Glossary
Win rate and expectancy
Last reviewed: 25 September 2026·Tradelyze
Win rate is the percentage of closed trades that made money, and expectancy is the average profit or loss per trade once the size of wins and losses is included. A strategy that wins 90% of the time with 1-point wins and 10-point losses still loses 0.1 points per trade: (0.9 × 1) − (0.1 × 10).
In plain English
Win rate tells you how often a strategy's trades made money, but not how much each one made or lost. Expectancy adds the size of the wins and the losses, so it tells you whether the average trade made or lost money. A strategy can win most of its trades and still lose money when its few losses are large.
New to this? Start with what backtesting is.
Rule you can use
Expectancy = (win rate × average win) − (loss rate × average loss). If it is negative, the strategy loses money however high the win rate is. Tradelyze does not show expectancy; download a trial's trades and average the P&L column.
Loss rate is the share of trades that lost money. Enter the average loss as a positive number. P&L means profit and loss, and a trial is one set of strategy settings that the Tradelyze optimizer tested. Step-by-step instructions are in What is expectancy and how do I use it?
Is a high win rate good?
Win rate is the share of closed trades that finished in profit. A very high win rate is not an achievement on its own. Experienced traders tend to read one as a warning sign. The reason is arithmetic. You can push a win rate up by shrinking the profit target and widening the stop. Those changes can make the strategy worse while the statistic looks better.
Take a widely repeated trading-forum example. A strategy takes one point of profit on the ES contract (the E-mini S&P 500 futures contract) against a ten-point stop. It wins 90% of the time. The expectancy arithmetic says it still loses money:
expectancy = (0.90 × 1) - (0.10 × 10)
expectancy = 0.90 - 1.00 = -0.10 points per trade
Nine trades out of ten are winners, and the account still bleeds a tenth of a point per trade before commission and slippage. The 90% figure hints at none of this. A win rate is a count, so it never sees the size of any trade.
Two mechanisms make a high win rate dangerous, not merely uninformative.
A smooth equity curve invites bigger positions. The equity curve is the running total of the account's profit and loss. At a 90% win rate the usual experience is a long run of small gains. The curve climbs in a near-straight line, and the largest drop so far stays small. Every risk statistic measured on that stretch looks excellent. Traders describe raising their position size at exactly that point. That is a practitioner account, not a measured finding, and why high win rate strategies blow up looks at it closely.
The large losses of a high-win-rate strategy have not stopped existing. They have not arrived yet. When they do, they hit a bigger position than the backtest assumed.
A win rate cannot show how big the losses are. Win rate is one number that counts how often trades won. It says nothing about the size of the rare worst losses, how much they vary, or whether they come in clusters. Two strategies can both report 90% and carry completely different risk. Whether the account survives is decided by the 10% of trades that lost.
The part that is usually left out
A rising win rate is only good news if the average win and the average loss stayed where they were. They rarely do. The changes that raise a win rate usually shrink wins and grow losses. Whenever a win rate improves, check the average win and the average loss at the same time. If expectancy fell while the win rate rose, the strategy got worse.
What win rate do I need for my risk-reward ratio?
Reward-to-risk is the average win divided by the average loss. Every reward-to-risk ratio has a breakeven win rate. That is the win rate at which the strategy makes exactly nothing before costs. It comes from setting expectancy to zero, so it is exact arithmetic, not a guideline.
Count the average loss as 1 and the average win as a multiple of it. Breakeven then needs the win rate W to satisfy W × reward-to-risk = (1 - W) × 1, which rearranges to:
(This is a ratio. The "R" in "expectancy in R", explained below, is something else: one unit of the money you risk on a trade.)
