returns often show more persistence than raw returns, while overlapping windows can create strong mechanical dependence because adjacent observations reuse much of the same data.

See also: Stationarity, Volatility regime, Mean reversion, Sample size.

The mean quantity traded per session over a specified lookback.

ADV adds the included sessions' recorded volume and divides by their count. A 20-session and a 50-session average can differ, and providers can differ on venue coverage, sessions, and adjustments. It supplies context for trading activity but does not count unique participants or guarantee how much liquidity a new order will find.

See also: Volume, Relative volume, Liquidity.

Buying more of a losing position to lower the average price paid.

It lowers the average entry price while increasing exposure and the money lost if price continues down. A price decline is not automatically evidence that a long thesis is false, but adding without a total-size cap or revised risk calculation can let one position dominate the account. A preplanned accumulation rule is testable; being preplanned alone does not make it prudent or profitable.

See also: Martingale, Pyramiding, Invalidation.

Testing rules against historical data.

A way to check whether an idea would have worked, and an easy way to fool yourself. A backtest is only meaningful with realistic costs, a large and varied sample, and rules fixed before looking at the results.

See also: Overfitting, Sample size, Look-ahead bias.

An erroneous price printed in the data feed.

A feed can contain an invalid, duplicated, mis-timestamped, misadjusted, or otherwise misleading record. Some outliers are genuine trades, including erroneous orders, so visual surprise alone does not prove bad data. Check the venue record, surrounding quotes, corrections, and corporate-action treatment before excluding an observation.

See also: OHLC, Liquidity.

How often something happens anyway, before your rule is applied.

If price rises 52% of the time regardless, a pattern preceding a rise 54% of the time adds almost nothing. Judging any pattern means comparing it against the base rate. Otherwise you're measuring the market's default behavior and calling it an edge.

See also: Edge, Sample size, Expected value (EV).

What the strategy has to beat to have been worth running.

A return means nothing without the alternative you gave up. Twelve percent is excellent against a flat market and a failure against a market that rose twenty, and a strategy that beat neither buy-and-hold nor a savings account has cost you time to underperform doing nothing.

See also: Expectancy, Profit factor, Drawdown.

Being measured against an index rather than against zero.

A manager down 12% in a year the index fell 18% has had a good year professionally, and a client who lost 12% may disagree. This measurement choice drives real behavior: it makes tracking the benchmark safe and deviating from it risky regardless of the merits, which is why institutional portfolios often look more alike than their managers' opinions do.

See also: Benchmark, Buy side, Sell side, Tracking error.

A broker's duty to seek the most favorable terms reasonably available.

In jurisdictions that impose it, best execution is generally a duty to use a reasonable process to seek favorable overall terms, not a guarantee of the best displayed price on every order. Relevant factors and review requirements vary by product and regulator and can include price, speed, likelihood, size, and cost. Payment for order flow is one possible conflict, not the only reason for the duty.

See also: Order routing, Transaction cost analysis, Sell side, Broker.

The highest price buyers are currently offering.

The best displayed bid is the highest current buy order in the quoted market. A marketable sell may execute there if enough size remains, but quotes can change and a larger order may fill across several prices.

See also: Ask (offer), Spread.

A moving average with bands based on a chosen multiple of recent standard deviation.

Common settings place upper and lower bands around a moving average using the rolling standard deviation of the same input. The lookback, multiplier, variance convention, and price series affect the result. A price outside a band is distant from that rolling center under the formula; it does not identify exhaustion, imply a normal-distribution probability, or require price to return inside.

See also: Moving average, Standard deviation, Volatility, Overbought and oversold.

A loan you make to a government or company.

Under its contract, an issuer commonly promises interest and repayment of principal at maturity, although zero-coupon, perpetual, convertible, and other bonds differ. A bondholder is a creditor rather than an owner, and payment remains subject to credit, rate, liquidity, call, currency, and contract risk.

See also: Stock (share).

Opening a position because nothing is happening and waiting is uncomfortable.

Most of a trading day contains nothing worth doing, and that is the normal state rather than a problem to solve. Boredom trading treats the discomfort of waiting as a signal, which it is not. It is worth naming separately from overtrading because the trigger is different: overtrading chases activity, while this specifically fills empty time, and it responds to a stated minimum standard for what qualifies rather than to a count.

See also: Overtrading, No-trade condition, Setup.

Whether shares can actually be borrowed to sell short, and what that borrow costs.

A short stock position generally requires a locate and borrowed shares. Scarcity can raise the borrow fee, and the lender or broker may recall or buy in the position under the account agreement, potentially at an adverse price. Hard-to-borrow status can contribute to squeeze risk, but it neither causes nor predicts a squeeze by itself.

See also: Short selling, Liquidity, No-trade condition.

Starting from an individual instrument and checking the market around it afterwards.

You find something interesting on a chart or in a filing and then ask what conditions it sits in. Its strength is that nothing gets screened out before you have seen it. Its weakness is order of operations: by the time you check the context you already like the idea, so the check becomes a search for permission rather than a filter. Most people do this and call it top-down.

See also: Top-down analysis, Confirmation bias, Setup.

Entering as price leaves a level or range, accepting a worse price for earlier confirmation of movement.

A long breakout enters after price crosses a defined boundary; a short breakout is the mirror. Entry is later than anticipation and may face a wider stop, slippage, or a false break. Some rules use relative volume as an additional observable condition, but volume does not reveal who funded the move and only testing can show whether that filter improves the rule.

See also: Pullback entry, Entry trigger, Slippage, Relative volume.

A firm that accepts, routes, executes, or arranges customer trades.

A broker can carry the account itself or use separate clearing and custody firms. Brokers differ in products, costs, conflicts, execution quality, reliability, protections, and regulatory status, all of which should be verified for the relevant entity and jurisdiction.

See also: Exchange, Market maker.

Institutions that invest money: funds, pensions, asset managers.

They hold positions and are measured on the performance of what they hold, usually against a benchmark rather than in absolute terms. That detail explains behavior that otherwise looks irrational, such as buying something they consider overvalued because not holding it is a larger risk to their measured performance than holding it is.

See also: Sell side, Benchmark-relative performance, Benchmark.

A company repurchasing its own shares under an authorized program or transaction.

Repurchases can use cash or debt, and acquired shares may be retired or held as treasury stock depending on the jurisdiction and company. If diluted shares outstanding fall while earnings stay constant, earnings per share rises mechanically; employee issuance or other changes can offset it. Value depends on price paid, funding, taxes, alternatives, and subsequent use of the shares, not the announcement alone.

See also: Shares outstanding, Dilution, Net income.

Annual return divided by the worst peak-to-trough loss over the period.

Where Sharpe asks about typical wobble, Calmar asks about the worst moment, which is closer to the thing that actually ends strategies and careers. Its weakness is that it rests on a single historical event: the largest drawdown that happened to occur in the window. A strategy tested over a calm period shows a flattering Calmar precisely because its worst case has not happened yet.

See also: Drawdown, Recovery factor, Sharpe ratio, Sortino ratio.

A bar showing one period's open, high, low, and close.

The body spans open to close. The wicks reach to the high and low. Direction comes from whether the close is above or below the open, never from color alone, since color schemes differ and not every reader can distinguish them.

See also: OHLC, Doji, Wick (shadow).

The largest size an approach can be traded at before its own market impact spoils it.

A method can degrade as size becomes material relative to available liquidity, especially when it needs fast entry or exit. Capacity is not one fixed, invisible ceiling: it varies with market conditions, execution method, holding period, and acceptable cost. Measure fill rate, delay, impact, and realized slippage across order size and stressed regimes.

See also: Liquidity, Slippage, Average daily volume.

A retrospective label for a sharp, high-volume decline near an apparent low.

A sharp decline with unusually high volume is sometimes labeled capitulation after price stabilizes or reverses. Price and volume alone cannot prove panic, forced liquidation, holder intentions, or that the decline is ending. In real time the same observations can accompany continuation.

See also: Climax, Margin call, Volume.

Earning the difference between what a position pays you and what it costs to hold.

A carry position seeks income from a yield, funding, or forward-price difference, while taking price, currency, credit, liquidity, and financing risk. Some carry strategies have produced negatively skewed returns, but the shape is not identical across instruments or periods. Carry must be evaluated together with changes in the asset's price and all holding costs.

See also: Tail risk, Leverage, Risk of ruin.

A brokerage account that does not borrow on margin to fund purchases.

With fully paid long shares, not borrowing on margin avoids margin interest and margin calls, and the investment loss is generally bounded by the amount paid, plus fees. Cash accounts can still offer products or create settlement obligations with different risks, so the account label alone does not guarantee that every possible position has bounded loss. Trading is also subject to settled-cash rules.

See also: Margin account, Settlement.

How fast a company is consuming cash it has not earned.

Historical burn divided into available liquidity gives a rough runway only if burn, receipts, restrictions, debt terms, and working capital remain similar. A company may cut costs, grow revenue, borrow, sell assets, receive milestones, or issue securities, so filings cannot reveal an exact financing date or guarantee dilution. Scenario ranges are more honest than one deadline.

See also: Free cash flow, Share offering, Dilution.

A company that holds crypto for you and matches trades internally.

A centralized venue commonly keeps customer keys and records balances on an internal ledger, although custody, segregation, regulation, and legal ownership differ by venue and jurisdiction. Failure can expose customers to freezes, insolvency, or creditor risk. On venues with weak surveillance or disclosure, reported volume and reserves deserve independent verification.

See also: Self-custody, Custody risk, Counterparty risk, Wash trading.

Chaikin money-flow volume divided by total volume over a fixed window.

Commonly calculated over 20 or 21 periods, CMF sums volume weighted by each close's location in its high–low range and divides by total volume for the window. A positive value means upper-range closes received more weight under that formula; it is not a direct measurement of net money entering, buyer dominance, or intent. The reading can also move materially when an old high-volume period leaves the window.

See also: Accumulation/distribution line, Money Flow Index, Volume.

A named shape formed by several swings of price, rather than by one or two candles.

Where a candlestick pattern describes one to three periods, a chart pattern describes a structure built from swing highs and lows over many of them: a triangle, a flag, a head and shoulders. The distinction matters because the evidence base differs. A shape that takes twenty periods to form is a claim about a stretch of trading, and it can only be tested at all once it is defined precisely enough for a machine to find it.

See also: Trend, Support and resistance, Pareidolia.

Entering after the move, at a worse price and with a wider stop, because you did not want to miss it.

Chasing is what fear of missing out produces once it reaches the order ticket. The damage is not that the direction is wrong, it often is not. It is that entering late puts the invalidation much further away, so the same account risk buys a far smaller position and the reward-to-risk of the trade has quietly halved. A trade taken at the right idea and the wrong price is a different trade.

See also: Fear of missing out, Per-unit risk, Breakout entry.

High volume producing very little net price movement.

A great deal of recorded trading and little net price progress. The combination is observable, but ordinary price and volume cannot establish who traded, why they traded, or whether the next move will be up or down.

See also: Absorption, Volume, Range.

One outcome being far more common than the other in the training data.

If 2% of bars precede a defined large move, a model that always predicts "no large move" is 98% accurate but has no ability to identify the rare class. Accuracy can therefore mislead on imbalanced labels. Precision, recall, calibration, payoff, base rates, and trading costs should be selected to match the actual decision.

See also: Accuracy, Data leakage, Expectancy, Base rate.

A retrospective label for an extreme activity surge near the end of a strong move.

A buying climax is labeled near an advance's end and a selling climax near a decline's end, usually after reversal or stabilization supplies the missing context. In real time, unusually high volume during a sharp move is also consistent with continuation. The observation does not establish in advance that the trend has ended.

See also: Capitulation, Volume, Outcome bias.

Two wandering series tied together so their gap keeps returning to normal.

Two or more nonstationary series are cointegrated when a specified linear combination is stationary under the model. This differs from short-run correlation and can support a relative-value hypothesis, but an estimated spread is not guaranteed to close on a tradable horizon. Parameters can be unstable and structural breaks can invalidate the historical relationship.

See also: Correlation, Spurious correlation, Stationarity, Structural break.

Assets pledged to cover what you might owe.

Broker margin and assets posted against derivatives are forms of collateral. Requirements can change with price, volatility, concentration, credit quality, and house or clearing rules, while pledged assets may gain or lose value independently. The resulting liquidity pressure can contribute to forced sales, but mechanics differ across retail securities, futures, cleared derivatives, and bilateral agreements.

See also: Margin call, Forced liquidation, Financing cost, Maintenance margin.

Growth that earns further growth on itself.

Each period's gains join the balance and can earn further returns, so growth is multiplicative rather than a flat amount. The same mathematics can increase debt when interest and fees are added, although payments and changing rates alter the path.

See also: Inflation, Rule of 72.

Trading that continues against your own intentions and damages the rest of your life.

The warning signs are behavioral rather than financial: trading larger to feel the same thing, hiding activity from people close to you, using money set aside for something else, being unable to stop after deciding to, and mood that tracks an open position. This is recognized as a form of problem gambling and is treated as one, which means help exists and works. Nothing in a trading course is a substitute for it, and the correct response to recognizing yourself in this list is a support service rather than a better plan.

See also: Gambling, Revenge trading, Risk capital.

Enough of the account resting on one name, sector, or factor that a single outcome decides the year.

Concentration is not automatically wrong, and it is always a decision, so it needs to be a deliberate one with a number attached. The reason to bound it is the shape of stock outcomes rather than a feeling about prudence: individual company returns are extremely skewed, so a concentrated position in one name is much more likely to be a bet on an ordinary company than on an exceptional one, whatever the story around it says.

See also: Diversification, Factor exposure, Correlated risk.

Whether the confidence you record before a trade matches how often those trades actually work.

Record a probability before the outcome, then compare each probability band with the frequency of that defined outcome over a large sample. Calling events 70% likely is well calibrated if about 70% occur; merely ranking high-confidence trades above low-confidence ones measures discrimination, not calibration. Trading usefulness also depends on payoff and costs, not win rate alone.

See also: Setup quality, Overconfidence, Sample size.

A range that expresses how uncertain an estimate is.

A confidence interval is produced by a method that, under its assumptions and repeated sampling, captures the parameter at its stated long-run rate. A wide interval around a win rate can include 50%, but 50% is not automatically "no edge" because payoff and costs matter. Uncertainty often shrinks roughly with the square root of independent sample size; dependence and changing regimes reduce the effective information.

See also: Sample size, Hypothesis test, P-value, Expectancy.

Waiting for follow-through before acting on a pattern.

A rule may require the next candle or another observable event to support a setup before entry. Waiting normally produces a later price and can filter some failed patterns, but whether it improves accuracy or expectancy is an empirical question that must be tested for that precise rule and market.

See also: Engulfing pattern, Base rate.

Seeking and believing evidence that supports what you already think.

It operates on the rating of evidence rather than on its collection, which is what makes it so hard to catch: you are not ignoring the disagreeing chart, you are finding it less convincing. Holding a position makes it markedly stronger, so anything designed to counter it has to run before entry. That is exactly what the counter-thesis is for, and it is why writing one afterwards does not work.

See also: Counter-thesis, Thesis, Analysis paralysis.

When someone's advice also happens to pay them.

A person recommending a broker, course, or asset may earn referral fees, course sales, or profit from your buying. The advice may still be sound, but you cannot evaluate it without knowing how the person gets paid.

See also: Pump and dump.

A summary, often a mean or median, of published analyst forecasts.

Consensus is a useful public proxy for expectations, not a complete measurement of every market participant's belief or everything embedded in price. Reactions can depend on the result versus the distribution of expectations, guidance, positioning, valuation, liquidity, and other news. Event analysis should state which expectation measure it uses.

See also: Surprise, Event-driven, Event filter.

A pause where price moves sideways in a narrowing or steady range after a move.

Consolidation is what most chart patterns are made of. Price advances, then spends a stretch going nowhere in particular while positions change hands, then either resumes or reverses. The shape that pause takes is what gets a name. The pause itself is the real subject, and it carries no promise about which way it ends.

See also: Range, Chart pattern, Volume dry-up.

An index tracking price changes for a weighted basket of household goods and services.

Statistical agencies update items, weights, quality adjustments, and methods over time, so the basket is not literally fixed forever. The index is an aggregate and any household's costs can differ. A common core measure excludes food and energy because their prices are often volatile, but exclusion does not mean their moves always reverse or do not affect households.

See also: Gross domestic product (GDP), Monetary policy.

Indicators whose overlapping inputs or calculations can produce dependent signals.

RSI, MACD, stochastics, and moving averages can reuse much of the same price history, although their inputs and transformations are not identical. Agreement can therefore reflect shared data rather than independent corroboration. Dependence and incremental value should be measured on representative out-of-sample data instead of inferred from the number of indicators that align.

See also: Indicator, Confirmation, Overfitting.

Separate positions exposed to overlapping drivers and therefore prone to moving together.

Three semiconductor names can share sector, factor, and market exposure, so counting tickers can overstate diversification. They are not literally one trade and their correlation is neither perfect nor stable. Scenario tests, factor exposures, and correlations across multiple regimes help estimate the overlap; adding planned losses is a conservative stress case, not an exact forecast.

See also: Portfolio heat, Diversification, Sector.

How closely two things have moved together, on a scale from -1 to +1.

