Glossary

Backtesting terms A to Z

Last reviewed: 14 September 2026·Tradelyze

A backtesting glossary is a list of the words used in backtests, optimization, prop firm rules and statistics, each defined in one or two plain sentences. It includes the labels on a Tradelyze report, such as Match Rate and Ruin Probability. Every entry links to the page that explains the term in full.

Which terms describe the backtest itself?

A backtest takes trading rules, price data, trading costs and a starting balance, and produces a trade list, an equity curve and summary numbers. The terms in this group name those inputs and outputs. Get them right before reading any result, because a backtest with the wrong costs or the wrong timeframe describes a different strategy from the one you would trade. What is backtesting? covers the whole process.

• BacktestRunning a strategy's exact rules on past prices to see which trades it would have taken and how the account would have changed. A backtest describes the past, not the future. What is backtesting?
• OHLCVOpen, high, low, close and volume: the five numbers recorded for every bar of price data. Tradelyze asks you to upload an OHLCV CSV file, because it holds no license to redistribute market data. Getting OHLCV data
• Bar (candle)One period of price, such as five minutes or one day, summarized by its open, high, low and close. A bar does not record the order in which price reached its high and low, which is why fills inside one bar are an assumption. Why trades do not match
• TimeframeThe length of each bar on the chart, such as 5 minutes, 1 hour or daily. Your price file must use the same timeframe as the chart you exported trades from, and a script that reads another timeframe needs an extra file for it. Extra timeframe files
• CommissionThe fee a broker or exchange charges on each filled order: a flat fee per order, a fee per contract, or a percentage of the order's value. At $1.00 per contract per fill, one contract in and out costs $2.00. What costs do per trade
• SlippageThe gap between the price an order was meant to fill at and the price it actually got. TradingView adds it as a fixed number of ticks against you on market and stop order fills, not limit orders. How much slippage to use
• PyramidingThe maximum number of entries a strategy may stack in one position. TradingView's Pine Script documentation gives a default of 1, meaning a strategy cannot add to a position it already holds. Raising it raises the size at risk. What pyramiding does
• Initial capitalThe account balance a backtest starts with. Percentage profit and drawdown are measured against it, so the same dollar loss reads 2% on $50,000 and 10% on $10,000. Tradelyze takes it from the script's strategy() declaration. Why script defaults matter
• EquityThe account balance plus the profit or loss on trades that are still open. An equity curve plots it over time; a curve that dips deeply on the way to the same end point is riskier than a smooth one. Maximum drawdown
• DrawdownA fall in the account from its highest point so far to a later low. Maximum drawdown is the largest such fall, usually shown as a percentage of the peak: $11,200 falling to $10,360 is a 7.5% drawdown. Maximum drawdown

Which terms describe trade matching?

Trade matching checks that a second backtest of your Pine Script takes the same trades as your TradingView report. Tradelyze does this first, at a stage called Baseline Matching, because every later number describes its own re-run. If the re-run trades differently, the optimization, walk-forward and prop firm results are about a different strategy. These labels appear on the Conversion Result card and the Best Metrics card. TradingView backtest accuracy explains each one with a worked card.

• Baseline MatchingThe first stage of a Tradelyze run. Tradelyze backtests your script at its default settings on its own Pine Script backtester and compares the trades with your TradingView export, one by one. Why Tradelyze re-runs your backtest
• Match RateThe share of trades the re-run and your TradingView export agree on: Matched divided by the larger of the two trade counts in the comparable window. Tradelyze colors it green at 90% or more, amber at 80% or more and red below, a display choice with no published standard. Match Rate explained
• MatchedPairs of trades that go the same direction and enter within five minutes of each other, one from each backtest. How trades are paired
• TV OnlyTrades in your TradingView export, inside the comparable window, that the re-run did not take. Why trades do not match
• BT OnlyTrades the re-run backtest (BT) took that have no TradingView partner, including re-run trades from before TradingView's first exported trade. Match Rate explained
• WarmupThe opening stretch of bars an indicator needs before it gives values; a 200-bar moving average says nothing on bar 50. Trades from before the other backtest's first trade are treated as warmup and left out of Match Rate. Indicator warmup
• TV Comparable FromThe first timestamp from which Tradelyze's figures can fairly be compared with TradingView's own for the same strategy. Before it, both backtests are still filling indicator history. It appears on the Best Metrics card and in Top Trials. TV Comparable From
• First TradeWhen the recommended settings placed their first trade. It is normally later than the start of your price file, because indicators need warmup bars first. The span actually traded runs from First Trade to the end of the data. Best Metrics tiles
• Continue anywayA button on the Optimization stopped card, shown when the match was too low. It optimizes the re-run as it is. In the card's words, every number the run reports will then describe that build, not the one your TradingView export came from. When to click Continue anyway