For example, suppose a strategy's winners average $200 and its losers average $100. Its reward-to-risk is 2, so it breaks even at 1 ÷ 3, a 33.3% win rate. Below that win rate it loses money before costs, and above it the strategy makes money. No amount of skill moves that floor.
| Reward-to-risk | Example: average win / average loss | Breakeven win rate |
|---|---|---|
| 1:1 | $100 / $100 | 50.0% |
| 1.5:1 | $150 / $100 | 40.0% |
| 2:1 | $200 / $100 | 33.3% |
| 3:1 | $300 / $100 | 25.0% |
| 5:1 | $500 / $100 | 16.7% |
| 10:1 | $1,000 / $100 | 9.1% |
| Arithmetic identity, exact | 1 ÷ (1 + reward-to-risk) |
The breakeven win rate is a floor, not a target. A strategy has to clear it by a margin that survives costs and bad luck. Going deeper covers which win rates traders report at 2:1 and 1.5:1, and which setting comes out ahead.
Check this on your own results
Costs raise the floor. Commission and spread come out of every winner and add to every loser. That shrinks the average win and grows the average loss at the same time, which lowers reward-to-risk. Recompute the breakeven win rate from average win and average loss net of costs, not from the gross target and stop distances.
Take a strategy with a one-point target and a ten-point stop. A cost of a quarter point per trade moves its effective reward-to-risk from 1:10 to 0.75:10.25. That raises the breakeven win rate from 90.9% to about 93.2%.
Why is win rate useless on its own?
Win rate says nothing about size. It counts how often a trade finished in profit and throws away every fact about size. It ignores how much each winner made and how much each loser lost. It cannot tell whether the losses were all alike or one was fifty times the others. Two strategies that work in completely different ways can report the same win rate.
The cleanest demonstration is a pair of strategies with the same expectancy and opposite win rates.
| Strategy | Win rate | Average win | Average loss | Expectancy |
|---|---|---|---|---|
| A — high win rate, low reward | 90% | +1 unit | -7 units | (0.9 × 1) - (0.1 × 7) = +0.20 |
| B — low win rate, high reward | 30% | +3 units | -1 unit | (0.3 × 3) - (0.7 × 1) = +0.20 |
| Both, after 100 trades | +20 units |
Strategy A wins nine times out of ten. Strategy B loses seven times out of ten. They have the same expectancy and finish the hundred trades at exactly the same equity. Yet their win rates differ by 60 percentage points. Judged on win rate alone they look like opposites, even though they earn the same on average.
They are not equivalent in path, which is the second thing win rate hides. A drawdown is a fall in the account from its peak to a later low.
The high-win-rate curve is the one that looks better for most of its length and is worse to own. Strategy A spends most of the sample making new highs, then surrenders more than its entire hundred-trade profit in three trades. Strategy B never looks impressive and never loses more than 10 units from a peak. Win rate ranks these two backwards.
What is expectancy and how do I use it?
Expectancy is the average profit or loss per trade, counting both how often trades win and how large the wins and losses are. It is the simplest number that tells you whether the average trade made or lost money.
where loss rate = 1 - win rate, and average loss is entered as a positive number
expectancy in R = expectancy per trade / average risk per trade
The first form is in whatever units the trades were measured in: currency, points or ticks. It answers "what does one more trade add, on average". The second form divides by the money at risk on a typical trade, producing a figure in R. Here R means one unit of the money you risk on a trade. It is not the reward-to-risk ratio used for the breakeven win rate. A strategy that makes an average of $30 per trade while risking $150 per trade has expectancy of $30 ÷ $150 = +0.2R.
An R-multiple is a trade's result expressed as a multiple of what was risked on it. Risking 200 units and making 400 is +2R. Being stopped out at the planned risk is -1R. The R framing and the practice of stating expectancy in R come from Van K. Tharp's Trade Your Way to Financial Freedom (2nd edition, McGraw-Hill, 2006).
Expectancy per unit of risk is the figure to compare across strategies. It removes two things that vary arbitrarily between them: position size and tick value, the money a one-tick price move is worth. A per-trade expectancy of 40 currency units means nothing without the position size that produced it. It also cannot be compared with a strategy trading a different instrument.