Pearson correlation of +1 means a perfect positive linear relationship, -1 a perfect negative one, and 0 no linear relationship in the measured sample. It can miss nonlinear dependence, is sensitive to the window and data treatment, and can change during stress. Correlations among many risky assets have sometimes risen in sell-offs, but this is not universal, and correlation does not establish cause.

See also: Correlated risk, Correlation regime, Spurious correlation, Diversification.

How much things are moving together right now, as against moving on their own reasons.

In a dispersed regime, instrument-specific differences are larger; in a correlated regime, shared factors explain more of the movement. Correlations can change quickly. Equity correlations have often risen during broad sell-offs, but not every pair moves toward +1: safe-haven, duration, currency, and hedge relationships can behave differently. Stress tests should include several correlation scenarios.

See also: Correlated risk, Diversification, Portfolio heat.

The strongest honest argument against your own trade, written before you take it.

Not a token doubt. The counter-thesis is what a competent person on the other side of your trade would say, argued as well as you can argue it. Writing it does two jobs: it surfaces the evidence you were skipping past, and it gives you the specific thing to watch for, because the counter-thesis coming true IS the invalidation. When you cannot construct one, that is information about how well you understand the setup rather than proof you are right.

See also: Thesis, Invalidation, Confirmation.

The chance that the other side of your arrangement fails to do what it promised.

Counterparty exposure arises in contracts, custody, settlement, and financing arrangements. Central clearing, collateral, segregation, netting, and regulation can reduce it but do not make it zero. In an uncleared or issuer-backed product, value can depend directly on the counterparty remaining able and willing to perform under the contract.

See also: Custody risk, Broker, Settlement.

Swapping a basket of the underlying shares for fund shares, and back.

Authorized participants can transact with an ETF in large creation units, delivering or receiving securities, cash, or both under the fund's rules. Arbitrage may help align market price and net asset value, but costs, risk, market closures, and hard-to-trade holdings can limit it. The mechanism encourages rather than guarantees a tight price.

See also: Net asset value, Premium or discount, ETF (exchange-traded fund).

One average crossing another, or price crossing an average.

A crossover is observed when the difference between two specified series changes sign under the chosen sampling rule. Because moving averages use current and earlier observations, their crossing describes those calculations rather than the cause of a move or the identity of traders. A crossover rule can trade repeatedly in oscillating prices, but frequency and net results must be measured after execution costs.

See also: Moving average, Lag, Range.

A quote of one currency priced in another, because a currency has no price by itself.

In EUR/USD the first currency is the base and the second is the quote, and the number says how many units of the quote one unit of the base buys. Buying the pair means buying the base and selling the quote at the same time, which is why every currency trade is two positions wearing one ticket. This is the structural difference from a share: there is no such thing as owning a currency's value in isolation, only its value against something else.

See also: Pip, Over-the-counter market, Rollover.

The risk attached to who is holding your assets and how they are segregated.

Assets held for you are supposed to be segregated from the firm's own, so that the firm failing does not take them with it. How well that holds depends on jurisdiction, on the compensation scheme covering the account, and on whether the firm actually did what its rules require. This is why the boring questions about regulation and insurance are risk questions rather than paperwork, and why they are worth asking before an account is funded rather than after.

See also: Counterparty risk, Platform risk, Broker.

A product rebalancing exposure to target a stated multiple for each trading day.

The daily objective is generally measured from one close to the next and remains subject to fees, financing, tracking difference, and market disruptions. Chaining daily leveraged returns is not the same as applying the multiple once to a multi-day return, so longer-period performance depends on the sequence of daily moves as well as the endpoint.

See also: Leveraged fund, Inverse fund, Path dependence.

A venue where orders are not displayed before they trade.

A dark venue does not publicly display eligible orders before execution. Non-display can reduce information leakage and market impact for some orders, but users, order types, access, matching, and reporting rules vary by venue and jurisdiction. Reported dark trades do not reveal participant identity or motive, and a lit order book therefore shows only part of the potential trading interest.

See also: Iceberg order, Fragmentation, Order routing, Level II.

Information from the future reaching a model during training.

Leakage can enter by scaling with full-sample statistics, filling a gap from later data, shuffling time-ordered observations, or overlapping labels and features. It can make historical performance look far better than deployable performance. Strong backtests can fail for many other reasons too, so time-aware validation and a pipeline audit are needed rather than diagnosing leakage from performance alone.

See also: Out-of-sample data, Overfitting, R-squared, Model drift.

The path market data takes from its source to your strategy's inputs.

It includes fetching, cleaning, adjusting, aligning timestamps, and storing data. Missing bars, duplicated adjustments, stale values, or timezone errors can corrupt signals even when strategy code is correct. Recording and testing every transformation makes results reproducible and helps trace a surprising signal back to its inputs.

See also: Reconciliation, Audit log, Bad tick, Data leakage.

Reusing the same data until something looks significant.

Searching many rules, periods, or universes and reporting the winner as if it were the only thing tried. The apparent quality of the best result depends entirely on how many were searched, and that count is exactly what disappears when a backtest gets shared.

See also: Multiple testing, Selection bias, Overfitting.

Short interest divided by average daily volume.

Also called the short-interest ratio, it scales reported short interest by a chosen average-volume window. A value of eight does not mean shorts can or will exit in eight days: not all volume is available to them, volume changes, and covering can affect price. It is a crowding and liquidity reference, not an exit forecast.

See also: Short interest, Short squeeze.

A protocol that uses smart contracts or distributed systems to facilitate trades.

Some protocols use automated liquidity pools and others use order books or aggregators. Users often retain key control until signing a transaction, but interfaces, bridges, governance, or contracts can still fail or block access. Public transaction ordering can create front-running and slippage risks, while legal recourse and decentralisation vary widely.

See also: Smart-contract risk, Centralized exchange, Self-custody.

How sound a choice was given what you knew at the time.

Assessed from the information, reasoning, and risk control available before the outcome was known. Good decisions sometimes lose. Bad decisions sometimes win. Reviewing decision quality separately from results is what makes improvement possible.

See also: Outcome bias.

A security being removed from the exchange it traded on.

Happens through bankruptcy, acquisition, or falling below listing requirements. It matters for testing because databases often stop recording a delisted name before its final decline, so even a dataset that includes failures can be missing how far they actually fell.

See also: Survivorship bias, Backtest, Bad tick.

A stablecoin losing the fixed value it was supposed to hold.

If an asset or account is denominated in that stablecoin, a depeg changes both its own value and the interpretation of quoted prices. Effects depend on which venues and pairs use it. Redemption limits, liquidity, network congestion, or withdrawal restrictions can worsen an exit, but they do not accompany every depeg.

See also: Stablecoin, Liquidity, Forced liquidation.

An existing holder's percentage falling because new shares were issued.

If outstanding shares increase by 25% while your holdings stay fixed, your ownership percentage becomes 1/1.25, or 80%, of its former level, a 20% reduction. Economic value need not fall by the same percentage because the company receives assets or services and future results can change. Options, convertibles, awards, and repurchases also affect basic or diluted share counts.

See also: Share offering, Shares outstanding, Cash burn.

Spreading exposure so one outcome has less influence on the whole portfolio.

Holding exposures whose drivers are not identical can reduce concentration in one company, theme, or risk factor. Correlations and liquidity can change, especially under stress, so diversification reduces selected risks but does not guarantee against loss or remove market-wide risk.

See also: ETF (exchange-traded fund).

A distribution a company declares for eligible shareholders.

Many dividends are cash distributions paid from current or accumulated resources, but distributions can also include stock or a return of capital. The board decides whether to declare them, and payments can be cut or stopped. Other things equal, the share price is expected to adjust downward by roughly a cash dividend when the stock begins trading ex-dividend, although market moves can obscure that adjustment.

See also: Stock (share), Adjusted price.

The annual dividend divided by the share price.

The quoted yield rises if the assumed payment rises or the share price falls. A very high trailing or indicated yield can reflect a depressed price and possible cut, a special distribution, stale data, or a different calculation convention. Check payment sustainability, declaration status, tax treatment, and whether the figure is trailing or forward-looking.

See also: Dividend, Ex-dividend date.

A candle that opens and closes at nearly the same price.

Its small body records little net change from open to close even if the high-low range was large. Calling it indecision is an interpretation; the candle alone does not identify beliefs, order flow, or what comes next.

See also: Candlestick.

A morning or evening star whose middle candle is a doji.

A long body, then a doji printed beyond its close, then an opposite candle closing past the first body's midpoint, after a declared prior move. Every condition except the middle candle matches the ordinary star. It's rarer than the plain star because open and close must nearly match.

See also: Morning star / evening star, Doji, Abandoned baby.

The chart whose answer wins when two timeframes disagree, decided before the disagreement happens.

Timeframes conflict constantly, and that is normal rather than a signal: a bullish hour inside a bearish week is a pullback, not a reversal, and the same picture inverted is a bounce in a downtrend. What turns the conflict from a problem into information is naming in advance which chart owns the direction question. Without that, whichever timeframe agrees with the trade you already want becomes the dominant one, and it changes from trade to trade.

See also: Timeframe, Market regime, Rule set.

Two failures at roughly the same price, separated by a move away and back.

A double top is two highs near one price with a trough between them; a double bottom is the mirror. The shape records two historical reversals without identifying who caused them or guaranteeing another response. Many definitions consider the pattern complete only after price breaks the intervening trough or peak, so criteria and timing must be fixed before testing.

See also: Support and resistance, Tweezer top / bottom, Chart pattern.

How far an account has fallen from its previous peak.

Measured as a percentage from the high-water mark. Recovery is not symmetric: a 50% drawdown requires a 100% gain to return to level, which is why limiting drawdown matters more than chasing large gains.

See also: Risk per trade, Position size, Variance.

A stated fall from the account's high water mark at which you stop trading for a period.

Daily, weekly, and monthly limits nest inside a maximum, and each one governs a different length of bad patch. Their purpose is not to prevent losses, which nothing prevents. It is to bound how much of the account any single stretch of bad decisions or bad luck can take, and to break the sequence in which one loss produces the behavior that causes the next. A limit only works if the resumption rule is written too.

See also: Drawdown, Risk of ruin, Kill switch.

A pooled investment product whose shares trade on an exchange.

An ETF can track a broad portfolio, a narrow sector, one commodity, derivatives, or an active strategy. Some diversify across many holdings and others concentrate or add leverage. Its objective, holdings, structure, liquidity, taxes, tracking, and ongoing costs matter more than the ETF label alone.

See also: Stock (share), Diversification.

The scheduled publication of a period's results.

A known date, with a published expectation attached to it, which is what makes the price react to the gap rather than to the figure. Holding a position through one is a decision about exposure rather than a forecast, because the direction of a repricing nobody has seen yet cannot be known.

See also: Consensus, Surprise, Guidance.

A measurable advantage expected to persist under defined conditions.

In trading, an edge is positive expected value after realistic costs for a specified rule, market, and horizon. It is supported, not permanently proved, by representative data, out-of-sample testing, and continued monitoring because conditions and implementation can change.

See also: Expectancy, Sample size.

Comparing traded volume with net price change over the same period.

The heuristic compares traded quantity with net price change. High volume with a large move and high volume with little progress are different observations, but neither identifies opposition, initiative, or participant intent from aggregate bars alone. Use the comparison as a description to test, not a prediction.

See also: Absorption, Churn, Volume.

Accessible savings set aside for emergencies and income disruption.

Its purpose is to keep a lost job, medical bill, or urgent repair from forcing an investment sale or expensive borrowing. Three to six months of essential expenses is a common starting guideline, not a universal rule; job stability, insurance, dependants, benefits, debt, and access to funds all affect an appropriate amount.

See also: Risk capital.

A recorded rating of your own state before and after a session.

Rating tiredness, stress, and urgency before trading creates a contemporaneous record that can later be compared with rule adherence and outcomes. A correlation in one person's journal does not establish cause and needs enough observations to interpret. The check-in is a self-monitoring aid, not a validated treatment or guarantee of better decisions.

See also: Trading journal, Post-loss routine, Decision quality.

A candle whose body completely covers the previous candle's body.

A common bullish definition has the second real body cover the prior down body's open-close range; bearish mirrors it. Definitions differ on gaps and equality. The geometry neither identifies intent nor guarantees reversal, so location and confirmation only matter if a precisely defined rule has evidence.

See also: Harami, Outside bar, Candlestick.

What it would cost to buy the whole company, counting its debt and its cash.

Market capitalisation values only the equity, and a buyer of the whole business inherits the borrowings. Adding debt and subtracting cash produces a figure that can be far larger than the capitalisation, and the gap is where heavily indebted companies show up. Two companies with identical capitalisations can be very differently sized once the lenders are counted.

See also: Market capitalisation, Net income.

The single observable event that turns a qualifying setup into an order.

A trigger is written so that two people watching the same chart would agree on whether it happened: a close above a level, a retest holding, a specific candle closing. Its purpose is to remove the moment of judgment at the point where judgment is worst, which is when money is about to be at risk. A setup with no trigger is an opinion, and an opinion gets acted on whenever the urge is strongest.

See also: Setup, Breakout entry, Pullback entry, Confirmation.

A rule excluding trades around scheduled events whose outcome you cannot read from a chart.

Earnings, central-bank decisions, and data releases can create volatility, gaps, and slippage that a chart cannot forecast reliably. A filter can avoid that exposure under a written rule. If a strategy deliberately holds through events, scenarios and smaller sizing can limit estimated damage, but no chosen "worst plausible" gap is a guaranteed bound.

See also: No-trade condition, Gap, Trading halt.

Trading built around a scheduled or announced event rather than around chart structure.

Examples include earnings, central-bank decisions, trial results, mergers, and corporate actions. Strategies may analyze expectations, probability, terms, or post-event reaction; none can assume the outcome or price response. Gaps, halts, and liquidity changes make scenario-based sizing and exit planning central, but risk control does not create an edge.

See also: Event filter, Gap, Trading halt.

The first day on which buying a share no longer earns the upcoming payment.

Under the market's settlement rules, a buyer on or after the ex-dividend date generally does not receive the declared distribution. Other things equal, price is expected to adjust by roughly the cash amount, but ordinary market movement, taxes, currency, and microstructure can make the observed change different. Buying solely to capture a dividend does not create a guaranteed gain or zero return.

See also: Dividend, Dividend yield, Adjusted price.

The venue that matches buyers with sellers.

An organized venue that lists products and matches orders under published rules. Regulation, transparency, clearing, and settlement protections vary by market and jurisdiction, so the word exchange alone does not guarantee that a product or venue is safe.

See also: Broker, Market maker.

What a repeated decision is worth on average per attempt.

Each possible outcome's value multiplied by its probability, all added together. Positive expectancy means the assumed outcome distribution has a positive average; it does not promise the next result or remain valid if probabilities, payoffs, or costs change. Estimates need enough comparable observations and must include trading friction.

See also: Expected value (EV), Variance, Edge.

The average loss on the bad days, given that a bad day happened.

At a chosen confidence level and horizon, expected shortfall estimates the average loss conditional on being in the tail beyond the corresponding VaR threshold. It describes breach severity that VaR omits, but it is not automatically more accurate: both depend on limited tail data, model choices, liquidity assumptions, and changing conditions.

See also: Value at Risk, Tail risk, Kurtosis, Risk of ruin.

The average result of a bet if you could repeat it many times.

Computed by multiplying each outcome by its probability and summing. A game paying $2 half the time and losing $1 half the time has EV of +$0.50 per play. EV describes the long run, not any single play.

See also: Expectancy, Variance, Sample size.

The annual fee a fund charges, as a percentage of what you hold.

It is deducted from fund assets under the prospectus rather than billed as a separate retail charge. It creates a predictable performance drag, but taxes, sampling, trading costs, securities lending, index methodology, and portfolio management also affect benchmark-relative results. Compare like-for-like funds and read the current disclosure.

See also: Tracking error, ETF (exchange-traded fund).

Being able to say why a model produced the answer it did.

Explanations can help inspect inputs, monitor dependencies, challenge a decision, and meet governance requirements. They do not prove causality, correctness, stability, or profitability, and an explanation method can itself be approximate. Opaque and interpretable models alike need validation, monitoring, limits, and documented failure procedures.

See also: Model drift, Residual, Overfitting, Thesis.

A moving average that weights recent periods more heavily than old ones.

An EMA applies exponentially declining weights, so recent observations contribute more than older ones and, under the usual recursive formula, older data retain progressively smaller influence. Compared with a simple average using the same nominal period, it commonly responds more to recent changes. That weighting does not make it inherently more accurate or predictive; results depend on the input, initialisation, parameter, rule, and market.

See also: Moving average, The smoothing trade-off, Lag.

The Federal Reserve committee that sets the target range for the federal funds rate and directs US monetary policy.

The Federal Open Market Committee normally holds eight scheduled meetings a year and can meet as needed. Each scheduled decision has a statement and press conference; a Summary of Economic Projections is released four times a year, not at every meeting. Prices can move throughout the announcement and conference, but which component matters most varies by meeting and prior expectations.

See also: Monetary policy, Risk-free rate.

How much of your book depends on one shared driver rather than on the individual names in it.

A factor is anything many instruments respond to at once: the broad market, interest rates, the price of oil, or a style such as high-growth against value. You can hold twelve unrelated-looking names and still have almost all of your risk in one factor, and on the day that factor moves, the positions do not behave like twelve decisions. Factor thinking is what turns the vague worry about correlation into something you can actually count.