Which terms describe optimization?

Optimization is backtesting the same strategy many times with different input settings and keeping the settings that score best. Every optimization term matters for one reason. The winning settings were chosen on the same prices they are scored on, so their numbers are the most flattering in the report. The labels below appear in Tradelyze's Run Optimization window and on the optimization results card. Strategy optimization covers each one in depth.

• Parameter (input)A number or choice the strategy's rules depend on, such as an RSI length or a stop-loss percentage. In Pine Script these are the input() values on the Inputs tab. What optimization changes
• TrialOne full backtest with one combination of settings. The more trials a search runs, the more likely its best result is partly luck. How many trials to run
• Parameter range (Fixed, Min, Max, Step)The values the optimizer may try for one input. Min and Max bound the range, Step sets the gap between values, and Fixed is the single value used when the input is switched off. An RSI length from 8 to 30 in steps of 2 gives 12 values. Setting ranges
• Pareto frontThe set of trials that no other trial beats on every goal at once. Tradelyze scores each trial on profit, Sharpe ratio and maximum drawdown, so one trial on the front may earn more while another risks less. Pareto front explained
• NSGA-IIThe search method Tradelyze uses, through the open-source Optuna library: a search that breeds new settings from the best previous ones. Tradelyze's Automatic trial budget is never fewer than 60 trials. How Tradelyze's optimizer searches
• Top TrialsA table of up to five of the best-ranked trials, so you can compare the recommended settings with their closest alternatives. Reading the Top Trials table
• FeasibleA Top Trials column. YES means every rule of the selected prop firm's rule set was checked against that trial and passed; NO means a rule failed or could not be checked. Feasible YES or NO
• Recommended parametersThe settings Tradelyze picks from the completed trials: for each selected firm, the highest-ranked trial that passes all of that firm's checked rules, or the closest attempt when none passes. They are in-sample, so treat their metrics as a ceiling, not a forecast. Best versus recommended parameters
• Verify TrialA Tradelyze tool that copies your Pine Script with one trial's settings written in as the defaults and opens a new strategy draft, so you can re-run it in TradingView. Submitting that draft costs 1 credit. What Verify Trial does

Which terms describe validation?

Validation asks whether an optimized result is likely to hold up, or was luck and overfitting. Walk-forward analysis scores settings on data the search did not tune on. The robustness checks stress-test the tuned trade list itself. These terms cover both, including the badges on Tradelyze's walk-forward and robustness cards. None of them proves a strategy will make money; each one shows a different way a backtest can fool you. Start with walk-forward efficiency and the robustness score.