Expectancy of +0.25R can be compared with any other expectancy in R. Both are measured against the same thing: the money the trader chose to put at risk.
| Strategy | Win rate | Expectancy per trade | Risk per trade | Expectancy in R |
|---|---|---|---|---|
| A — wins 1, loses 7 | 90% | +0.20 units | 7 units | +0.029R |
| B — wins 3, loses 1 | 30% | +0.20 units | 1 unit | +0.20R |
| Same expectancy per trade, seven times apart per unit of risk | 0.20 / 0.029 ≈ 7 |
This is where the two strategies stop being equivalent. Per trade they are identical at +0.20 units. Per unit of risk, strategy B earns +0.20R while strategy A earns +0.029R. Strategy A has to risk seven units to earn the same 0.20. Sized to the same risk budget, B earns roughly seven times what A does. The per-trade figure hid that completely, and the win rate pointed the other way.
Using this with Tradelyze
Tradelyze shows a Win Rate tile on the Best Metrics card. The Top Trials table has Win Rate % and PF (profit factor) columns. Tradelyze does not calculate expectancy, but you can work it out from a trial's own trade list:
- In the Top Trials table, click the trial's row to open it, then click Download Trades. The button appears only when Tradelyze kept that trial's trade list, so some older results do not have it.
- The file has one row per trade, and each trade's profit or loss is in the P&L column. The PnL column repeats the same values; cumulative_pnl is a running total, so do not average it. Count the trades with a P&L above zero (wins) and below zero (losses). Divide each count by the total number of trades to get the win rate and the loss rate.
- Average the positive P&L values to get the average win. Average the negative ones and drop the minus sign to get the average loss.
- Expectancy = win rate × average win − loss rate × average loss. The result is in the same units as the P&L column. It always equals the plain average of the whole P&L column, which is a quick check on your arithmetic.
A worked example with made-up numbers: 40 trades, of which 22 won an average of 150 and 18 lost an average of 100. That gives a win rate of 55% and a loss rate of 45%. Expectancy is (0.55 × 150) − (0.45 × 100) = 82.5 − 45 = +37.5 per trade. As a check, (22 × 150 − 18 × 100) / 40 = 1,500 / 40 = 37.5.
If the button shows two counts, as in Download Trades (N of M), the file holds only the first N of the trial's M trades. Your figure then covers only those trades. The file has no column for the amount you planned to risk, so turning expectancy into R needs your own stop distance.
What expectancy still does not tell you
Expectancy is an average, and an average has no path. It says nothing about the order the trades arrive in or how deep the account falls before it recovers. It also cannot show whether losses cluster. A strategy with positive expectancy can still ruin an account sized as though a deep drawdown could not happen. The chance of hitting a loss the account cannot come back from is called risk of ruin. A Monte Carlo simulation reorders or resamples the same trades many times. It is the standard way to see the range of paths one expectancy figure allows.
Why do high win rate strategies blow up?
A strategy blows up when it loses so much that the account, or the challenge, is finished. When a high-win-rate strategy does blow up, the usual account has two stages, and neither shows in the win rate. This describes how such failures happen, not how often high-win-rate strategies fail.
Stage one: the losses are packed into a few large trades
A strategy with a 90% win rate and a reward-to-risk of 1:7 does not distribute its risk evenly across trades. Ninety percent of its trades are near-costless and ten percent of them carry the entire loss budget.
Take strategy A from the constructed equity-curve figure on this page. It wins 1 unit on 90 of its 100 trades and loses 7 units on the other 10. Three of those ten losses landed back to back, at trades 70, 71 and 72. That one cluster produced a 21-unit drawdown, more than the strategy's entire hundred-trade profit of 20 units.
Nothing unusual occurred. Ten losses spread at random over a hundred trades will sometimes land next to each other. How much strategy A loses in a bad stretch depends mostly on whether its rare large losses cluster. The win rate contains no information about that.