See also: Correlated risk, Concentration risk, Sector.

Stating a conclusion more exactly than the evidence behind it can support.

"This goes to 62.40 by Thursday" carries four decisions' worth of certainty on evidence that supports maybe one. The damage is not the wrong number, it is that precision reads as confidence and confidence sets position size, so a habit of over-specifying quietly becomes a habit of oversizing. The honest form of a view states a direction, a condition that would disprove it, and a rough magnitude, and it says which of the three is the least certain.

See also: Confidence calibration, Overconfidence, Thesis.

The pull to enter a move already underway, because watching it run feels worse than losing on it.

It is driven by anticipated regret rather than a tested trigger. It can arise after a move has extended, when an evidence-based invalidation may be farther away and reward-to-risk less favorable, but that timing is not universal. Compare the proposed entry with the written setup and risk rules rather than treating urgency as evidence.

See also: Chasing, Entry trigger, Overtrading.

What it costs to hold a position funded on credit.

Margin interest, stock-borrow fees, and funding embedded in derivatives can accrue with time and change during a holding period. They must be placed on the same notional, horizon, and return basis as the strategy before comparison. A positive gross trading result can become negative after financing, but the cost may also be offset partly by cash interest or favorable carry.

See also: Collateral, Leverage, Margin account, Carry.

A short, shallow pause after a sharp move: a small channel against the move, or a small triangle.

Both describe the same idea with different geometry. After a fast directional move, price pauses in a tight range that drifts slightly against the direction it came from. Drawn with two parallel lines it is called a flag. Drawn with two converging lines it is a pennant. The defining features are that the pause is SHORT and SHALLOW relative to the move before it. A pause that lasts as long as the move, or gives back most of it, is describing something else.

See also: Consolidation, Chart pattern, Measured move.

The shares actually available for public trading.

Float generally excludes restricted or closely held shares under a provider's definition. A smaller float can contribute to limited depth, volatility, wider spreads, and market impact, but turnover, ownership concentration, order flow, venue, and current liquidity also matter. Float is not the number of shares that will actually be offered for sale.

See also: Market capitalisation, Liquidity.

The broker closing your positions for you, at their timing and their choice of what to sell.

When an account breaches maintenance or another contractual risk threshold, the broker may liquidate positions, sometimes without first issuing a call or allowing time to deposit funds. Which assets are sold, when, and in what order depends on the agreement, product, and jurisdiction. Execution can occur in stressed conditions and does not guarantee enough proceeds to eliminate a deficit.

See also: Margin call, Maintenance margin, Leverage.

Betting a fixed fraction of the Kelly size, commonly a half or a quarter.

Using half or quarter of an estimated Kelly fraction reduces exposure to drawdowns and to error in estimated probabilities or payoffs. In simple idealized models, half Kelly retains about three quarters of maximum expected log growth while volatility scales roughly with size; those figures are not universal for changing, correlated trading returns. Fractional Kelly reduces risk but does not make an uncertain edge safe.

See also: Kelly criterion, Position size, Risk per trade, Risk of ruin.

The same instrument trading across many separate venues at once.

Many securities trade across exchanges, alternative systems, dealers, and internalisers, while some products are more centralized. A view from one venue can therefore omit other liquidity. Consolidated and proprietary feeds differ in coverage, latency, depth, and cost, so the data subscription and venue scope should be checked rather than inferred from a screen.

See also: Order routing, Dark pool, Level II, Limit order book.

Cash left after running the business and paying for equipment and facilities.

A common definition is operating cash flow minus capital expenditure, but companies and data providers may use different formulas. It is derived from the cash-flow statement, not the bank balance, and is not all automatically distributable because debt, working capital, leases, acquisitions, and other needs remain. Reconcile the definition and multi-period drivers before comparing firms.

See also: Net income, Revenue, Cash burn.

Current token price multiplied by a stated maximum or fully diluted supply.

FDV extrapolates today's marginal price to a larger supply; it is not the price at which all those units could actually trade and is not a forecast of future network value. A gap from circulating market capitalisation flags potential supply expansion, but burns, emissions, governance, vesting, demand, and liquidity can change. Supply schedules and holder concentration must be checked separately.

See also: Token unlock, Dilution, Market capitalisation.

The recurring payment between long and short holders of a perpetual.

When the contract trades above spot, longs pay shorts, and the reverse when it trades below. It is what keeps a contract with no expiry anchored to the underlying price. For a holder it behaves like rollover in currencies: a cost or credit accruing with time rather than with price, which can quietly dominate the result of a position held through a crowded period.

See also: Perpetual, Financing cost, Rollover, Positioning.

An agreement to buy or sell something at a set price on a set future date. Both sides are obliged.

Unlike an option holder, both sides of a futures contract have obligations under standardized exchange and clearing rules. Positions are commonly marked to market daily and can settle financially or by delivery. Low performance-bond requirements permit high leverage, but a trader can post additional capital; contract size, margin, settlement, limits, and loss potential must be checked.

See also: Option, Leverage, Open interest.

Expecting independent outcomes to even out, so a run of losses makes a win due.

Nothing is due. If trades are independent, the previous five results carry no information about the sixth, and the run does not create a debt the market has to repay. It is dangerous because it is the reasoning that justifies sizing up during a losing streak, which is the martingale wearing a plausible argument. Where results are not independent, the dependence usually runs the wrong way: a losing run degrades your state, which makes the next decision worse rather than better.

See also: Martingale, Sample size, Risk of ruin.

Putting money at risk with no edge and negative expectancy.

The defining feature is not the venue or the excitement. It is the mathematics. Where there is no edge, the expected result of repeating the wager is a loss, and repetition makes that loss more certain rather than less. Trading without an edge is gambling regardless of what the screen looks like.

See also: Trading, Investing, Edge, Expectancy, Expected value (EV).

A jump between one period's close and the next period's open.

Markets are closed for hours or days at a time, and news arriving in between is priced in the moment trading resumes. With no opportunity to trade at the levels skipped over. This is why stops cannot promise their exit price.

See also: Stop order.

Strategies built around the price jump between one session's close and the next one's open.

A session gap can reflect news, order imbalance, changed liquidity, corporate-action adjustment, or trading on other venues while the displayed session was closed. Rules may fade or follow selected gaps, but neither has automatic validity. The definition, adjusted data, eligible events, costs, and selection criteria must be fixed before testing.

See also: Gap, Premarket and after-hours, Event filter.

The constant rate that would have produced the actual ending balance.

Chain the period returns together multiplicatively rather than adding them, then take the appropriate root. This is the only average that answers what a pound left in the strategy actually became, which is why it is also called the compound annual growth rate. When a marketing document quotes an average return, the single most useful question is which of the two averages it is, because the difference is not a rounding matter on anything volatile.

See also: Arithmetic return, Compounding, Volatility drag, Drawdown.

The total value of everything an economy produces, reported as a rate of change.

GDP estimates the market value of final goods and services produced within an economy over a period. Release frequency, revision schedule, and whether growth is annualised vary by country; a level and several growth rates may all be reported. GDP measures aggregate output, not its distribution, household welfare, or a guaranteed market reaction.

See also: Consumer price index (CPI), Monetary policy.

The total value of every position, long and short added together as positive numbers.

Long $60,000 and short $40,000 is $100,000 of gross notional exposure. It is a useful leverage and activity measure, but equal notionals can carry very different risks because volatility, beta, duration, delta, liquidity, and correlation differ. Gross exposure should therefore be read alongside product-specific risk measures and stress tests.

See also: Net exposure, Leverage, Portfolio heat.

A company's own statement about the periods still to come.

Guidance may be issued with results or separately and can cover revenue, profit, costs, or other operating measures. Markets may weigh it heavily because valuation is forward-looking, but its importance and the price reaction depend on prior expectations, credibility, range, assumptions, and other disclosures. Lowered guidance does not prove that all good performance is over.

See also: Earnings announcement, Surprise, Consensus.

A small body at the top of the range with a long lower wick, after a decline.

The low is far below the body and the close lies near the upper part of the range. Conventional naming calls the shape a hammer after a decline and a hanging man after an advance. OHLC alone cannot identify which participants caused the move or establish a reversal.

See also: Hanging man, Pin bar, Candlestick.

The hammer's shape, but appearing after an advance.

Identical anatomy to a hammer with the opposite conventional reading, because it follows a rise rather than a decline. That two names describe one shape is the clearest proof that context, not shape, carries the meaning.

See also: Hammer, Candlestick.

A small candle whose body sits inside the previous larger body.

The opposite arrangement to an engulfing: a large move followed by a small contained one, conventionally read as momentum pausing. When the second candle is a doji it is called a harami cross.

See also: Engulfing pattern, Inside bar, Doji.

Three peaks with a higher middle one, and a neckline drawn under the two troughs.

A conventional definition requires three peaks with the middle highest, two intervening troughs, and a neckline through them; the inverse form mirrors this. Some rules require a close beyond the neckline before completion. Reliability cannot be inferred from the name or popularity: geometry, tolerance, confirmation, costs, market, and sample must be defined for testing.

See also: Chart pattern, Neckline, Double top and double bottom.

A position or contract intended to offset some risk in another exposure.

A hedge can be close or imperfect depending on the instrument and exposure. Premiums, spreads, financing, opportunity cost, changing sensitivities, and basis risk can make protection differ from the amount expected. It can reduce a specified risk while introducing other risks; its behavior should be defined and stress-tested rather than treated as complete protection.

See also: Tail risk, Net exposure, Correlated risk.

Doing what a crowd is doing, treating their agreement as evidence.

People can follow others because of social proof, information, mandates, benchmarks, or shared signals. Crowded positions may face one-sided exits and reduced liquidity under stress, but not every holder acts together and crowding data are incomplete. Popularity is context to investigate, not proof that a view is right or wrong.

See also: Correlated risk, Liquidity, Pump and dump.

A price area assigned unusually high volume within a profile.

A thick shelf showing that the profile assigned relatively high volume near that price under its selected data and settings. Traders may use it as a reference, but the historical node does not show current orders or promise that price will slow, stall, or reverse there again.

See also: Volume profile, Low-volume node, Point of control.

A spinning top whose range is much wider than the candles before it.

A small but visible body with long wicks on both sides, inside a range much wider than the recent average. The width is measured against the preceding candles, so the name can't be applied to a candle shown on its own. It's conventionally read as neutral, and its range records distance traveled, not participation.

See also: Spinning top, Doji, Range.

Believing an outcome was obvious once you know what it was.

Any chart looks readable from the right-hand edge, which is why reviewing your own trades is harder and more valuable than it appears. It corrupts a journal specifically: a review written after the result is known records what you now think you knew, not what you actually had. The defense is to record the reasoning and the state at entry, before the outcome exists to rewrite them.

See also: Trading journal, Outcome bias, Decision quality.

How long a position is intended to be held.

Holding period influences which risks and costs matter: transaction costs apply at entry and exit, while financing, borrow, decay, events, and opportunity cost accrue with time. For a daily-reset product, even one-day performance can differ from the target because of fees and tracking, while multi-day performance also depends strongly on the path.

See also: Daily reset, Position size, Financing cost.

A procedure comparing observed data with a specified null model.

You specify a null hypothesis, a test statistic, assumptions, and a decision rule, then ask how compatible the observed statistic is with that model. Rejecting the null does not prove a strategy works, quantify the probability it is true, or establish economic value. Data-driven test selection and repeated testing require appropriate correction and honest disclosure.

See also: P-value, Confidence interval, Data snooping, Multiple testing.

A large order that shows only a small piece of itself in the book.

For example, an order may display 100 units while reserving more that can replenish under venue rules. This can reduce information leakage, but availability and priority differ by venue. Displayed depth is neither the full depth nor a guaranteed lower bound on executable depth, because hidden size may exist while displayed orders can cancel before a trade.

See also: Limit order book, Absorption, Level II, Dark pool.

A unique tag on an order so a retry cannot become a second order.

A network timeout does not tell you whether the order arrived. Retrying without a key risks a duplicate position. Not retrying risks no position at all. Attaching a key you generate means the broker recognizes the second attempt as the same instruction and returns the original result, which turns an unanswerable question into a safe one.

See also: Reconciliation, Kill switch, Operational risk.

Believing effort or attention influences an outcome that does not respond to either.

Watching a position does not change what it does, and neither does understanding it deeply. What that mistaken feeling produces is real: more screen time, more intervention, more small adjustments to plans that were fine, and a stop moved because sitting still felt passive. You control entry, size, exit rules, and whether you trade today. The outcome of any single trade is not on that list and never joins it.

See also: Decision quality, Process goal, Trading plan.

The gap between the price when you decided and the price you actually got.

A full implementation-shortfall calculation compares the actual portfolio result with a paper portfolio executed at the decision price, including explicit costs and an opportunity-cost treatment for unfilled quantity. Definitions differ, and the result is sensitive to the decision timestamp, benchmark, cancellations, and attribution method. It is comprehensive only to the extent those inputs are complete and consistent.

See also: Transaction cost analysis, VWAP, Market impact, Slippage.

The history a rule was developed on.

Whatever you looked at while building, tuning, and debugging. A rule is expected to perform well here because it was partly chosen by what performed well here, so a good in-sample result is close to meaningless on its own.

See also: Out-of-sample data, Overfitting, Backtest.

How much of a fund sits in its largest few holdings.

A holding count is a count, and the weights decide the exposure: five hundred companies with thirty per cent in the top five is far less spread than it sounds. Capitalisation weighting produces this by rule rather than by judgment, since the fund holds most of whatever has already grown largest. The published holdings answer the question in about a minute.

See also: Diversification, Sector, ETF (exchange-traded fund).

A fund that holds the constituents of an index so its value tracks that index.

The bridge between a measurement and an investment. An index cannot be bought. A fund tracking it can. The distinction matters because a fund carries things the index does not: a fee, a tracking difference, and a fund manager who has to actually hold and rebalance the positions.

See also: Market index, ETF (exchange-traded fund), Diversification.

A calculated series used to summarize market data.

Many technical indicators transform price or volume so a property such as trend, momentum, or volatility is easier to compare. Others use different inputs, including breadth, options, fundamentals, or sentiment. An indicator cannot create information absent from its inputs, and two indicators built from the same data are not independent confirmation.

See also: Moving average, OHLC, Confirmation.

Price making a new extreme while the indicator does not.

For example, price can make a higher high while a specified oscillator makes a lower high. The comparison is defined only after both extrema and depends on how they are selected. It does not reveal participant intent or establish reversal timing or direction; predictive value must be tested for the exact indicator, parameters, market, horizon, and costs.

See also: Relative strength index (RSI), MACD, Volume divergence.

The gradual rise in prices that shrinks what money buys.

If the same basket becomes 5% more expensive, unchanged cash buys about 4.8% less of it. Cash can be stable in nominal terms while losing purchasing power when inflation exceeds the interest earned; the result depends on the holder's spending basket, taxes, rate, and period.

See also: Compounding.

A candle whose whole range sits within the previous candle's range.

Its high is no higher and its low no lower than the prior bar under the chosen equality rule. This records range contraction, not disagreement or a guaranteed volatility expansion. Any claim about what follows needs a precisely defined, out-of-sample test.

See also: Outside bar, Volatility.

A reportable transaction in company securities by a director, officer, or other covered insider.

Reports can include open-market trades, awards, option exercises, tax withholding, gifts, and plan-based sales. The form and transaction code matter, and the filing does not reveal motive or prove legality, information advantage, or a directional view. Reporting rules and covered persons vary by jurisdiction.

See also: Trading plan, Regulatory filing, Material announcement.

The share of a company held by funds and other large managers.

Ownership reports are snapshots published under jurisdiction-specific thresholds and delays, and some omit shorts, derivatives, or positions outside the reporting rule. They rarely disclose motive, so an index holding, hedge, and active view can look alike. Index-linked holdings can still change through flows, rebalances, corporate actions, lending, and tracking decisions.

See also: Position disclosure, Float, Index concentration.

The gap between two countries' interest rates, which drives currency carry.

Differences in relevant interest rates help determine forward points and financing, but broker charges and which maturities are compared matter. Currency reactions to central-bank communication can reflect the expected future path of rates, inflation, growth, risk appetite, and positioning, not only the current differential or announced decision.

See also: Rollover, Carry, Currency pair.

A recurring link between separate markets, such as rates and long-duration equities.

Equities, government bonds, credit, currencies, commodities, and volatility products respond to overlapping drivers, so movement in one is often informative about the others. The relationships are tendencies rather than laws: they hold for long stretches, they weaken, and they occasionally invert entirely when the driver behind them changes. Using them means knowing the mechanism, because a relationship you cannot explain is one you cannot tell has broken.

See also: Factor exposure, Correlation regime, Sector.

A central bank deliberately trading its own currency to move the rate.

A central bank, treasury, or monetary authority may buy or sell currency, sometimes with other policy actions, to influence the exchange rate. Timing, size, coordination, and effectiveness vary. Intervention or a broken peg can create abrupt moves and slippage through stop triggers, but neither guarantees a particular path or gap through every price.

See also: Currency pair, Counterparty risk, Gap.

The observable event that says the planned thesis is no longer supported.

Invalidation is a prewritten condition, not simply a price that feels painful. "Price closes back below the breakout level" can invalidate a specific breakout plan without proving what price must do next. A loss limit is a separate account constraint. If the planned exit sits far away, position size must make the estimated loss affordable while allowing for gaps and slippage.