• In-sampleThe stretch of price history a strategy's settings were tuned on. In-sample results show how well the settings fit that past. In-sample vs out-of-sample
• Out-of-sampleA separate stretch of price history the settings never saw, used only to score them once they are frozen. The out-of-sample result is the more honest one and is usually weaker. In-sample vs out-of-sample
• Overfitting (curve fitting)Settings that describe the random noise in one price history rather than behavior that repeats. An overfit strategy looks excellent in-sample and fades on new prices. Overfitting and sample size
• Sharpe ratioAverage return above a risk-free rate divided by the standard deviation of returns: profit per unit of swing. Tradelyze leaves out the risk-free rate, and shows Sharpe (Bar) and Sharpe (Daily) on its Best Metrics card. Which Sharpe Tradelyze shows
• Walk-forward windowOne tuning segment followed by one test segment. Tradelyze's default is 2 rolling windows, each tuned on its first 70% of bars and tested on the remaining 30%. Walk-forward analysis
• WF EfficiencyTradelyze's walk-forward efficiency: the average annualized Sharpe ratio on the test windows divided by the average on the tuning windows, shown as a ratio. 0.50 means half the tuned edge survived on unseen data. Above 0.5 is labeled Generalizes well, 0 to 0.5 Likely overfit, and below 0 Inverted — lost out-of-sample. The tile is blank on runs graded with fixed settings, which do not compute it. WF Efficiency labels
• PASS, FAIL and NO VERDICTThe walk-forward badge. PASS needs WF Efficiency above 0.5, more than 60% of usable test windows profitable, at least one usable window, and no more than half the windows excluded. FAIL means a condition failed; NO VERDICT means that method produced no pass or fail, so read the per-window results. Older results read Confirmed or Not Confirmed. What the badge requires
• ROBUST, ACCEPTABLE, MARGINAL, FRAGILEThe robustness card's verdict. ROBUST needs all five checks run and passed and a score of at least 80; ACCEPTABLE needs 70 or more, MARGINAL 50 or more, and anything lower is FRAGILE. Robustness verdicts
• Monte Carlo simulationReplaying a strategy's real trades many times in different orders and mixes to see how deep the drawdown could have gone. Tradelyze builds 1,000 versions by default from short runs of your trades, so losing streaks stay together. Tradelyze's Monte Carlo numbers
• Ruin ProbabilityThe share of simulated runs whose drawdown went past the total drawdown limit of the rule set you selected, or 10% when no firm is named. Tradelyze passes its Monte Carlo check below 20%, a line it chose rather than a published standard. Ruin Probability
• MC Max DD Real→P95Two drawdowns: your trades in the order they happened, then the 95th percentile across the simulated runs. 7.4% → 12.9% means a less lucky order could plausibly have fallen 12.9%. Size an account against the second number. MC Max DD Real→P95
• Permutation p-valueThe share of coin-flip versions of your trades, with each win or loss randomly flipped, whose Sharpe ratio matched or beat yours. The bar starts at 0.05 and gets stricter the more settings were tried, and the test needs at least 20 closed trades. Permutation test
• Deflated Sharpe ratio (DSR)A probability from 0 to 1 that your Sharpe ratio beats what the best of that many trials would reach by luck. Tradelyze reads over 0.95 as Significant and needs at least 5 closed trades. Deflated Sharpe ratio
• Minimum backtest lengthThe history needed to trust a Sharpe ratio found by a search of that size. Tradelyze compares Required Years with Available Years, and Required Years is never under one year. Minimum backtest length
• Parameter sensitivityA check that nudges the winning settings slightly and re-runs the backtest. Tradelyze reads Stable when the Sharpe ratio falls less than 20% on average and the original Sharpe ratio was positive. Parameter sensitivity
• NOT RUN and NOT RESOLVEDGrey badges for a robustness check that did not happen (NOT RUN) or ran without reaching an answer (NOT RESOLVED). Nothing failed, but the verdict can then be ACCEPTABLE at best. An optimization under 50 trials skips parameter sensitivity. What to do about NOT RUN

Which statistics words appear in reports?

Statistics words in a backtest report describe a sample of trades or days, not a law of the market. Each definition below uses a trading example and the plain meaning, not the textbook proof. Two habits prevent most misreadings: check how many trades sit behind a number, and ask whether one extreme trade is pulling it. The five trades used in the examples are constructed, not measured: −$150, −$100, +$50, +$300 and +$400.