Strategy B in the same figure wins 3 units on 30 of its 100 trades and loses 1 unit on the other 70. Its losses are spread across seventy small events. Matching strategy A's 21-unit drawdown would take at least 21 of those 1-unit losses with few wins in between. That is possible but unlikely. Strategy B's worst drawdown in the figure is 10 units.
Stage two: the smooth curve pulls in leverage
Stage two is a practitioner account, not a measured finding, and no study is cited for it.
On that account, a high-win-rate strategy produces long stretches with no real setback. Leverage here means trading a bigger position for the same account. Every risk statistic measured on a quiet stretch looks flattering: the largest drawdown so far, the longest losing streak, the worst month. The losses that would fill those statistics have not happened yet. After a run of clean months, traders describe increasing their position size.
The size increase is based on the results seen so far. The loss that eventually arrives comes from results not yet seen. Position size grows during the quiet stretch, and the cluster of large losses then lands on the bigger position. A drawdown the account could survive at the original size may not be survivable at three times that size. No single decision along the way looked reckless.
Stage one's arithmetic holds whether or not stage two describes a given trader.
On a prop firm challenge, a cluster of large losses is even less forgiving than in a personal account. A prop firm challenge is a paid evaluation with a profit target and loss limits. It usually ends when the account touches its daily loss limit or its maximum drawdown limit. A high-win-rate strategy then gets no chance to earn back a hole like strategy A's 21 units. The win rate says nothing about whether three large losses in a row fit inside those limits. Prop firm rules and backtest metrics shows which backtest number to compare with each rule.
Check this on your own results
For any strategy with a high win rate, compute two things the win rate hides. First, divide the average loss by the average win. If that ratio is larger than the win rate divided by the loss rate, the strategy loses money however high the win rate is. At a 90% win rate that limit is 0.9 / 0.1 = 9. Losses averaging more than nine times the average win mean a losing strategy.
Second, divide the largest drawdown by the average win. A strategy that wins 1 unit and loses 7 needs 21 winning trades to recover from three losses in a row. If that number is large, the strategy's real limit is its rare large losses, not its win rate.
How do win rate, profit factor and the Sharpe ratio differ?
Win rate, expectancy, profit factor and the Sharpe ratio answer four different questions about the same set of trades. None of them contains everything the others do. What separates them is exactly the information each one throws away.
| Metric | What it measures | What it ignores | Comparable across account sizes? |
|---|---|---|---|
| Win rate | How often a trade closed in profit | The size of every win and every loss, the order of trades, and how big the rare worst losses are | Yes, but it says nothing about how large wins and losses are, which is why it cannot be read alone |
| Expectancy | Average profit or loss per trade, combining frequency and size | The order of trades and how widely results swing, so it cannot show how far the account falls before it recovers | Only in the R form; the per-trade form is in currency or points |
| Profit factor | Gross profit divided by gross loss over the whole sample | Trade count, the order of trades, and whether one outlier supplied the gross profit | Yes — it is a ratio of two money totals |
| Sharpe ratio | Return per unit of volatility, meaning how much returns swing around | The direction of volatility — large winners are penalized the same as large losers | Yes, but the value depends on how often returns are measured: per trade, per bar or per day. It also depends on whether the figure is annualized, meaning converted to a yearly figure. |
The practical order for judging a strategy is expectancy in R first. It is the only one of the four that combines how often trades win, how big they are, and the money at risk. Profit factor is a useful cross-check because it has no units, so it reads the same on any instrument without knowing the risk per trade. Like expectancy, profit factor can be inflated by one unusually large win. Check how much of the gross profit came from the single biggest trade before trusting either figure.
The Sharpe ratio adds how widely results swing, which none of the others show. The cost is that it treats big winning swings and big losing swings the same way. Win rate is the one to read last. It describes what holding the strategy feels like, not whether the strategy works.
How many trades before a win rate means anything?