See also: Thesis, Structural stop, Position size.

A fund that aims to deliver the opposite of an index's move for one day.

Many inverse funds reset daily and are path-dependent over longer periods. A volatile benchmark that ends near flat can leave both a leveraged long fund and an inverse fund below their starts, but combining them is not literally the same bet twice: net exposure, leverage multiples, rebalance timing, fees, and sizing determine the result. A hedge must be analyzed as a portfolio.

See also: Leveraged fund, Daily reset, Path dependence.

Owning something productive and giving it years to compound.

A company earns profits, a bond pays interest, a property collects rent. You are not predicting next week. You are accepting a share of what an asset produces over a long period. The time horizon is what separates it from trading, not the instrument used.

See also: Trading, Gambling, Compounding, Stock (share).

The bet size that maximizes long-run compound growth, given a known edge.

Under a specified repeated-bet model with known probabilities and payoffs, Kelly selects the fraction that maximizes expected logarithmic wealth. Real trading inputs are estimated, change over time, and include costs and dependence, so estimation error can lead to severe overbetting. Full-Kelly paths can have large drawdowns, but their probability is model-specific rather than a universal 50% rule.

See also: Fractional Kelly, Position size, Risk of ruin, Expectancy.

Two opposite candles separated by a gap, with no body overlap.

Price closes one period in one direction, then opens beyond it in the other and continues, usually around news. Rare, and largely a description of a gap and the repricing that caused it.

See also: Gap, Engulfing pattern.

A pre-decided rule that stops all trading immediately when a stated condition is met.

Down a stated percentage in a day, three losses in a row, an unexplained fill, or any suspicion that the platform is misbehaving: the switch says stop everything and check. Its value is that it is decided in advance, so it does not require you to make a good judgment in the exact conditions where judgment is worst. Like every rule in a plan, it only counts if it is written with a number in it, because a switch that fires when things feel bad never fires.

See also: Platform risk, Drawdown limit, No-trade condition, Market-wide circuit breaker.

How much of the risk is concentrated in rare, extreme moves.

Kurtosis measures the fourth moment and is sensitive to both tails and the center of a distribution; high excess kurtosis often accompanies more extreme observations than a fitted normal model predicts. Many financial return series are heavy-tailed, but frequency depends on the asset, horizon, sample, and model. Kurtosis does not identify when or on which side an extreme will occur.

See also: Tail risk, Skewness, Return distribution, Value at Risk.

The delayed response produced when an indicator smooths observations over time.

A moving average combines new data with earlier observations, so one new move has only part of the weight. A shorter lookback or heavier recent weighting generally responds faster but also follows more short-lived variation. The effective delay depends on the formula and price path, and no amount of smoothing adds knowledge of the next observation.

See also: Moving average, Indicator, The smoothing trade-off.

The delay between deciding and the order arriving.

Signal computation, network time, and the broker's own processing all sit between your rule firing and your order reaching the book. A backtest treats these as zero. For a rule holding weeks that assumption is harmless, and for one reacting to a print it is the whole strategy.

See also: Slippage, Market impact, Market order.

Spoofing spread across several price levels to look like genuine depth.

Rather than one large deceptive order, a series of them are stacked at successive prices so the book appears to have real support or resistance building. The intent and the illegality are identical to spoofing. The multi-level structure exists to look less like a single suspicious order. Regulators treat the two together and detect them the same way, through cancellation rates and the timing of the trader's real orders on the opposite side.

See also: Spoofing, Market manipulation, Limit order book.

A view of the order book beyond the best bid and offer.

Level I shows the top quote; Level II displays additional quoted prices and sizes for the venues included in the feed. It does not reveal intent or guaranteed executable depth: orders can cancel, size can be hidden, and coverage and latency vary. Its usefulness for execution or prediction is an empirical question tied to the product, feed, horizon, and rule.

See also: Limit order book, Order book imbalance, Iceberg order, Fragmentation.

Controlling a larger position than your own money would allow.

Leverage magnifies gains, losses, financing costs, and sensitivity to price movement. Ignoring fees, maintenance requirements, and earlier liquidation, a 20% adverse move against exposure equal to five times equity would consume that starting equity; gaps can create an even larger deficit. Leverage changes the distribution of account outcomes, not the underlying strategy's gross signal quality.

See also: Margin account, Margin call, Position size.

A fund that aims to deliver a multiple of an index's move for one day.

Most such funds target a stated multiple for one trading day, before or after specified fees depending on the objective. Over longer periods, compounding makes return depend on the path: a consistent trend can produce more than the simple period multiple, while volatility can produce less, but neither outcome is guaranteed. Check the prospectus, tracking, costs, and exposure when sizing.

See also: Daily reset, Inverse fund, Path dependence.

A price band beyond which an instrument is not allowed to trade for a period.

Futures markets and some others define a daily band, and once price reaches it trading pauses or can only occur at prices inside the limit. The intent is to slow a disorderly move and give participants time to react. The consequence for anyone holding is that a market can be limit-locked against you with no ability to exit at all, which means a stop order has nothing to execute against and the position keeps losing while you watch.

See also: Trading halt, Gap, Stop order.

Trade only at your price or better, or not at all.

A buy limit sets the highest acceptable price and a sell limit sets the lowest. The order may never execute, may fill only partly, and merely seeing trades at the limit does not prove your order reached the front of the queue.

See also: Market order.

A venue's record of resting buy and sell interest by price and priority.

Displayed orders are ranked under the venue's matching rules; some size can be hidden, routed elsewhere, or absent from the feed. The best displayed bid and offer are current order prices, not an intrinsic value. The book helps describe available displayed liquidity, but cancellations, fragmentation, and hidden interest limit what it says about future fills.

See also: Matching engine, Price-time priority, Bid, Spread.

A single line connecting each period's closing price.

The simplest chart: it discards open, high, and low, keeping only closes. That loss of detail is sometimes an advantage. It strips intraperiod noise and makes the overall path easier to see.

See also: Candlestick, OHLC.

Fitting the straight line that best predicts one series from another.

The output is a slope, saying how much the first thing tends to move when the second moves by one unit, and an intercept. It is the workhorse of quantitative finance and it is honest only when you also look at what it did not explain. A regression will always return a line, including for data with no relationship at all, and reporting the slope without the uncertainty around it is one of the most common ways a weak result is dressed up as a finding.

See also: R-squared, Residual, Correlation, Confidence interval.

How easily you can trade near the quoted price.

A liquid market usually combines depth, tight spreads, resilient quotes, and enough activity for the intended order size. In an illiquid market both entry and exit can be costly, partial, delayed, or unavailable. Liquidity changes with time, direction, size, venue, and stress.

See also: Spread, Volume, Slippage.

How easily size can move in and out right now, which changes far faster than most traders check.

The same instrument can be deep at ten in the morning and hollow at two in the afternoon, deep in June and thin in late December, and deep on an ordinary day and gone entirely during a stressed one. Liquidity is the regime that fails without warning and takes your exit with it, and it is the one people are least likely to notice changing, because it is invisible until you need it.

See also: Liquidity, Slippage, Capacity.

A price axis where equal distances mean equal percentage moves.

On a linear axis, $10 to $20 and $110 to $120 look identical although one doubled and the other rose 9%. A log axis spaces by percentage instead, which is usually the honest way to view long periods or large price ranges.

See also: Timeframe, Candlestick.

Accidentally using information that wasn't available yet.

Testing a rule with data that only existed after the decision point: a closing price used mid-session, or a later revision of a figure. It produces spectacular results that cannot be repeated in real time.

See also: Backtest, Survivorship bias.

Feeling a loss more sharply than a gain of the same size.

The asymmetry is measurable and it produces two opposite errors from one cause. Winners get closed early, because a gain that could evaporate feels urgent, and losers get held, because closing one makes it real. Together they invert the distribution a strategy depends on, cutting the winners short and letting the losers run, which is the precise opposite of what almost every written plan says to do.

See also: Partial exit, Invalidation, R-multiple.

A convention for expressing currency trade size.

Retail spot FX commonly calls 100,000 base-currency units a standard lot, with mini and micro lots often one tenth and one hundredth, but broker and product conventions can differ. Units, pair, price, and account currency determine pip value and notional exposure, so verify the contract specification before sizing.

See also: Pip, Position size, Leverage.

A company with very few shares available to trade.

A small tradable float can contribute to limited depth, volatility, wider spreads, and market impact, but float alone does not determine liquidity because turnover and order flow matter. Gaps and manipulation risk can be elevated in some low-float securities; neither is guaranteed, so current volume, spread, ownership, disclosures, and venue also need checking.

See also: Float, Pump and dump, Spread.

A price area assigned unusually low volume within a profile.

A thin area showing that the profile assigned relatively low volume near that price under its selected data and settings. It may coincide with a fast historical move, but it neither reveals the cause nor guarantees fast movement when price returns.

See also: Volume profile, High-volume node, Value area.

The difference between two exponential moving averages, with a signal line and histogram derived from it.

The MACD line is usually a faster EMA minus a slower EMA; the signal line is an EMA of the MACD line, and the histogram is MACD minus that signal line. All three are transformations of the selected price input, so agreement with another price-derived indicator is not independent evidence. A histogram change describes these differences, not participant momentum or a guaranteed next price move.

See also: Moving average, Exponential moving average (EMA), Indicator divergence.

The minimum equity a margin account must keep, below which the broker demands money or sells your positions.

Initial margin is required to open a position; maintenance margin is the minimum equity required to keep it. For a long margined stock, a price fall reduces both market value and account equity, and equity can fall below the required percentage even though the requirement's dollar amount also changes. Brokers may impose higher house requirements and change them under the account agreement.

See also: Margin account, Margin call, Forced liquidation.

An account that lets you trade with borrowed money.

The broker lends against your holdings, which magnifies both gains and losses and adds interest. If the account value falls below a required level, the broker can demand more money or sell your positions for you, at their timing rather than yours.

See also: Leverage, Margin call, Cash account.

A demand for more money when a leveraged account falls too far.

If equity falls below a required minimum, a broker may demand funds, restrict activity, or liquidate positions under the agreement. Some brokers can liquidate without issuing or waiting on a call. Timing, thresholds, and deficit liability depend on the product, broker, and jurisdiction.

See also: Margin account, Leverage, Maintenance margin, Forced liquidation.

How many instruments are participating in a move, as against how far the index went.

An index can rise while most members fall if a few large constituents rise enough. Breadth measures observations such as advances versus declines, new highs versus lows, or the share above a moving average. It adds information about how widely a move is distributed, but its value for future returns depends on the exact definition, horizon, and tested rule.

See also: Market index, Participation, Relative strength.

Share price multiplied by the number of shares outstanding.

A measure of a company's total market value. Two companies with the same share price can be wildly different sizes. Price alone says nothing about scale, which is why a $5 stock is not automatically 'cheap'.

See also: Stock (share), Float.

The scheduled period when an exchange matches orders normally.

Each exchange has a defined regular session, plus holidays and occasional early closes. Some products also trade in extended or nearly continuous sessions. Liquidity, spreads, order types, protections, and data can differ outside regular hours and must be checked for that venue and product.

See also: Premarket and after-hours, Gap, Liquidity.

The price moving because of your own order.

Above a certain size relative to available liquidity, buying pushes the price up as you buy. A backtest assumes your order changed nothing, which is close enough to true at small size and a serious overstatement once your order is a meaningful share of the volume.

See also: Slippage, Liquidity, Volume.

A number tracking a chosen basket of companies. A measurement, not something you can own.

An index is a calculation: pick companies, decide how much each one counts, and publish the total. That last decision changes what the number means. The Dow Jones Industrial Average is price-weighted, so a company with a high share price moves it more regardless of the company's size. The S&P 500 is weighted by market capitalisation, so the biggest companies move it most. Neither is buyable. What you buy is a fund that tries to track one.

See also: ETF (exchange-traded fund), Market capitalisation, Index fund.

A firm that regularly quotes prices at which it may buy and sell.

Market makers post bids and offers under venue-specific obligations, helping other participants trade without finding one another directly. Quotes can change or be withdrawn, especially in stressed markets. Their revenue and risk come from several sources, including the spread, fees or rebates, inventory changes, and hedging costs.

See also: Spread, Liquidity.

Trading intended to create a false impression of supply, demand, or price.

The unifying element across every form of it is intent to deceive other participants, rather than any particular trade being unusual. This is why the same order can be legitimate or criminal depending on what the sender meant by it, and why prosecutions rest on messages, code, and cancellation patterns rather than on prices. It is prosecuted, it carries prison sentences, and no part of this curriculum treats it as a technique.

See also: Spoofing, Layering, Wash trading, Marking the close.

An instruction that prioritizes prompt execution over a price limit.

A market order normally seeks the best prices available when it reaches the market, but it does not guarantee a particular price and execution itself can be delayed or prevented by a halt or lack of liquidity. In fast or thin markets, one order can fill at several prices far from the last quote.

See also: Limit order, Slippage.

The kind of behavior a market is currently showing, which decides what has any chance of working.

Regimes are described along several axes at once: trending or mean-reverting, calm or volatile, liquid or thin, and correlated or dispersed. Naming the current one is what connects Worlds 13's families to reality, because a strategy is a bet on a regime as much as on a direction. Regimes are also only identifiable with a lag, which is the honest limitation: you know what regime you were in more confidently than the one you are in.

See also: Trend, Range, Volatility regime, Correlation regime.

A pause across a whole venue when a broad index falls by a set amount.

In the US equity market, specified S&P 500 declines can trigger coordinated pauses of different lengths and, at the deepest level, halt trading for the day. Thresholds, reference indexes, times, affected products, and order handling differ across markets. A circuit breaker delays execution; it does not guarantee liquidity, prevent further decline, or prove why participants wanted to trade.

See also: Trading halt, Volatility band.

Trading near the close to push the closing price to a wanted level.

Closing prices can affect valuations, benchmarks, derivatives, and fees. Trading intended to create an artificial close can be manipulation, while a large closing-auction order can also be legitimate rebalancing or execution. Intent, context, and applicable law matter; unusual closing volume alone does not prove misconduct.

See also: Auction, Market manipulation, Wash trading.

Doubling the stake after a loss, on the reasoning that a win must eventually come.

It produces a long run of small wins and one catastrophic loss, and it feels like it is working right up until it is not. Two things defeat it, and both are unavoidable: capital is finite, so the sequence of doublings ends at a bet you cannot make, and brokers and exchanges impose size limits that end it sooner. The reason it survives as folklore is that the failure is rare and total, so most people who use it have only ever seen the part that works.

See also: Anti-martingale, Averaging down, Risk of ruin.

A candle with a full body and almost no wicks.

The open and close lie near opposite extremes of the period. OHLC does not show whether price traveled one way continuously, and the shape does not measure conviction or predict the next candle without additional data and a tested rule.

See also: Candlestick, Doji.

The exchange system that pairs incoming orders with resting ones.

It applies the venue's rules to determine which compatible orders trade, in what quantity, priority, and price. An aggressive buy consumes resting sell interest and may sweep several prices; a passive buy instead waits for a seller to trade against it. Auction, pro-rata, midpoint, and other mechanisms can follow different rules.

See also: Limit order book, Price-time priority, Aggressive order, Slippage.

An unscheduled publication of a fact that would change a reasonable investor's view.

Acquisitions, leadership changes, major contracts, regulatory actions, and financings may require prompt disclosure under applicable rules. Materiality, deadline, format, and permitted delay vary, and dissemination is not literally simultaneous for every participant. Price can react before, during, or after publication for many reasons, including expectations and leakage.

See also: Regulatory filing, Trading halt, Insider transaction.

The furthest a trade went in your favor, and the furthest it went against you, while you held it.

Two numbers per trade, usually written MFE and MAE. Favorable excursion says how much was available at the best moment. Adverse excursion says how close the position came to your stop before it worked. Together they measure the gap between the trade the market offered and the trade you actually took, which is the one thing a profit and loss column can never show. A winner that reached 3R and closed at 1R and a winner that never exceeded 1.1R are recorded identically without them.

See also: R-multiple, Trailing stop, Trading journal.

Trading on the expectation that price stretched far from an average will return toward it.

The reference average, definition of a stretch, holding period, and exit must all be specified. Some mean-reversion rules produce many small gains and occasional large losses, but that shape is not inherent to every implementation. A move can keep extending or the historical mean can change, so position limits and invalidation rules matter.

See also: Range trading, Trend following, Structural stop.

A target taken by projecting the pattern's own height from its breakout point.

The conventional way to derive a target from a chart pattern: measure the height of the formation, then add it to the break level. It is a rule of thumb, not a forecast, and it has one honest use. It tells you in advance whether the idea has room to be worth the risk, which lets you reject a setup before entering rather than discover it afterwards.

See also: Chart pattern, Risk-to-reward ratio, Flag and pennant.

A model quietly getting worse as the market stops resembling its training data.

Nothing breaks and no error is raised. The relationship the model learned simply weakens. Because degradation is gradual and losing runs are normal, drift is genuinely hard to distinguish from bad luck, which is why the thresholds for retiring or retraining a model have to be written down before the losing period starts. Retraining on recent data is not automatically a fix, since it also relearns whatever was temporary.

See also: Stationarity, Structural break, Data leakage, Residual.