• Average (mean)Add up every value and divide by how many there are. The five example trades average $100 each. One very large winner can lift the average a long way. Expectancy, the average result per trade
• MedianThe middle value once the values are sorted: half are smaller and half are larger. The five example trades have a median of $50. The NIST/SEMATECH statistics handbook notes that extreme values distort the mean but not the median.
• Standard deviationHow widely values spread around their average. Daily results of +$200, −$200, +$200 and −$200 average $0 with a standard deviation of about $231; +$10, −$10, +$10 and −$10 also average $0 but give about $12. It is the bottom half of the Sharpe ratio. Sharpe ratio
• PercentileThe value a given share of results falls at or below. Of 1,000 simulated drawdowns, about 950 are no deeper than the 95th percentile, so only about 1 in 20 goes past it. Choosing a confidence level
• p-valueThe probability of a result at least this good if a stated no-edge assumption were true. The American Statistical Association's 2016 statement warns that p-values “do not measure the probability that the studied hypothesis is true.” What the permutation p-value means
• Sample sizeHow many observations a number rests on; in a backtest, usually the closed trades. Tradelyze's Trade Count tooltip treats fewer than 30 trades as unreliable, a rule of thumb with no primary source. How many trades you need
• AnnualizedConverted to a yearly figure so results from different test lengths compare. A daily Sharpe ratio of 0.10 annualizes to 0.10 × √252 = 1.59 using 252 trading days; Tradelyze's Sharpe (Daily) counts the sessions a year in your own data instead. How to annualize

Which terms describe prop firm rules and futures?

A prop firm challenge is a paid evaluation with a profit target and loss limits. The terms below name those rules, plus the futures contract details that turn price moves into dollars. Prop firm rules decide pass or fail on fixed limits. A small definition gap, such as trailing versus static drawdown, can end an account you paid for. Firms word and change their rules often, so read your firm's current terms. Prop firm rules and backtest metrics maps each rule to the backtest number that predicts it.

• Prop firm challengeA paid evaluation account from a proprietary trading firm. Passing its profit target without breaking its loss limits leads to a funded or payout-eligible account. How prop firm challenges work
• Daily loss limitA cap on how much the account may lose within one trading day. Tradelyze checks it on the max_daily_drawdown row, from closed trades, per calendar day in your Exchange Timezone, as a share of the starting account size. Daily loss limit
• Static drawdownA total loss limit whose floor is set once from the starting balance and never moves. A $50,000 account with a $2,500 static limit fails only below $47,500. Static versus trailing
• Trailing drawdownA loss limit whose floor follows the account's highest balance upward and never moves back down. On a $30,000 account with a $1,500 trailing limit, a climb to $31,000 lifts the floor from $28,500 to $29,500. Trailing drawdown
• End-of-day trailing drawdownA trailing limit that raises its floor only from the balance at the daily close, so a gain that fades before the close never moves it. It is gentler than a floor that trails during the session. Intraday versus end-of-day trailing
• Trailing Realtime (and Trailing EOD, EOD Balance, Static)The four drawdown types in Tradelyze's custom rule form. Trailing Realtime follows the highest balance after each closed trade; despite the name, it follows closed trades, not live equity. Trailing EOD follows the highest end-of-day balance, EOD Balance measures from the starting balance at each day's close, and Static from the starting balance. How Tradelyze checks each type
• Consistency ruleA cap on how much of the profit may come from the single best day. Tradelyze divides the best day's profit by total profit, so a $1,200 best day in $3,000 of profit is 40%. Some firms state their cap as a share of the profit target instead, so check the wording. Prop firm consistency rule
• Profit targetThe gain an account must reach to pass a phase, usually a percentage of the starting balance. Tradelyze compares the backtest's total return with it. Prop firm rule types
• Minimum trading daysThe number of days an account must trade before it can pass. Tradelyze counts the days on which at least one trade closed, dated in the Exchange Timezone. What Tradelyze checks
• Exchange TimezoneThe strategy setting Tradelyze uses to decide where one trading day ends and the next begins, which moves daily loss, consistency and trading-day figures. It is separate from Chart Timezone, the clock your uploaded files are written in. Which timezone to choose
• Qualifies and Not FeasibleThe badge on each prop firm card. Qualifies means every rule Tradelyze checked passed; Not Feasible means at least one failed. Tradelyze does not check evaluation time limits, news-trading or weekend-holding rules. What Qualifies covers
• Point valueHow many dollars one futures contract gains or loses when price moves one full point. Tradelyze's futures lookup table lists $2 for Micro E-mini Nasdaq-100 (MNQ) and $50 for E-mini S&P 500 (ES); confirm against the exchange. Point value and tick size
• Tick sizeThe smallest step a price can move. MNQ moves in ticks of 0.25 points, so one tick is worth 0.25 × $2 = $0.50 per contract. From ticks to dollars at risk
• Continuous contractIn TradingView's help center's words, “an artificial financial instrument derived by linking multiple individual futures contracts with different expiration dates.” MNQ1! follows the front contract, and the back-adjustment setting changes its past prices. Continuous futures and back-adjustment