A win rate measured on a few trades is mostly noise. The number of trades needed is larger than most backtests contain. The useful comparison is a fair coin. If a coin would often produce the same win rate, the win rate is not evidence of anything.
| Trades | Wins needed for 60% | Chance a fair coin does it | Reading | Source |
|---|---|---|---|---|
| 20 | 12 | 25.2% | Indistinguishable from a coin flip. One backtest in four hits it by chance. | Chance: exact binomial arithmetic. Reading: this page's judgment; no primary source. |
| 50 | 30 | 10.1% | Still common enough that a handful of tested variants will produce one. | Chance: exact binomial arithmetic. Reading: this page's judgment; no primary source. |
| 100 | 60 | 2.8% | Unlikely for a single pre-specified test, but not for a search over many variants. | Chance: exact binomial arithmetic. Reading: this page's judgment; no primary source. |
| 200 | 120 | 0.28% | Rare by chance for one test. Trial count still has to be accounted for. | Chance: exact binomial arithmetic. Reading: this page's judgment; no primary source. |
| Method | Exact binomial tail probabilities at p = 0.5, computed arithmetically. Not an estimate and not a convention. |
A 60% win rate over 20 trades is indistinguishable from a coin flip. A fair coin produces 12 or more heads in 20 flips about 25.2% of the time. That win rate describes a sample, not a property of the strategy. At 100 trades, a fair coin reaches 60% only about 2.8% of the time. That is meaningful for a single test decided in advance. It means much less if the result was the best of fifty tested variants, meaning versions of the strategy's settings.
Testing many variants is the bigger of the two problems, and more trades do not solve it. The more variants you try, the more likely some variant clears a threshold by luck. So record how many settings were tried as well as how many trades were taken. Both requirements are covered in how many trades a backtest needs.
Where this appears in Tradelyze
In a Tradelyze report, this is the Win Rate tile in Best Metrics and the Win Rate % column in Top Trials. Both come from the same price data the optimizer searched, so they are in-sample figures, the most optimistic version of the number. Tradelyze does not show expectancy. The section What is expectancy and how do I use it? shows how to calculate it from a trial's downloaded trades.
To judge the whole report, not one tile, use the pre-trade checklist.
Tradelyze re-runs an uploaded TradingView Pine Script strategy from your exported trade list and price data. It then runs parameter optimization, walk-forward analysis, a four-check robustness score and prop firm rule checks. It does not place trades, give financial advice or guarantee a challenge pass, and it is in beta.
Create an account. Already a user? Open your strategies.
Going deeper
The sections below go deeper: the win rates one trader reports at 2:1 and 1.5:1 reward-to-risk, and which of the two settings comes out ahead in expectancy. You can skip them and still read your own report.
Which win rates do traders report at 2:1 and 1.5:1 reward-to-risk?
The breakeven win rate is exact arithmetic. The win rate traders actually achieve at a given reward-to-risk is a softer, reported number. The figures below come from one trader's account of long-run experience in EliteTrader thread 337793. They are a practitioner report, not a study, so read them as a rough expectation rather than a measurement.
| Reward-to-risk | Breakeven win rate | Win rate reported in the same pairing | Source |
|---|---|---|---|
| 1:1 | 50.0% | — | No primary source. At 1:1 every cost comes directly out of a razor-thin margin. |
| 1.5:1 | 40.0% | about 60% | EliteTrader thread 337793: traders reporting around a 60% win rate describe reward-to-risk falling toward 1.5. The report runs in that direction, not the other one. |
| 2:1 | 33.3% | 45–50% | EliteTrader thread 337793: roughly 45% to 50% wins reported alongside about 2:1. |
| 3:1 | 25.0% | — | No primary source for a typical achieved win rate at 3:1. |
| Reported column | A pairing, not a prediction. Read it as "traders reporting these win rates describe running about this reward-to-risk", never as "choose this ratio and expect this win rate". The rows are indexed by reward-to-risk because the breakeven column is; the reported figures were volunteered the other way around. |
The pattern in the reported column is a trade-off, not a menu. The account pairs roughly 45% to 50% wins with about 2:1 reward-to-risk. It pairs around 60% wins with reward-to-risk sagging toward 1.5. Because it is one trader's report, the pattern is worth more than the exact figures.