Buying what has already risen most, on the observation that recent relative performance has tended to persist over intermediate horizons.

It differs from trend following in what it compares against. A trend follower asks whether this instrument is rising. A momentum trader asks whether it is rising more than the alternatives, and holds the strongest. That makes it inherently a ranking exercise across a universe rather than a judgment about one chart. The effect has been documented in academic work across long samples, and it has also produced severe, sudden reversals, which is the part usually left out of the summary.

See also: Relative strength, Trend following, Sector.

How a central bank sets the price and availability of money.

Central banks can use policy rates, asset operations, reserve tools, lending facilities, and communication to influence financial conditions and economic activity. Higher discount rates can reduce the present value of a fixed future cash-flow stream, other things equal, but policy also changes growth, inflation, currencies, credit, and risk premiums. Market reactions depend on expectations and many simultaneous channels.

See also: FOMC, Gross domestic product (GDP), Consumer price index (CPI).

A bounded oscillator built from price and volume together.

MFI multiplies each period's typical price by volume, classifies that value as positive or negative according to the change in typical price, and compares the window's positive and negative sums on a 0–100 scale. The term money flow is a formula label, not a measurement of net cash entering or leaving the asset. A high reading neither identifies buyers nor establishes that a reversal is due.

See also: Chaikin Money Flow, Volume, Volume divergence.

Reshuffling the same trades to see the range they could have produced.

Your equity curve is one ordering of your trades. Resample them thousands of times and you get a distribution of curves, most of which look nothing like the one you saw. It measures how much of your result was the order things happened in, and the answer is usually more than feels comfortable.

See also: Variance, Sample size, Drawdown.

Three candles: strong move, small pause, strong move the other way.

A morning star conventionally follows a decline with a down body, a small body, then an up body closing into the first; an evening star mirrors it. Definitions differ on gaps and thresholds. The middle candle's size does not prove momentum stalled or that reversal follows.

See also: Doji, Engulfing pattern, Candlestick.

A rolling average of a chosen price series over a specified lookback.

A trailing moving average combines the current observation with earlier ones, using equal or specified weights. The input, lookback, weighting, and data treatment affect the line. Smoothing can delay its response to new movement, and the calculation by itself does not forecast the next price.

See also: Trend, Timeframe.

Running many tests, so some pass by chance.

At a one-in-twenty threshold, twenty independent tests of pure noise produce a passing result about 64% of the time. Nothing is wrong with running many tests. What breaks the arithmetic is reporting the one that passed without the number that were run.

See also: Data snooping, Sample size, Overfitting.

The line drawn through the reaction lows (or highs) of a reversal pattern.

In a head and shoulders it connects the two troughs. In a double top it runs through the trough between the peaks. A breach conventionally completes the pattern, but a chart does not show how many orders sit there and a neckline is neither a guaranteed trigger nor guaranteed support or resistance.

See also: Head and shoulders, Double top and double bottom, Support and resistance.

What one fund share's holdings are actually worth.

The value of everything a fund holds divided by the fund shares in issue. The fund's market price is a separate number set by supply and demand for the fund itself, and the gap between the two is a premium or a discount. Creation and redemption is the mechanism that normally keeps them close.

See also: ETF (exchange-traded fund), Premium or discount, Creation and redemption.

Longs minus shorts: which direction the account leans, and by how much.

Long $60,000 and short $40,000 is $20,000 net long by notional value. That does not by itself predict the effect of a broad market move because the two sides may have different betas, deltas, currencies, or factors. A low net figure can coexist with high gross exposure and substantial basis, correlation, liquidity, and short-specific risk.

See also: Gross exposure, Diversification, Factor exposure.

Accounting profit after recognized expenses, interest, and tax for a period.

Net income is a bottom-line measure under the relevant accounting rules, but not every economic cost or item in other comprehensive income appears in it. Earnings per share starts from income attributable to the applicable shareholders and share-count rules. Accruals and noncash items mean net income and cash movement can differ legitimately.

See also: Operating profit, Free cash flow, Price-to-earnings ratio.

A circumstance under which you have decided in advance not to trade at all.

Earnings tomorrow, a spread wider than a third of your target, the first fifteen minutes, three losses today, an instrument you have never traded. Each of these is a decision made once, in a calm state, so it does not have to be made again in a state where you would decide differently. No-trade conditions are the cheapest risk control there is, because a trade never taken cannot slip, gap, or be held too long.

See also: Event filter, Trading plan, Trade checklist.

How many jobs employers added or cut in a month, excluding farms.

The US establishment survey estimates payroll jobs, so a person with two payroll jobs can be counted twice. Initial estimates are revised as responses arrive and later benchmarked; the release also reports earnings and hours. Farm workers and several other categories are outside this payroll measure, while employment of people is measured separately in the household survey.

See also: Unemployment rate, Consumer price index (CPI).

Open, high, low, close: the four prices that summarize a period.

Open is the first traded price of the period, high and low are its extremes, close is the final price. Nearly every chart type is built from these four numbers.

See also: Candlestick, Timeframe.

A running total that adds a period's volume on up closes and subtracts it on down closes.

If the close is above the prior close, the usual calculation adds the period's entire volume; if it is below, it subtracts it, and an unchanged close commonly leaves the total unchanged. That sign assignment is not buyer-minus-seller volume and does not identify participant class or motive. Analysts compare OBV direction or divergence with price, but usefulness depends on the precise rule, data, horizon, and out-of-sample results.

See also: Accumulation/distribution line, Volume divergence, Volume.

The number of derivative contracts currently outstanding, not the number traded.

Volume counts contracts traded during a period; open interest counts contracts that remain open. Open interest rises when both sides create a new contract and falls when both close one, while transfers between an opening and a closing trader can leave it unchanged. It is reported for derivatives such as options and futures and does not identify whether the open positions are bullish, bearish, hedged, or speculative.

See also: Volume, Expectancy.

What is left after the cost of the product and the cost of running the business.

Computed before interest to lenders and before tax, which makes it the cleanest measure of how the business itself performs, independent of how it is financed. Two companies with identical operations and different amounts of debt report the same operating profit and very different net income.

See also: Revenue, Net income, Free cash flow.

Losses caused by failed processes, systems, or people rather than by the market.

Examples include an erroneous order, failed data feed, mishandled corporate action, fraud, cyber incident, or undocumented dependency. Historical return models may omit these losses. Checklists, access controls, reconciliation, tested limits, backups, incident response, and separation of duties can reduce operational risk but cannot eliminate it.

See also: Reconciliation, Kill switch, Audit log, Platform risk.

A contract giving the right, but not the obligation, to buy or sell at a set price by a set date.

A call gives its holder the right to buy and a put the right to sell under the contract's terms. A buyer who only holds the option can generally lose the premium and transaction costs; exercising it creates the risks of the resulting position. A seller accepts an obligation and can face much larger losses. Remaining time value generally erodes toward expiry, but price, volatility, rates, dividends, exercise style, and settlement terms all matter.

See also: Futures contract, Leverage, Volatility.

Noticeably more resting size on one side of the book than the other.

It compares displayed bid and offer size over specified levels. Orders can cancel, hidden liquidity is omitted, and fragmented feeds can show only part of the market. Some studies find short-horizon relationships in particular markets, but their duration, tradability, costs, and stability must be tested; imbalance alone is not a directional instruction.

See also: Level II, Spoofing, Iceberg order, Absorption.

The decisions about where your order is sent to be executed.

Between pressing buy and a fill, a broker chooses among exchanges, dark venues, and internal matching, and those choices affect what you pay. Some routes pay the broker for the flow, which creates a conflict of interest that best-execution obligations exist to constrain. The practical consequence for a retail trader is that the venue was decided for you, and the quality of that decision is largely invisible in your fill confirmation.

See also: Best execution, Fragmentation, Dark pool, Broker.

An indicator designed to fluctuate around a center or within a range.

Some oscillators are bounded, such as RSI from 0 to 100, while others fluctuate around zero without a fixed numerical ceiling. When a bound exists, it belongs to the formula rather than to market price. An extreme reading therefore does not by itself show participant exhaustion or require price to reverse.

See also: Relative strength index (RSI), Overbought and oversold, Indicator.

History deliberately held back, and looked at once.

Data that had no influence on the rule, used for a single honest estimate of what the rule does on something it has not seen. Adjusting the rule after seeing the result converts this data into in-sample data permanently, and there is no partial version of that.

See also: In-sample data, Validation data, Overfitting.

Judging a decision only by how it turned out.

The habit of calling a decision 'good' because it won and 'bad' because it lost. In markets, where randomness dominates short-term results, outcome bias teaches exactly the wrong lessons: it rewards reckless winners and punishes disciplined losers.

See also: Variance, Decision quality.

A candle whose range engulfs the previous candle's entire range.

Its high exceeds the prior high and its low falls below the prior low. That shows a wider range, not participant disagreement or future direction. The close is another observable feature whose usefulness must be tested rather than assumed.

See also: Inside bar, Engulfing pattern.

Trading through dealers rather than on a central exchange.

Over-the-counter structures, quotation tiers, reporting duties, and investor protections differ by jurisdiction and security. Some quoted issuers provide little current public information, and a delisted security may continue OTC with lower liquidity and wider spreads. The venue label does not by itself describe disclosure quality, dealer risk, or whether an exit is available.

See also: Delisting, Market maker, Regulatory filing.

Conventional labels for indicator readings beyond specified upper or lower thresholds.

For RSI, 70 and 30 are common thresholds, but indicators and settings use different conventions. The label describes the indicator's position under its formula; it does not show excess inventory, identify buyers or sellers, or mean price must reverse. Whether a threshold has predictive or trading value is an empirical question for a defined market, horizon, and rule.

See also: Relative strength index (RSI), Oscillator, Trend.

Holding a belief more strongly than the evidence behind it supports.

In trading it can show up as size: the position grows because this one feels different. Winning streaks can raise confidence even when the sample is too small to distinguish skill from variance. Predefined risk caps and recorded forecasts make confidence testable, but grading a setup should affect size only if that grading system has demonstrated useful calibration.

See also: The four stages of competence, Sample size, Setup quality.

Tuning rules until they fit past data, and only past data.

Add enough conditions and any history can be made to look profitable. Such a strategy has memorized noise instead of learning a pattern, and it typically fails immediately on data it has not seen.

See also: Backtest, Sample size, Edge.

Taking more trades than the strategy generates, usually to feel productive.

Trades outside a defined strategy are untested rather than automatically worse on every outcome. They still add spread, fees, slippage, risk, and another opportunity for error. Comparing actual trades with the rule set and measuring net results by trade type can show whether activity, not merely a recent loss, is the issue.

See also: Boredom trading, Slippage, No-trade condition.

The probability, assuming a specified null model, of a result at least as extreme as the one observed.

A p-value of 0.05 means that, if the null model and test assumptions hold, outcomes at least as extreme as the observation have 5% probability. It is not the probability that the strategy works or that the null is true, and it does not measure effect size. With twenty independent true-null tests at a 5% threshold, the expected number below the threshold is one, but the realized count can differ and dependent searches require more careful treatment.

See also: Hypothesis test, Multiple testing, Confidence interval, Data snooping.

Buying one instrument and shorting a related one, betting on the gap between them rather than on direction.

Matching long and short sensitivities may reduce some broad-market exposure, but notionals alone do not make it neutral. The relationship, hedge ratio, beta, borrow, dividends, financing, and two sets of execution costs all matter. Historical correlation or cointegration can change gradually or abruptly and does not guarantee spread convergence.

See also: Relative value, Hedge, Correlated risk.

Running a strategy live without money at stake.

Tests the operational half honestly: whether the signals fire, whether you can act in time, whether the data arrives. It cannot test the half that breaks people, because a paper drawdown is a number and a real one is your money, and no amount of pretending closes that gap.

See also: Backtest, Out-of-sample data, Trading plan.

Whether nearby settings work nearly as well.

A robust parameter sits on a plateau: change it by 10% and the result barely moves. A fitted one sits on a spike, best at exactly one value and poor either side, which is the signature of a number chosen to match past noise rather than a real effect.

See also: Overfitting, Backtest, Walk-forward analysis.

Seeing meaningful patterns in randomness.

The tendency that finds faces in clouds finds textbook formations in noise. Charts are especially prone to it: a long enough history contains every shape somewhere, and the eye searches until it finds one.

See also: Base rate, Overfitting.

Closing part of a position while leaving the rest to run.

Selling half at the first target reduces the average result of your winners and raises the share of trades that end green. That is a real trade-off rather than a free lunch: it buys the psychological ability to hold the remainder, at the cost of the tail that pays for a strategy where most trades lose. Whether it helps depends on your own records, which is a reason to record it as a rule rather than improvise it.

See also: Scaling, R-multiple, Trailing stop.

How broadly and heavily a move is occurring across the measured instruments.

A rise on high traded volume across many names is different from the same index move concentrated in a few names. Volume and breadth describe quantity and distribution, not the identities, beliefs, or agreement of the people trading. Participation measures are context, not a directional prediction.

See also: Market breadth, Relative volume, Volume.

An order that rests in the book and waits to be traded against.

A nonmarketable limit order can add displayed or hidden liquidity and may qualify for a maker rebate on some venues. A fill can avoid crossing the quoted spread, but it does not guarantee a profit equal to the spread: price may move adversely, fees vary, and the order may never fill or may fill only partly. Fill outcomes are conditional rather than a neutral sample.

See also: Aggressive order, Adverse selection, Queue position, Spread.

A result that depends on the route taken rather than only the destination.

An index can rise ten per cent, give it back, and finish flat while a double-long fund finishes down, because the larger fall was applied to a larger base. The same arithmetic that produces volatility drag on any compounded series is multiplied here by the leverage. In a steady trend it works the other way and the fund beats its multiple.

See also: Daily reset, Leveraged fund, Volatility drag.

The average winner divided by the average loser, both measured in R.

It answers one question: when you are right, how much more do you make than you lose when you are wrong? A payoff of 1.5 means the average winner is one and a half times the average loser. It is half of what decides whether an approach makes money, the other half being the win rate, and the two trade against each other constantly: methods that win often tend to win small, and methods that win rarely tend to need the large tail to survive at all.

See also: Expectancy, Win rate, R-multiple.

The distance between entry and stop, expressed per share or per contract.

For a long stock entered at $52 with a planned stop at $49.50, the price distance is $2.50 per share. Account risk divided by estimated loss per unit gives a baseline size, but contracts, FX, options, fees, slippage, gaps, and nonlinear payoffs require their own value and scenario calculations. The stop is not a guaranteed loss bound.

See also: Position size, Risk per trade, Structural stop.

A leveraged crypto contract with no expiry date.

A perpetual derivative has no fixed expiry and commonly uses funding payments plus an index, margin, and liquidation system to keep its price near spot. Leverage limits, legal availability, settlement, and protections vary by venue and jurisdiction. Liquidation can be rapid and may leave losses, fees, or deficits beyond the posted margin under some terms.

See also: Funding rate, Forced liquidation, Leverage, Maintenance margin.

A two-candle reversal where the second candle recovers past the midpoint of the first.

A piercing pattern follows a decline: a down candle, then an up candle closing above the midpoint of the first body. Dark-cloud cover is its mirror after an advance. Both are weaker relatives of engulfing. The recovery is partial rather than complete.

See also: Engulfing pattern, Candlestick.

Any candle whose wick dominates a small body.

An umbrella term for hammers, shooting stars, and similar geometry. It records an extreme away from the open-close body, but describing that as rejection is an interpretation of incomplete OHLC data. It is not a signal on its own.

See also: Hammer, Shooting star, Wick (shadow).

A conventional unit for measuring a currency pair's price change.

For many pairs one pip is 0.0001, and for many yen pairs 0.01; brokers may quote smaller fractional pips and some instruments use different conventions. Monetary value depends on pair, position size, account currency, and price. A small-sounding pip move can therefore create a large percentage gain or loss when leverage is high.

See also: Currency pair, Lot, Leverage.

Everything that can stop you acting: outages, failed logins, bad data, and rejected orders.

A broker, venue, network, device, or data feed can fail or reject an instruction, including during busy or volatile conditions. A contingency plan can list verified contact methods, order-status checks, and approved alternatives, but a second route and resting orders introduce their own risks and may not be available. Never assume that an exit can be sent or filled on demand.

See also: Kill switch, Custody risk, Stop order.

The price level or bucket with the greatest assigned volume in a volume profile.

The widest bar in a volume profile: the price bucket to which the selected data and calculation assign the greatest volume. It is a historical reference, not a magnet, forecast, or measurement of current resting supply and demand.

See also: Volume profile, Value area, High-volume node.

The sum of the planned loss to each open position's stop.

Four positions each planned to lose 1% at its stop have 4% of nominal heat. This is a useful budget, not a worst-case guarantee: gaps, slippage, changing correlations, nonlinear products, and shared factors can make realized losses larger. A heat cap complements per-trade limits by considering simultaneous exposure.

See also: Correlated risk, Risk per trade, Diversification.

A required announcement that a holder has crossed a size threshold.

Some jurisdictions and instruments require holders to report when specified ownership or control thresholds are crossed. The filing tells you what the rule requires as of a reporting date; it may not reveal the build path, current position, economic exposure, or hedges. Thresholds, deadlines, amendments, and exemptions differ.

See also: Institutional ownership, Regulatory filing, Float.

How much you commit to a single trade.