How do these terms fit together on one report?

The terms fit together as a reading order. First check whether the re-run is your strategy, then how many trades back it. Then check whether the result held up, and finally whether it survives the firm's limits. Here is one constructed report, not measured data, for a 5-minute MNQ strategy on a $50,000 account with a $2,000 (4%) trailing drawdown limit.

Read together, the terms tell a different story from the Qualifies badge. The real trade order squeezed under the limit, but a quarter of plausible orders did not, and the tuned edge mostly faded on unseen data. For your money, that means not paying the challenge fee yet. Test a smaller position size first, since dollar drawdowns shrink with size, though not exactly in proportion (why size does not scale exactly). Then deal with the walk-forward FAIL without re-optimizing until it passes. What if the verdicts disagree? covers this exact mix.

Frequently asked questions about backtesting terms

Sources

• NIST/SEMATECH, e-Handbook of Statistical Methods, section 1.3.5.1 “Measures of Location”: the mean and median definitions and the note that extreme values distort the mean but not the median. itl.nist.gov/div898/handbook/eda/section3/eda351.htm, retrieved September 2026.
• NIST/SEMATECH, e-Handbook of Statistical Methods, section 1.3.5.6 “Measures of Scale”: standard deviation as the square root of the variance, with an N−1 divisor. itl.nist.gov/div898/handbook/eda/section3/eda356.htm, retrieved September 2026.
• NIST/SEMATECH, e-Handbook of Statistical Methods, section 7.2.6.2 “Percentiles”. itl.nist.gov/div898/handbook/prc/section2/prc262.htm, retrieved September 2026.
• Ronald L. Wasserstein and Nicole A. Lazar, The ASA Statement on p-Values: Context, Process, and Purpose, The American Statistician 70(2), 2016, pages 129–133, doi:10.1080/00031305.2016.1154108. The wording of principle 2 was checked against the American Statistical Association's own copy of the statement, retrieved September 2026.
• TradingView, Pine Script User Manual, “Strategies”: pyramiding defaults to 1 and limits the open trades from strategy.entry() calls in one position. tradingview.com/pine-script-docs/concepts/strategies, retrieved September 2026.
• TradingView Help Center, “Strategy properties”: slippage is a value in ticks added to the fill price of market or stop orders, and pyramiding affects only strategy.entry() orders. tradingview.com/support/solutions/43000628599, retrieved September 2026.
• TradingView Help Center, “What are 1! and 2! continuous futures contracts?”: the continuous contract definition quoted on this page and 1! as the front contract. tradingview.com/support/solutions/43000483493, retrieved September 2026.
• Tradelyze implementation, reviewed September 2026: the on-screen labels Match Rate, Matched, TV Only, BT Only, TV Comparable From, First Trade, Continue anyway, Feasible, Verify Trial, WF Efficiency, PASS, FAIL and NO VERDICT, MC Max DD Real→P95, Ruin Probability, NOT RUN and NOT RESOLVED, Qualifies and Not Feasible, and the custom rule drawdown types Static, EOD Balance, Trailing Realtime and Trailing EOD; the result-card tooltip text for First Trade and TV Comparable From; and the MNQ and ES entries in the futures lookup table. The thresholds quoted for Match Rate colors, walk-forward, robustness and Monte Carlo checks are Tradelyze's own and have no primary source.
• All worked examples on this page, including the five example trades, the standard deviation figures, the prop firm account figures and the report in the worked-example table, are constructed illustrations, not measured data.