Win rate and reward-to-risk move against each other because they are two views of the same exit decision. Holding a trade for a larger target means more trades turn around and hit the stop. Taking profit earlier turns would-be losers into small winners.
Compare these pairings in expectancy, where R is one unit of the money you risk on a trade, because points above the breakeven floor are still a win-rate measure that ignores trade size:
2:1 at a 50% win rate: (0.50 × 2) - (0.50 × 1) = +0.500R
1.5:1 at a 60% win rate: (0.60 × 1.5) - (0.40 × 1) = +0.500R
At the top of the reported 2:1 range the two settings are exactly equal. At the bottom of it, 1.5:1 is ahead by 43% (+0.500R against +0.350R). On this evidence the ratio is a style choice within a band rather than a free upgrade in either direction. Nothing here supports treating 2:1 as the better setting, and the ranking is only as good as the reported win rates behind it.
Stage 2 · step 5 of 18. Next in the learning path: Profit factor
Frequently asked questions about win rate and expectancy
What is a good win rate in trading?
There is no good win rate independent of the size of the wins and the losses. A 30% win rate is excellent at three-to-one reward-to-risk and fatal at one-to-one. The only figure that combines both is expectancy: (win rate × average win) - (loss rate × average loss). Judge a strategy on expectancy per unit of risk, never on the win rate alone.
Is a 90% win rate good?
Not on its own, and experienced traders read it as a warning rather than an achievement. A widely repeated trading-forum example is a strategy that takes one point of profit on the ES futures contract against a ten-point stop. It wins 90% of the time and still loses money, because 0.9 × 1 - 0.1 × 10 = -0.1 points per trade before costs.
What is the trading expectancy formula?
Expectancy = (win rate × average win) - (loss rate × average loss), where the loss rate is 1 minus the win rate and the average loss is entered as a positive number. The result is the average profit or loss per trade in currency or points. Dividing it by the average risk per trade converts it to R, which is comparable across instruments and account sizes.
Does Tradelyze calculate expectancy?
No. Tradelyze shows win rate as the Win Rate tile on the Best Metrics card and the Win Rate % column in Top Trials, and profit factor in the PF column, but it does not show expectancy. To calculate it, download a trial's trades from Top Trials and work out (win rate × average win) - (loss rate × average loss) from the P&L column, or simply average that column.
What win rate do I need for a 2:1 risk-reward ratio?
The arithmetic breakeven win rate at two-to-one reward-to-risk is 33.3%, from the identity breakeven win rate = 1 ÷ (1 + reward-to-risk). That is a floor before costs, not a target. A trader posting in EliteTrader thread 337793 reports win rates of roughly 45% to 50% alongside two-to-one, which is a pairing rather than a prediction. Commissions and slippage raise the floor.
Can a strategy with a 30% win rate be profitable?
Yes, and the arithmetic is not marginal. At three-to-one reward-to-risk a 30% win rate gives expectancy of 0.3 × 3 - 0.7 × 1 = +0.2 units per trade. That is exactly the same expectancy as a 90% win rate that wins one unit and loses seven. The two strategies finish in the same place through completely different equity curves.
Why do high win rate strategies blow up?
Because the win rate hides where the risk sits. A high-win-rate, low-reward strategy concentrates its risk into a small number of large losses, so the equity curve looks smooth for long stretches. On the usual practitioner account, smoothness then invites leverage, applied before the rare large losses have shown up, and the first cluster of losses lands against a larger position than the backtest assumed.
What is an R-multiple?
An R-multiple expresses a trade's result as a multiple of the money risked on that trade. A trade risking 200 units that returns 400 is +2R, and one stopped out at the planned risk is -1R. Expectancy in R is the average R-multiple per trade. Because it is denominated in risk rather than currency, it compares across instruments, position sizes and accounts.
Is win rate or risk-reward ratio more important?