Position size is one of the main risk choices a trader can set before an order. It changes exposure to ordinary losses, gaps, costs, liquidity, and losing sequences; the same strategy can be tolerable at one size and financially unsustainable at another. Fill uncertainty and nonlinear products mean realized exposure still needs monitoring.

See also: Risk per trade, Leverage, Drawdown.

Holding for weeks to months, on the largest structure a chart shows.

Trading less often can make entry and exit friction a smaller share of a position's intended move, but spreads, market impact, borrow, financing, taxes, and gaps can still be material. Samples accumulate slowly, and each position remains exposed to events and changing conditions throughout the holding window.

See also: Swing trading, Sample size, Event-driven.

How much of the market is already on one side of a trade, which changes how it responds to news.

Crowded positioning can make exits less liquid and can contribute to asymmetric reactions, but available positioning data are partial, delayed, and often do not reveal hedges or motives. Good news can still produce a large rise and bad news need not trigger an unwind. Treat positioning as one conditional risk input, not a deterministic reaction model.

See also: Herd behavior, Short squeeze, Correlated risk.

A fixed sequence performed after a losing trade, before anything else.

A pause, order-status check, record of rule adherence, and emotional check can interrupt an impulsive follow-up trade. The useful duration and steps differ by person, and the routine is a behavioral guardrail rather than a guarantee or clinical treatment. Repeated inability to stop calls for professional support and removal of trading access, not a stronger checklist.

See also: Revenge trading, Kill switch, Emotional check-in.

A schedule set up in advance that sells an insider's shares automatically.

A US Rule 10b5-1 plan is intended to be adopted in good faith when the insider is not aware of material nonpublic information and to specify trades in advance, subject to current conditions and disclosures. Plans can be amended or terminated and the label does not prove compliance or make the transaction meaningless. Other jurisdictions use different rules.

See also: Insider transaction, Regulatory filing.

Trading sessions before the open and after the close.

Extended sessions can have fewer displayed orders, wider spreads, different order rules, and prices that differ from the later regular-session open. Conditions vary by instrument and event, so an extended-hours move should be evaluated with its volume, spread, venue coverage, and liquidity rather than treated as a forecast.

See also: Market hours, Liquidity, Volume.

The gap between a fund's market price and the value of its holdings.

A market price above estimated net asset value is a premium and one below is a discount. Stale or asynchronous underlying prices, market closures, transaction costs, taxes, liquidity, and creation constraints can all contribute. Because NAV can be an estimate, an apparent gap is not automatically an arbitrage opportunity or proof of mispricing.

See also: Net asset value, Creation and redemption, Tracking error.

Better prices trade first. Among equal prices, whoever arrived first.

On a price-time venue, more competitive prices rank first and earlier orders rank ahead at the same price, subject to the venue's detailed rules. Other markets use pro-rata, size, dealer, or hybrid priority. Improving a limit by one tick can gain price priority but accepts a less favorable execution price, so matching rules must be checked rather than assumed.

See also: Queue position, Limit order book, Matching engine, Passive order.

The share price divided by earnings per share.

A high ratio can reflect growth expectations, low risk, or temporarily depressed earnings; a low one can reflect perceived value, risk, or expected earnings decline. Trailing, forward, basic, diluted, adjusted, and negative-earnings treatments differ. Comparisons are most interpretable among reasonably similar businesses using consistent definitions, but the ratio is not a value verdict.

See also: Net income, Shares outstanding, Market capitalisation.

The secret that controls a crypto wallet, and effectively is ownership.

A private key can authorize transactions for its address; a seed phrase commonly derives one or more keys and should not be treated as an ordinary password. Multisignature, smart wallets, custodians, issuer controls, and recovery designs can distribute or limit authority. Base-layer transfers are often hard to reverse, but protections and remedies depend on the network, asset, wallet, and jurisdiction.

See also: Self-custody, Custody risk, Centralized exchange.

A target describing behavior you perform rather than a result you receive.

"Take only setups meeting all four written conditions" is a process goal; "make $1,000 this month" is an outcome goal. The difference that matters is that a process goal can be satisfied on a day the market gives you nothing, so it never forces a trade in order to be met. It also produces the records that eventually answer whether the approach has an edge, which is the question no outcome target can settle.

See also: Decision quality, Trading plan, Trading journal.

Everything you made divided by everything you lost, across a set of trades.

A profit factor of 1.35 means gross gains were 1.35 times gross losses in the sample. Below 1 indicates a net trading loss before any costs omitted from those gains and losses; exactly 1 indicates break-even on that basis. Like expectancy, the ratio does not become trustworthy merely because it describes the aggregate: sample size, dependence, regime coverage, outliers, and costs still matter.

See also: Expectancy, Payoff ratio, Sample size.

Waiting for price to return toward the level it left, buying the retest rather than the break.

You get a better price and a tighter stop, because the level you are wrong against is right beside you. The cost is the trades you never get into: a strong move often does not come back, and waiting for a retest means missing it entirely. Neither entry is better. They trade fill quality against participation rate, and a plan has to say which one it uses so the choice is not made by mood.

See also: Breakout entry, Throwback and pullback, Entry trigger.

Promoters hype an asset they already own, then sell into your buying.

Promoters make misleading claims or manufacture excitement, often in group chats or social posts, while holding the asset. They then sell into demand they helped create. Price can fall sharply when promotion or buying fades, leaving later buyers exposed; the exact path is not guaranteed, but the deceptive conflict is the defining risk.

See also: Conflict of interest.

Adding to a winning position in decreasing amounts, with the stop moved up each time.

A pyramiding rule may use smaller additions and raise a planned stop to keep nominal stop-based risk near a cap. Recalculate the whole position after each add: gaps, slippage, correlation, and product mechanics can still make actual loss exceed that amount. Favorable movement supports rather than proves a thesis, and adding changes both exposure and average price.

See also: Scaling, Averaging down, Anti-martingale.

Where your resting order sits in line at its price level.

On a price-time venue, earlier orders generally fill before later ones at the same price, although hidden size, amendments, venue rules, and feed limitations can obscure exact position. A trade at your limit does not guarantee your order filled. Back-of-queue orders may require more opposing flow and can experience different fill and adverse-selection patterns, which must be measured rather than assumed.

See also: Price-time priority, Adverse selection, Passive order, Limit order book.

A result measured in units of the amount risked, rather than in currency.

If you risked $200 and made $600, that is a 3R win, whether the account is $2,000 or $2 million. Recording results in R is what makes a hundred trades comparable: it strips out account size, position size, and the price of the instrument, leaving only how the trade actually performed against what it cost to be wrong. Expectancy expressed in R is the number that says whether a strategy is worth repeating.

See also: Risk-to-reward ratio, Expectancy, Risk per trade.

The share of one series' movement that the fitted line accounts for.

In a standard regression with an intercept, an R-squared of 0.3 means the fitted model accounts for 30% of the sample variation around the mean. Adding predictors cannot lower ordinary in-sample R-squared, which is why adjusted and out-of-sample measures matter. A low or high value alone says neither that a strategy is tradable nor that data leaked; uncertainty, specification, stability, payoff, and costs decide that.

See also: Linear regression, Residual, Overfitting, Data leakage.

Price oscillating between a recognizable floor and ceiling.

A range is defined from repeated movement between selected boundaries over a chosen period. Those boundaries are historical observations rather than exact order locations or promises of another reversal, and different drawing rules can produce different ranges.

See also: Trend, Support and resistance.

Buying near the bottom of an established range and selling near the top, while the range holds.

It is a form of mean reversion whose boundaries and breach rule must be defined rather than drawn after the fact. A breakout can cause a loss, but the last trade before a range ends is not mechanically guaranteed to lose: exits, gaps, false breaks, and whether an order was open all matter. Costs and a prewritten regime-change rule belong in the test.

See also: Range, Mean reversion, Support and resistance.

Profit or loss recognized after all or part of a position is closed or settled.

Closing or settling a position commonly realizes the change between its recognized cost and proceeds. Partial closes, derivatives marked to market, income, fees, tax rules, and accounting methods can change when and how a result is recognized, so broker and tax reports may not use identical figures.

See also: Unrealized profit or loss.

Trading back to target weights after prices move them apart.

Without flows or rebalancing, relative performance changes portfolio weights and can increase concentration in outperformers. Restoring targets generally sells assets above target and buys those below it. Frequency, bands, taxes, spreads, market impact, and whether target risk itself changed determine the trade-off; rebalancing does not guarantee better returns.

See also: Risk parity, Volatility weighting, Concentration risk, Diversification.

Weighting the most recent outcomes far more heavily than the longer record.

A short run of losses can feel like proof that a strategy broke, while a short run of wins can feel like proof it was solved. How informative either run is depends on the strategy's expected distribution, dependence, and whether conditions changed. Reviewing complete records against predefined thresholds is more reliable than judging from memory alone.

See also: Sample size, Strategy hopping, Trading journal.

Checking that what your system thinks it holds matches what the broker holds.

Positions, cash, transactions, and open orders are compared with broker or custodian records on a schedule. A mismatch can reflect timing, identifiers, corporate actions, fees, or an error and should be investigated before relying on affected risk figures. Reconciliation is an important control, but it does not catch failures that leave both compared records consistently wrong.

See also: Idempotency key, Audit log, Kill switch, Operational risk.

Total profit divided by the worst drawdown suffered along the way.

It asks how much you were paid for the worst pain the strategy inflicted. A recovery factor of 1 means the total profit merely matched the size of the deepest hole, which is a poor trade for the years spent in it. Like Calmar it depends on one historical extreme, so it should always be read alongside how long the sample was and whether it contained a genuine crisis.

See also: Calmar ratio, Drawdown, Profit factor.

A document a listed company is required to publish, on a schedule or on an event.

Applicable rules can require periodic reports, material-event notices, insider transactions, and ownership disclosures. Publication channels, deadlines, coverage, and access differ by jurisdiction and form, so not every filing reaches every participant at the same moment. Reading primary documents reduces reliance on summaries but does not by itself create a trading edge.

See also: Annual report, Material announcement, Insider transaction.

How an instrument is performing measured against a benchmark rather than against its own past.

A stock up 3% on a day the index rose 5% underperformed that benchmark for the day. Relative-strength measures can use ratios or return differences over specified windows. Some momentum strategies rank on them, but a position losing relative strength is an exit only if the tested rule defines it that way.

See also: Momentum, Market index, Sector rotation.

A 0-100 oscillator comparing the size of recent gains to the size of recent losses.

Wilder's RSI uses smoothed average gains and losses over a chosen lookback and maps their ratio to 0–100. A high value records that gains outweighed losses under that calculation; it does not identify buyers, measure demand, or establish that continuation or reversal comes next. Despite the name, RSI compares an instrument with its own recent changes rather than with another instrument or benchmark.

See also: Oscillator, Overbought and oversold, Indicator divergence, Relative strength.

Trading the price difference between related instruments rather than the price of either one.

The general form of pairs trading can compare bond maturities, an index and a basket, or economically related assets. A spread can move because the thesis works, because financing and carry accrue, or because the assumed relationship changes or breaks. Being early and being wrong can look identical for a long time, so leverage, model risk, costs, and a suspension rule matter.

See also: Pairs trading, Leverage, Counterparty risk.

Today's volume measured against what is normal for this instrument at this time of day.

Usually a ratio: 2.0 means twice the usual traded quantity for the comparison window. The time-of-day part matters because volume is not spread evenly through a session. Comparing 10am volume against a full-day average makes every morning look quiet and every afternoon look busy. Relative volume shows unusual activity; it does not reveal how many participants traded or why they did so.

See also: Average daily volume, Volume, Volume dry-up.

What a model got wrong: the gap between what it predicted and what happened.

Residuals are where a model tells you the truth about itself. If they are scattered randomly, the model has captured what it can. If they show a pattern, drifting over time, clustering in volatile periods, all large in one regime, then the model is systematically wrong in a way its headline fit statistic hid. Looking at residuals is the cheapest available check and the one most often skipped, because the headline number is the one people want to report.

See also: Linear regression, R-squared, Model drift, Stationarity.

The full shape of every result a strategy produced, not just its average.

Two strategies can share an average return and have nothing else in common: one grinds out small gains and occasionally loses everything, the other loses small constantly and occasionally wins big. The distribution is the picture that separates them, and it is what an average, by construction, throws away. Every summary statistic in this world is an attempt to describe some feature of this shape in one number, and every one of them loses something in the process.

See also: Skewness, Kurtosis, Standard deviation, Expectancy.

Trading to recover a loss rather than because a setup appeared.

The tell is that the reason for the trade references the previous trade. Size is usually raised to make the recovery arrive faster, so the largest position of the day gets taken in the worst state of the day, which is the combination that ends accounts rather than merely damaging them. It is what a daily loss limit written as a count of losses is actually for: not to cap the money, but to end the sequence before the third decision.

See also: Drawdown limit, Kill switch, Risk per trade.

Everything customers paid, before any cost is subtracted.

The top of the chain, and the figure that says least about whether a business works. A company can grow revenue while losing more money on every sale. It is the starting number that gross profit, operating profit, net income and free cash flow are each computed from by subtracting a different set of costs.

See also: Operating profit, Net income, Free cash flow.

Combining several shares into one, at a proportionally higher price.

Ten shares at 1 can become one share at 10, mechanically preserving proportional ownership and value at the effective moment apart from fractional-share treatment and market movement. Companies may use reverse splits for listing-price compliance or other capital-structure reasons. The arithmetic is neutral, while costs, liquidity, options adjustments, and the circumstances around it can matter.

See also: Stock split, Shares outstanding, Delisting.

A published figure being restated later, or expectations being changed before the announcement.

Economic statistics are frequently revised after release, sometimes substantially, so the number that moved the market can turn out not to have been the number. Forecast revisions run the other way: analysts move their estimates in the weeks before an announcement, which shifts what counts as a surprise. Both matter for the same reason, which is that the reference point a market is measuring against is itself moving.

See also: Consensus, Surprise, Bad tick.

Money you could lose completely without changing your life.

Money left over after your needs, your goals, and your emergency fund are funded. Money whose total loss would not cost you rent, food, school fees, or safety. Rent money, borrowed money, and emergency savings are never risk capital, no matter how good an opportunity looks.

See also: Emergency fund, Expectancy.

A shared driver that explains why many instruments move together.

Size, value, momentum, quality, and the market itself are the classic ones. Factors are useful because a portfolio of twenty names can turn out to be one bet on a single factor wearing twenty labels, which is the most common way diversification is imaginary. They are also the subject of an enormous published literature that is heavily affected by multiple testing, so the existence of a documented factor is weaker evidence than the volume of papers suggests.

See also: Factor exposure, Diversification, Concentration risk, Multiple testing.

The probability that a losing sequence takes the account below the point it can recover from.

It depends on the definition of ruin, starting capital, position sizing, payoff distribution, costs, dependence between trades, changing conditions, and the time horizon, not only win rate and average win or loss. Smaller risk per attempt generally reduces estimated ruin probability, but gaps, leverage, and clustered losses can make simple independent-trade formulas too optimistic.

See also: Drawdown, Expectancy, Risk per trade.

Sizing positions so each contributes similar risk, not similar money.

Equal capital in a lower-volatility and a higher-volatility asset generally does not create equal estimated risk. Full risk-parity calculations use both volatility and correlation so marginal contributions target equality; simple inverse-volatility weighting is only an approximation. Estimates can change sharply in stress, and some implementations use leverage on lower-volatility assets, adding financing and liquidity risk.

See also: Volatility weighting, Portfolio heat, Leverage, Correlated risk.

The money you accept losing if a single trade fails.

Often expressed as a percentage of account equity using an entry, planned exit, and position size. Setting a cap before entry limits nominal planned loss and reduces confidence-based sizing, but gaps, slippage, fees, liquidity, and product mechanics can make realized loss larger.

See also: Position size, Stop order, Drawdown.

A reference return treated as having negligible default risk for a chosen currency and horizon.

Short-dated sovereign bills in the matching currency are a common proxy, not literally risk-free in every respect. The rate is an opportunity-cost benchmark used in measures such as Sharpe. A strategy returning less with greater risk is unattractive on that comparison, although taxes, liquidity, horizon, liabilities, and currency can make other comparisons relevant. The chosen proxy and period should be stated.

See also: Sharpe ratio, Benchmark, Expectancy.

What you stand to gain compared with what you risk.

Risking $100 to make $300 is a 1:3 ratio. It only means something alongside how often the trade works: a great ratio with a poor hit rate can still lose money, which is why expectancy matters more than either number alone.

See also: Expectancy, Expected value (EV), Risk per trade.

The nightly interest adjustment for holding a currency position.

A leveraged spot-FX position held across the broker's rollover time may receive or pay a financing adjustment based on both currencies, broker terms, day count, and market rates; multi-day adjustments can occur around weekends or holidays. It can materially affect a long-held position. Carry strategies can have skewed crash risk, but their realized distribution is not universally negative-skewed.

See also: Interest-rate differential, Carry, Currency pair, Financing cost.

A project's insiders taking the money and abandoning it.

Insiders may withdraw controlled liquidity, misuse upgrade keys, mint tokens, or sell concentrated holdings and abandon the project. Conduct can be fraudulent, but legal treatment depends on facts and jurisdiction. Anonymous control, unaudited code, unlocked insider holdings, concentrated permissions, and removable liquidity are risk indicators, not proof that a project will fail.

See also: Token unlock, Smart-contract risk, Pump and dump.

Divide 72 by a growth rate to estimate years to double.

A mental shortcut, not exact arithmetic. At 8% a year, roughly 9 years to double (72 ÷ 8). At 24% interest on a debt, roughly 3 years for the amount owed to double.

See also: Compounding.

The twelve fields that turn a strategy family into something a stranger could execute identically.

Market, timeframe, regime, setup, trigger, volume condition, stop, exit, size, event exclusions, transaction-cost assumptions, and suspension rule. A rule set is complete when somebody who has never spoken to you could trade it and get the same trades you would. Anything less is a description of a preference, and the reason it matters is mechanical rather than moral: an incomplete rule set cannot be backtested, cannot be simulated, and cannot be reviewed, because there is no fixed thing to test, run, or check compliance against.

See also: Trading plan, Setup, Entry trigger.

How many results you're judging from.

Small samples can be dominated by chance and may omit important conditions. The amount of data needed depends on effect size, variability, dependence, and regime coverage, and costs must be included before calling anything an edge.

See also: Variance, Edge.

Building or reducing a position in planned pieces rather than all at once.

Scaling in or out divides execution across prewritten prices, times, or conditions. Later evidence may support a thesis without proving it, and an added position changes total risk. Planned rules can be tested and sized; improvised additions or exits can instead reflect loss aversion, anchoring, or uncontrolled risk.

See also: Partial exit, Position size, Portfolio heat.

Taking many very short trades for small moves, usually within minutes.

Small targets and frequent decisions can make spread, fees, slippage, latency, queue position, and market impact large relative to gross profit. Attention and execution requirements depend on whether the method is discretionary or automated. Viability is an empirical net-of-cost question, and paper fills can be especially optimistic at short horizons.

See also: Turnover, Slippage, Spread.

A claimed tendency for returns to depend on the time of year, month, or week.

Some seasonal patterns have plausible mechanisms, such as tax calendars or growing cycles, while searching many calendar slices raises the chance that one looks strong by luck. A mechanism helps but does not validate a rule; the test also needs multiple-testing controls, out-of-sample evidence, realistic costs, and stability across periods.

See also: Overfitting, Sample size, Backtest.

A grouping of companies doing broadly the same kind of business.

Standard classifications split the market into about eleven sectors: technology, health care, financials, energy, industrials, consumer staples, consumer discretionary, materials, real estate, utilities, and communication services. The grouping is useful because companies in one sector often move together in response to the same news, which means a portfolio that feels diversified across ten names can be one bet if all ten sit in one sector.

See also: Diversification, Market index, Stock (share).

Moving capital between industry groups as their relative strength changes.

A rule ranks sectors using defined data and shifts exposure as those rankings change. Relative performance does not reveal who is moving money, and sector membership explains different amounts of a stock's return in different periods. Rotation can create concentration, turnover, and whipsaw, so its benchmark, rebalance schedule, costs, and risk limits must be explicit.

See also: Sector, Relative strength, Concentration risk.

Choosing what to include after seeing what it does.

Picking the instruments, the date range, or the exclusions once you already know which choices flatter the result. Each individual decision usually has a reasonable defense, and the pattern is that every one of them happened to point the same way.

See also: Data snooping, Survivorship bias, Backtest.

Holding your own keys instead of leaving assets with an exchange.

Withdrawing to a wallet you control reduces ongoing custody exposure to an exchange, while adding key loss, phishing, malware, transaction, smart-contract, and inheritance risks. Venue exposure can return whenever assets are deposited or traded there, and token issuers or protocols may retain controls. Neither custody model is risk-free; verify the actual legal and technical arrangement.

See also: Private key, Custody risk, Centralized exchange.

Banks and brokers that provide markets, research, and execution.

Sell-side firms can make markets, execute client orders, underwrite securities, lend, and publish research. Their roles, conflicts, and regulations vary. Market makers manage inventory and hedges while earning or losing from spreads, fees, rebates, and price changes; a quote is not necessarily a simple directional view.

See also: Buy side, Market maker, Best execution, Broker.

When ownership and cash actually change hands after a trade.

Executing a trade and completing it are different moments. Until settlement, some brokers restrict how the proceeds can be reused, which is why an account can show a balance that is not yet fully usable.

See also: Broker, Cash account.

The repeatable configuration of conditions you wait for before you will consider a trade.

A setup is a pattern of circumstances you can name and recognize again: a regime, a level, a structure, and usually a participation requirement. It is not yet a reason to act. The distinction that makes a plan work is setup against trigger: the setup says the situation qualifies, and the trigger says now. Collapsing the two is how people enter early, before the thing they were waiting for actually happened.

See also: Entry trigger, Thesis, Setup quality.

Grading how completely an opportunity meets predefined setup rules before entry.

The value is not the letters but a checklist applied before the result is known. Missing a required condition should normally disqualify the setup rather than be rescued by confidence. Any rule that varies size by grade needs evidence that the grades predict net outcomes; until then, keep size at or below the ordinary risk limit.

See also: Position size, Risk per trade, Trading plan.

One of several kinds of share in the same company, carrying different rights.

A company can issue more than one class, and the classes differ in votes, in distributions, or in where they rank if the company is wound up. The common arrangement puts ten or more votes on a founder class and one on the listed class, so an outside holder can own most of the economics and none of the control. It is disclosed rather than hidden, and it matters whenever a decision is actually put to a vote.

See also: Voting rights, Stock (share), Shares outstanding.

A sale of shares after the initial offering, which may or may not raise money for the company.

A primary follow-on offering issues new shares and can dilute existing percentages while raising capital for the company. A secondary sale by existing holders transfers already-issued shares and normally does not raise company cash or increase shares outstanding. Value effects depend on price, amount, costs, use of proceeds, seller, signaling, and available alternatives, not one inferred motive.

See also: Dilution, Cash burn, Shares outstanding.

How many shares of a company currently exist.

The denominator under every per-share figure, and it moves. It rises when the company issues stock or employee awards vest, and it falls when the company buys shares back and cancels them. A per-share number from two years ago may be measuring a differently sized slice than today's, which is why any change in a per-share figure deserves the question of whether the numerator or the denominator moved.

See also: Float, Market capitalisation, Dilution, Buyback.

Return above the risk-free rate, divided by the volatility taken to get it.

Sharpe divides average excess return by the standard deviation of returns. The calculation treats upside and downside deviations alike; the ratio itself does not require normal returns, but common annualisation and statistical comparisons can rely on assumptions that fail with skew, autocorrelation, changing volatility, or fat tails. It should be paired with drawdown, tail, cost, and liquidity measures over comparable samples.

See also: Sortino ratio, Risk-free rate, Standard deviation, Skewness.

A small body low in the range with a long upper wick, after an advance.

The high is far above the body and the close lies near the lower part of the range. The same geometry after a decline is conventionally called an inverted hammer. The candle does not identify who traded or guarantee that the preceding move reverses.

See also: Hammer, Wick (shadow), Pin bar.

The number of shares currently sold short and not yet bought back.

Short interest is reported on a schedule and is often compared with float or shares outstanding, whose definitions and dates must match. Open shorts normally require later delivery or closing, but positions can transfer and corporate events can change the obligation. The measure omits timing, entry price, hedges, and motive, so it does not predict a squeeze or direction by itself.

See also: Days to cover, Short squeeze, Borrow availability.

Selling borrowed shares, hoping to buy them back cheaper.

Profits if the price falls. Unlike buying, the potential loss has no natural ceiling, because a price can keep rising indefinitely. Shorting also involves borrowing costs and the risk that the lender recalls the shares.

See also: Leverage, Stock (share).

Rising prices forcing short sellers to buy, which raises prices further.

Short sellers who close must buy shares, and recalls or margin liquidations can make some of that buying urgent. If it helps push price higher, further covering can create a temporary feedback loop. High short interest alone neither proves that covering caused a rise nor guarantees a squeeze, and fundamentals, news, liquidity, options hedging, and ordinary demand can all contribute.

See also: Short selling, Borrow availability, Margin call.

Whether the rare, extreme results sit on the winning side or the losing side.

Positive skew has a longer or heavier right tail; negative skew has a longer or heavier left tail. Some trend-following implementations have produced positive skew and some short-option strategies negative skew, but strategy labels do not determine a distribution. Skewness must be measured over a representative net-return sample and interpreted with tail size, dependence, and drawdown.

See also: Return distribution, Kurtosis, Tail risk, Risk of ruin.

The difference between the price you expected and the one you got.

It can result from quote changes, latency, order size, queue position, market impact, or limited depth. Slippage is often larger in fast or thin markets but can be favorable or adverse, and the reference price used to measure it must be stated.

See also: Market order, Liquidity, Spread.

The risk that the code running a protocol has a flaw somebody can exploit.

A bug, unsafe upgrade, oracle failure, governance action, or unexpected interaction can cause loss. Audits reduce uncertainty but cannot prove safety. Some systems have admin controls, pauses, insurance, forks, or legal remedies and others do not, so reversibility and recourse must be checked rather than assumed.

See also: Decentralized exchange, Custody risk, Platform risk.

Like Sharpe, but only counting downside volatility as risk.

Built on the observation that an investor is not harmed by upside surprises, so measuring them as risk is strange. It divides excess return by downside deviation instead of total deviation. It fixes one of Sharpe's problems and inherits the rest: it still summarizes a whole distribution in one number, still says nothing about the largest loss, and is still computed over a period somebody chose.

See also: Sharpe ratio, Calmar ratio, Skewness, Standard deviation.

A small body with visible wicks on both sides.

The period's high and low extend beyond a relatively small open-close body. It records a close near the open relative to the range, but not the intraperiod path, participant balance, motive, or next direction.

See also: Doji, Candlestick.

Placing orders you intend to cancel, to make others think demand exists.

A large visible bid is entered with no intention of trading it, other participants react to the apparent demand, and the order is pulled once the price has moved. It is illegal in major jurisdictions, explicitly prohibited by name in US law since 2010, and has produced criminal convictions and nine-figure penalties. What makes an order spoofing is the intent to cancel, which is why traders have been convicted on the evidence of their own messages and code.

See also: Layering, Market manipulation, Order book imbalance, Level II.

The gap between the best displayed bid and ask.

Crossing the spread to buy and then sell immediately creates a round-trip trading cost before fees and market movement. Wide spreads often accompany lower liquidity or greater uncertainty, but order type, fill price, size, fees, and price improvement determine the cost actually paid.

See also: Bid, Ask (offer), Liquidity, Slippage.

Two series that track each other for no reason connecting them.

Search enough pairs of series and you will find ones that match beautifully by luck alone: the number is real, the relationship is not. Two things trending in the same direction over the same period correlate strongly whether or not they have anything to do with each other, which is why a correlation found by searching needs a mechanism, an out-of-sample period, or both before anyone should size a position on it.

See also: Correlation, Cointegration, Data snooping, Overfitting.

A token designed to hold a fixed value, usually one US dollar.

Stablecoins are widely used for trading and settlement, but prevalence differs by venue and pair. Support can include cash and securities reserves, crypto collateral, contractual claims, algorithms, or combinations. The peg, redemption, reserves, issuer, custody, smart contracts, and regulation all carry risk; the name is not a guarantee of value.

See also: Depeg, Centralized exchange, Counterparty risk.

A measure of how far results typically fall from their average.

Take each result's distance from the mean, square those distances, average them using the chosen population or sample formula, then take the square root. Large outliers have substantial influence. Standard deviation describes dispersion for any finite sample, but by itself does not describe skewness or tail probabilities; using it in a normal-distribution model can underestimate extremes when returns are heavy-tailed.

See also: Variance, Volatility, Kurtosis, Return distribution.

Whether a series' statistical behavior stays the same over time.

Under weak stationarity, the mean and variance are constant and covariance depends on the lag rather than the date. Many price levels are nonstationary, so analysts often model transformations such as returns, but returns can still have changing volatility, tails, or relationships. Stationarity is an assumption to test and monitor, not a property guaranteed by differencing.

See also: Autocorrelation, Structural break, Model drift, Market regime.

A unit of ownership in a company.

A common share is a residual ownership claim whose voting, dividend, and liquidation rights depend on its class and governing documents. Shareholders can benefit if the business grows or makes distributions, but profits do not create an automatic payment and both price and principal can fall to zero.

See also: Bond, ETF (exchange-traded fund).

Dividing existing shares into more, cheaper shares.

A 2-for-1 split doubles each holder's share count and mechanically halves the per-share price, leaving proportional ownership and value unchanged at that moment, apart from rounding or cash paid for fractional shares. An unadjusted chart can resemble a sudden 50% fall, so the data's corporate-action treatment must be checked.

See also: Adjusted price, Stock (share).

An order that becomes active when its specified trigger condition occurs.

A standard stop commonly becomes a market order after the broker or venue's trigger rule is met; trigger rules can use trades or quotes and vary by provider. It can automate an exit but cannot guarantee the fill price: a sell stop can fill below its trigger after a gap, and a buy stop can fill above it.

See also: Market order, Gap, Slippage.

Abandoning an approach after a losing run and adopting a new one, repeatedly.

Frequent switching can prevent a comparable sample from accumulating and can mistake ordinary variance for evidence that a rule failed. A written review schedule, minimum evidence standard, and predefined suspension conditions help separate justified changes from reactions to a recent outcome. A minimum trade count alone is not enough if regimes or rules changed.

See also: Sample size, Expectancy, Recency bias.

Working out what your current positions do under a specified bad day, before it happens.

Specify shocks such as a broad decline, sector move, gap, volatility jump, correlation change, or unavailable broker, then estimate their effect on current positions. Results depend on pricing, liquidity, nonlinear exposures, and scenario assumptions, so they are conditional estimates rather than exact worst cases. Multiple scenarios reveal concentrations that isolated position limits can miss.

See also: Portfolio heat, Tail risk, Gross exposure.

A point where a market's behavior changed and stayed changed.

A rule change, a new participant type, a shift in who provides liquidity, a regime that ends. Unlike a regime shift that reverses, a structural break makes the earlier data a description of a market that no longer exists. It is the reason more history is not automatically better history, and identifying one is largely a question about mechanism rather than statistics, because the test that detects a break cannot tell you whether it will revert.

See also: Stationarity, Market regime, Model drift, Cointegration.

A stop placed where the chart's structure says the thesis has failed.

Examples include beyond a chosen swing low or back inside a broken range. The analyst, not the market, chooses which structure invalidates the setup, and different definitions produce different distances. Position size should adapt to the planned exit and estimated slippage so account risk remains within its limit.

See also: Invalidation, Volatility stop, Position size, Risk per trade.

Continuing because of what has already been spent, which cannot be recovered by continuing.

Past time and unrecoverable cost should not justify taking new risk by themselves. A useful review asks whether the position still fits its thesis and portfolio today, while also considering taxes, transaction costs, liabilities, hedges, and constraints that can make closing different from opening a fresh position. Those current consequences are not sunk costs.

See also: Anchoring, Time stop, Invalidation.

Price areas near which the market previously stalled or reversed.

Zones, not exact lines, where price previously stalled or reversed under the chosen drawing rule. Traders may use them as candidate reference areas, but the chart does not reveal current resting orders or guarantee another stall, reversal, or break.

See also: Range, Trend.

The difference between an announced figure and a specified expectation.

A positive numerical surprise is above the chosen estimate and a negative one is below. Price need not move in the same direction or in proportion: guidance, estimate dispersion, quality of earnings, valuation, positioning, liquidity, and simultaneous disclosures can matter. Pre-event price movement does not by itself reveal what was anticipated.

See also: Consensus, Revision, Gap.

Studying only the things that lasted.

Testing a strategy on companies that still exist ignores every one that failed and was removed, so the results describe a group selected for having survived. The same error makes successful traders look more representative than they are.

See also: Backtest, Sample size.

Holding for days to weeks, aiming to capture one leg of a larger move.

Longer intended moves can make one round trip a smaller share of the target than in very short-term trading, but this depends on spread and instrument. Decisions may still require market-hours attention. Holding overnight creates gap and event exposure, so risk scenarios must allow actual loss to exceed a stop-based estimate.

See also: Position trading, Gap, Timeframe.

Slicing an order evenly through time to average out the price.

A basic time-weighted schedule distributes quantity across time without targeting the market's volume curve. Implementations can vary slice sizes, intervals, limits, and participation constraints to reduce signaling or adapt to liquidity. It does not guarantee the benchmark price or a complete fill.

See also: VWAP, Transaction cost analysis, Implementation shortfall, Market impact.

The rare, large move that ordinary volatility measures underestimate.

Many financial return series have heavier tails and changing volatility, so large moves can occur more often than a simple independent normal model implies. A volatility estimate does not automatically understate every bad case; the model, horizon, data, and assumptions decide that. Conservative sizing, stress tests, liquidity planning, limits, and suitable hedges are complementary responses rather than guarantees.

See also: Volatility, Stress test, Hedge.

A model of learning: not knowing what you lack, then knowing, then doing it deliberately, then doing it automatically.

Unconscious incompetence, conscious incompetence, conscious competence, unconscious competence. It is useful for locating yourself in a learning curve, and it is worth knowing it is a teaching model rather than a research finding. It is frequently mislabelled as the Dunning-Kruger effect, which is a separate and specific result about low performers overestimating their own rank.

See also: Decision quality, Trading journal.

More smoothing usually reduces short-term variation while slowing response to new data.

Shorter lookbacks or heavier recent weights usually track changes sooner; longer lookbacks usually produce a steadier line. Neither setting guarantees fewer losing trades or better forecasts, because results depend on the data, market, horizon, signal rule, and costs. Parameter choice should therefore be specified before evaluation and tested out of sample.

See also: Lag, Moving average, Overfitting.

The claim about what price should do next, and why, stated before the trade.

A thesis is a testable statement with a reason and horizon: for example, "if price closes above the range on unusually high activity, I expect continuation over the next three sessions." It is a hypothesis, not a claim about hidden buyers or sellers. Writing it makes the evidence, invalidation, and review criteria explicit before the result is known.

See also: Counter-thesis, Invalidation, Setup.

A harami or an engulfing pair followed by a third candle that closes further in the new direction.

Three inside up is a bullish harami followed by an up candle closing above the first candle's open. Three outside up is a bullish engulfing pair followed by an up candle closing above the second close. The down versions mirror both after a declared advance. The third candle makes each name available one period later than the pair it extends.

See also: Harami, Engulfing pattern, Confirmation.

Three consecutive strong candles in the same direction.

Conventional rules look for three sizeable same-direction bodies with progressively higher or lower closes and often opens within prior bodies. The formation is identified only after three periods and records a move already made; it does not prove pressure continues.

See also: Marubozu, Trend.

Three same-direction candles, then one opposite candle that closes beyond where the three began.

In the bullish version, three up candles each close higher, then a down candle opens at or above the third close and closes below the first open. Conventional sources label it a continuation despite the last candle's direction. The label is a convention, not a tested result, and the pattern is rare.

See also: Three white soldiers / three black crows, Engulfing pattern, Chart pattern.

Price returning to the broken level shortly after breaking it.

A return toward a level after an upward break is commonly called a throwback; after a downward break it is often called a pullback, although terminology varies. The return may not happen, and reaching the old level does not guarantee a hold, reversal, or continuation. Break and retest entries therefore have different fill, timing, and risk characteristics that require separate testing.

See also: Support and resistance, Chart pattern, Confirmation.

A count of quote or price updates in a feed, used as an activity proxy when centralized trade volume is unavailable.

In fragmented markets such as spot foreign exchange, a platform may count updates received from its own feed instead of units traded across the whole market. Tick volume measures those observed updates, not transaction quantity: an update need not identify a trade and contains no traded size. Its relationship with broader activity varies by feed, market, and period.

See also: Volume, Bad tick, Liquidity.

The running record of trades that actually happened.

A feed of reported executions showing fields such as price, size, time, venue, and trade condition. Coverage, timestamps, corrections, aggregation, and permitted reporting delays vary, and an aggressor-side label may be reported or inferred from trades and quotes. The tape records transactions rather than participant identity, motive, complete market-wide activity, or future direction.

See also: Level II, Absorption, Limit order book, Dark pool.

An exit taken because the move did not happen in the window it was supposed to happen in.

A thesis should state a horizon before entry. If its prewritten window passes without the required move, a time stop exits even though the price stop was not reached. The horizon is part of the hypothesis rather than a universal claim about how quickly every breakout or setup must work.

See also: Invalidation, Thesis, Portfolio heat.

How long an account stays below a previous high before making a new one.

Depth is only half of a drawdown and duration is the half that decides whether anybody survives it. A 9% fall recovered in a fortnight is an incident. The same 9% taking eight months is a period during which every rule gets questioned and most approaches get abandoned. Measured in trades or in calendar time, it is the honest answer to how long you would have to keep doing this while nothing appeared to be working.

See also: Drawdown, Drawdown limit, Expectancy.

How much time each candle on the chart represents.

A daily chart's single candle contains a whole day of five-minute candles. Trends at different timeframes can point in different directions without either being wrong. They answer different questions.

See also: OHLC, Trend.

A scheduled date when previously locked tokens become sellable.

A vesting or lockup schedule may make additional tokens transferable, although disclosures can change and unlocked holders need not sell. An unlock can increase circulating supply or potential selling pressure, but price impact depends on size, recipients, liquidity, expectations, hedges, and demand. Verify the contract and current schedule rather than relying on a calendar alone.

See also: Fully diluted value, Dilution, Float.

Starting from the broad market and narrowing to one instrument.

Market, then sector, then name, then chart. Its strength is that it stops you falling in love with a company inside a market that is falling, because the first question filters out most candidates before you have looked at any of them. Its weakness is the mirror of that: a genuinely exceptional individual situation gets screened out because its neighborhood looks poor, and you never find out what you passed on.

See also: Bottom-up analysis, Market regime, Sector rotation.

The variability of a fund's return difference from its benchmark over time.

Tracking difference is the fund return minus benchmark return; tracking error commonly measures how much that difference varies across periods. Fees, taxes, cash, sampling, rebalancing, securities lending, and trading costs can affect one or both. A fee is a known drag, but it does not mechanically equal each year's observed lag because other contributions can offset or add to it.

See also: Expense ratio, Premium or discount, ETF (exchange-traded fund).

The fixed list of conditions confirmed before an order is sent.

Its value is not that it contains anything you did not know. It is that it forces the knowledge to be applied in the exact moment it is hardest to apply, which is when a move is already running and you are afraid of missing it. Checklists work in the same way and for the same reason in every field that has adopted them: they defend against skipped steps under pressure, not against ignorance.

See also: Trading plan, No-trade condition, Trading journal.

A shorthand for a default behavioral tendency: cautious, impulsive, or selectively aggressive.

Borrowed from how poker describes players. Tight-passive hesitates and lets opportunities pass, biased toward loss aversion. Loose-aggressive forces trades and swings between extremes, biased toward overconfidence. Tight-aggressive waits and then acts decisively. The point is not to be sorted into a box but to notice that your default exists and costs you something specific, which you can then plan around.

See also: Setup quality, Trading plan, Decision quality.

Trying to profit from shorter-term price changes.

Days, hours, sometimes minutes. It can be done with real discipline and it is genuinely difficult: costs are paid on every round trip, and most short-term traders lose money after those costs. What makes it trading rather than gambling is a repeatable edge and a positive expectancy, not the speed.

See also: Investing, Gambling, Edge, Expectancy, Swing trading, Scalping.

A pause imposed by an exchange or regulator.

Trading can be suspended for pending news, volatility controls, operational problems, or regulatory reasons. Executions in the affected security or venue are ordinarily unavailable during the halt even if orders can be entered, canceled, or routed under specific rules. Stops cannot execute until an eligible market reopens, when price may be far from the trigger.

See also: Stop order, Gap.

A record of the reasoning, the emotion, and the result of each decision, kept separately.

Beyond profit and loss, a journal can preserve the information, forecast, risk plan, state, actions, and result of each decision. Contemporaneous records make process and outcome easier to review separately, but they remain self-reported evidence and work best with broker records, predefined fields, and periodic analysis.

See also: Decision quality, Outcome bias, Trading plan.

A written statement of what you trade, when, how much you risk, and when you stop.

Its purpose is to move decisions out of the moment when you are least able to make them well. A plan written calmly on Sunday and followed on Wednesday is one person's considered judgment surviving contact with another person's adrenaline. It is not a prediction, and it does not have to be sophisticated. It has to be specific enough that somebody else could tell whether you followed it.

See also: Risk per trade, Risk-to-reward ratio, Trading journal.

A stop whose trigger ratchets with favorable price movement under preset rules.

For a long position, the trigger can rise as the reference price rises and remain fixed when it falls; a short position works in reverse. Broker rules differ on the reference price, adjustment method, trigger, and order submitted afterwards. It automates an instruction, not a profit floor: gaps, slippage, and partial fills can produce a result worse than the trailing distance.

See also: Stop order, Risk per trade, Slippage.

Measuring what execution actually cost against a chosen reference price.

Compare achieved prices and quantities with a benchmark such as decision price, arrival price, VWAP, or another stated reference, then include explicit and implicit costs. Delay, spread, fees, market impact, and opportunity cost can dominate in different orders. Because benchmark and methodology materially change the result, both must accompany the number.

See also: Implementation shortfall, VWAP, TWAP, Market impact.

A repeated staircase of higher highs and higher lows, or lower ones.

An uptrend makes successively higher peaks and higher troughs. A downtrend does the reverse. Trends are descriptions of what price has done, and they end without warning.

See also: Range, Timeframe.

Entering in the direction a market is already moving and holding while it continues.

It does not require predicting a target, but entry still expresses a conditional belief that an established move may persist. Many trend-following rules accept frequent small losses in pursuit of fewer large winners, although the exact win rate and return shape depend on the rules, market, horizon, sizing, and costs. Entries, exits, and risk control all affect the result.

See also: Trend, Trailing stop, R-multiple.

A consolidation whose highs and lows converge, drawn as two lines meeting at an apex.

An ascending triangle has a roughly flat upper boundary and rising lows; a descending triangle has a roughly flat lower boundary and falling highs; a symmetrical triangle has both boundaries converging. Those are geometric descriptions chosen with tolerances, not observations of participant beliefs or intent. Direction after the pattern is not fixed by its name and must be evaluated with a defined rule and evidence.

See also: Consolidation, Apex, Chart pattern.

How much trading you do relative to the size of the account.

Higher turnover applies spreads, slippage, commissions, taxes, and market impact more frequently, although the exact cost depends on what and how you trade. Tracking turnover beside net expectancy helps show whether additional activity is adding enough gross return to cover its incremental friction.

See also: Slippage, Spread, Overtrading.

Two or more candles sharing almost the same high or low.

The candles record repeated highs or lows within a chosen tolerance. This is a geometric observation rather than proof that a participant defended the level, and its predictive value depends on an exact definition, context, and test.

See also: Support and resistance, Candlestick.

The share of people looking for work who have not found it.

In the US headline measure, unemployed people are those without work who are available and actively searched recently, plus certain people on temporary layoff; the rate divides them by the labor force. Someone who stops searching can leave both numerator and denominator, but that does not guarantee the rate falls. Definitions differ across countries, so read it with employment and participation measures.

See also: Non-farm payrolls.

A paper gain or loss on a position you still hold.

It moves with every price tick and can shrink or disappear entirely before you act. Treating unrealized gains as spendable income is a classic way to be surprised later.

See also: Realized profit or loss.

A measure of how much movement the options market is pricing in for the near future.

Derived from options prices, it reports expected volatility rather than direction. It usually rises when markets fall, which is why it gets called a fear gauge, and that nickname oversells it: it measures the price of protection, not an emotion. A high reading says movement is expected to be large, and says nothing about which way.

See also: Volatility, Option, Market index.

The average price of a session weighted by the volume traded at each price.

Volume-weighted average price. Every included trade contributes in proportion to its size, so the result describes the average recorded transaction price for the session rather than the midpoint of its range. It does not reveal the average current holder's cost: every transaction has both a buyer and a seller, and positions can be opened, closed, or transferred repeatedly. VWAP is commonly used as an execution benchmark; predictive use requires separate evidence.

See also: Anchored VWAP, Volume, Moving average.

The middle slice, used to choose between candidates.

Sits between development and the final holdout, and is for deciding which of several surviving rules to keep. Every comparison leaks a little information into your choices, so after enough of them the validation data has been partly fitted too, which is why it is not the last word.

See also: In-sample data, Out-of-sample data, Multiple testing.

The price band containing the bulk of a profile's volume, conventionally about 70%.

A range calculated to contain a chosen share of the profile's assigned volume, often 70%. The percentage and calculation method are conventions, not laws of markets, and different platforms can produce different boundaries. It describes the selected historical sample rather than fair value or future support and resistance.

See also: Volume profile, Point of control, Low-volume node.

A loss level that should only be exceeded on a small percentage of days.

A one-day 99% VaR of 20,000 estimates, under the chosen model and current inputs, a 1% chance of losing more than 20,000 over one day. It does not say that exactly one day in each hundred will breach or how severe a breach will be. Historical, parametric, and simulation methods make different assumptions, so backtesting and stress testing are essential.

See also: Expected shortfall, Kurtosis, Tail risk, Drawdown.

How much individual results scatter around the average.

The reason a good decision can lose and a bad decision can win. High variance means single results tell you very little about the quality of the process that produced them.

See also: Expected value (EV), Sample size, Outcome bias.

How much and how quickly a price moves around.

A measure of movement, not direction. High volatility means larger swings in both directions, which widens the range of outcomes. It is a description of risk, not evidence of an opportunity.

See also: Variance, Range, Liquidity.

A price range outside which trading pauses automatically.

Under the US LULD plan, price bands are calculated around a reference price and executions outside them are prevented; persistent limit states can trigger a trading pause. Rules, durations, order handling, reopening mechanics, and exceptions are venue- and product-specific. A reopening can gap through a stop trigger, but a limit order still retains its stated execution-price bound.

See also: Trading halt, Market-wide circuit breaker, Stop order.

The gap compounding opens between an average return and the actual result.

Lose 50% and gain 50%: the average is zero and you are down 25%. That gap is volatility drag, and it grows with the size of the swings, not with their direction. It is the mathematical reason a smoother strategy can end up ahead of a higher-averaging volatile one, and the reason cutting volatility is worth real return rather than being purely a comfort measure.

See also: Geometric return, Arithmetic return, Compounding, Drawdown.

Whether a market is currently moving more or less than its own normal, and for how long that has been true.

Volatility clusters: quiet periods follow quiet periods and violent ones follow violent ones, far more than chance would produce. That persistence is what makes the regime worth naming, because it means the recent past is genuinely informative about the near future in a way direction is not. Everything downstream depends on it: stop distances, position sizes, target sizes, and which strategy families have room to work.

See also: Volatility, Market regime, Volatility stop.

A stop distance scaled to a measure of the instrument's recent variability.

A multiple of ATR or another volatility estimate gives more room to an instrument that has recently moved more. This can reduce exits caused by routine fluctuation, but no multiple prevents them and recent volatility can change abruptly. A complete rule specifies the estimator, lookback, multiple, structural context, and position size.

See also: Structural stop, Volatility, Position size.

Taking smaller positions in things that move more.

Dividing a risk budget by an estimate of volatility gives smaller weights to instruments estimated to move more. This can equalize standalone volatility, but portfolio contributions also depend on correlations. The estimate is backward-looking and can be low before a sudden change, so caps, correlation-aware calculations, and stress tests remain necessary.

See also: Risk parity, Position size, Volatility, Per-unit risk.

How much was actually traded in a period.

Volume records the quantity of shares, contracts, or units traded under a market's counting convention. It does not count unique participants, reveal their motives, or by itself predict whether price will rise or fall.

See also: Liquidity.

Price making a new extreme while a volume-based measure does not.

For example, price can set a higher high while a specified volume measure does not. This is a relationship between two derived observations; it does not identify who traded, why activity differed, or when or whether price will reverse. Any predictive use depends on the exact measure, comparison rule, market, horizon, costs, and out-of-sample evidence.

See also: On-balance volume, Confirmation, Money Flow Index.

Recorded volume declining during a pullback or pause.

A pullback or pause with progressively less traded volume shows declining activity relative to the chosen baseline. It does not prove that sellers are absent or unmotivated, and it is not a directional signal on its own: quiet markets can stay quiet or move sharply when new orders arrive.

See also: Relative volume, Volume, Trend.

A histogram assigning recorded or estimated volume to price levels or buckets.

A time-based volume bar groups activity by period; a volume profile groups it by price. Some feeds can use transaction-level data, while chart platforms may allocate lower-timeframe bar volume across price buckets, so range, row size, source, and method affect the shape. High- and low-volume areas describe the selected data; they do not reveal participant intent or guarantee how price will behave on a return.

See also: Point of control, Value area, High-volume node.

The say a share carries in the decisions a company puts to its owners.

One of the three things traveling under the word share, alongside the residual claim and any distribution. Rights vary by class, and preferred shares usually carry none at all. Whether the class you can buy carries the votes is one of the two questions that describe a company's share structure.

See also: Share class, Stock (share).

Repeatedly fit on one window, test on the next, then roll.

Instead of one split, a sequence of them: fit on 2016 to 2018 and test on 2019, then fit on 2017 to 2019 and test on 2020, and so on. It tests whether the fitting procedure keeps working, which is a stronger and more relevant question than whether one particular fitted rule did.

See also: Out-of-sample data, Parameter stability, Backtest.

Trading with yourself to manufacture the appearance of volume.

Buying and selling the same instrument with no change in beneficial ownership, producing volume that represents no real transfer. It is used to make an instrument look actively traded, and it is prohibited. It is a particular problem in venues without surveillance, where reported volume can be substantially fictitious, so volume figures from unregulated markets should never be treated as equivalent to exchange-reported volume.

See also: Market manipulation, Volume, Spoofing.

The thin line showing the period's extremes outside the open-close body.

An upper wick spans from the higher of open or close to the high; a lower wick spans from the lower of them to the low. A long wick records intraperiod movement and a close away from the extreme, but OHLC alone does not reveal the path, who acted, or what happens next.

See also: Candlestick.

The share of trades that end profitably.

On its own it says very little. A 30% win rate is profitable if the wins are large enough, and a 90% win rate loses money if the rare losses are catastrophic. Win rate is only interpretable next to average win and average loss.

See also: Expectancy, Risk-to-reward ratio, Expected value (EV).