Neither, because neither can be read alone. Win rate says how often a trade wins, never by how much. Reward-to-risk says how much a win pays compared with what a loss costs, never how often. Only the combination carries information, which is what expectancy computes. A strategy is defined by the pair, so any change that raises one while lowering the other is a trade-off, not an improvement.
How many trades before a win rate means anything?
More than most backtests contain. A fair coin flipped 20 times produces 12 or more heads about 25.2% of the time, so a 60% win rate over 20 trades is not distinguishable from luck. At 100 trades the same 60% figure falls to roughly a 2.8% chance under a fair coin. Both figures are exact binomial calculations rather than estimates.
What is the difference between win rate and profit factor?
Win rate counts trades and profit factor weighs money. Profit factor is gross profit divided by gross loss, so it accounts for how large the winners and losers were, which win rate discards entirely. A strategy can have a 90% win rate and a profit factor below 1.0. Profit factor is closer to expectancy than win rate is, but it still ignores the order the trades arrived in.
Does a high win rate mean lower risk?
No, and it often means the opposite. The usual way to raise a win rate is to widen the stop or shrink the target, and both concentrate loss into fewer, larger events. Risk lives in how big the rare worst losses are, not in how often they happen. A strategy that loses rarely but loses ten times its average win is a high-risk strategy with a comfortable-looking statistic.
How do trading costs change the breakeven win rate?
Costs raise it, and they raise it most for strategies with small targets. Commission and spread are subtracted from every winner and added to every loser, so they shrink the average win and grow the average loss at the same time. A strategy targeting one point against a ten-point stop has almost no room to absorb them, which is why high-win-rate scalping degrades fastest between backtest and live trading.
Sources
- EliteTrader thread 231131, What is a win rate? (retrieved 28 July 2026) — the source of the widely repeated trading-forum example in the section on whether a high win rate is good: the one-point target against a ten-point stop, posted as a rhetorical question: "Win rate means nothing without looking at the risk to reward ratio with it. What good is a 90% win rate for 1 point es gains if your stop is 10 points." The poster does not state that the strategy loses money and does not do the arithmetic; the expectancy calculation on this page is this page's own, and it is shown in full so it can be checked.
- EliteTrader thread 337793, 50-60% win rate, 1:2 R:R (retrieved 28 July 2026) — both reported pairings, in one passage: "If your avg win is 2x your avg loss, over the long-term, in a winning system, you can expect your avg winning pct to gravitate more towards 45-50%. When you get at the avg 60% winners level, long-term, expect your win to loss ratio to drop more towards 1.5." One trader's account of long-run experience in a public forum, not a study, and it should be read as a rough expectation. Used under Going deeper, in which win rates traders report.
- Van K. Tharp, Trade Your Way to Financial Freedom, 2nd edition, McGraw-Hill, 2006 — expectancy stated in R-multiples, where R is the money risked on a trade.
- Breakeven win rate = 1 ÷ (1 + reward-to-risk), where reward-to-risk = average win ÷ average loss: arithmetic identity obtained by setting expectancy to zero. No citation required and none exists. The cost example (1:10 becoming 0.75:10.25, breakeven rising from 90.9% to about 93.2%), the dollar examples in the breakeven table, and the expectancy comparison of the reported 2:1 and 1.5:1 pairings (+0.350R, +0.500R, +0.500R) are this page's own arithmetic, applied to win rates that are practitioner reports rather than measurements.
- Coin-flip probabilities in the sample-size table: exact binomial tail probabilities at p = 0.5, computed directly. 12 or more heads in 20 flips is 25.2%; 30 or more in 50 is 10.1%; 60 or more in 100 is 2.8%; 120 or more in 200 is 0.28%.
- Equity curves in the figure: constructed illustrations, not measured results. Strategy A takes 90 wins of 1 unit and 10 losses of 7 units; strategy B takes 30 wins of 3 units and 70 losses of 1 unit. Both total +20 units over 100 trades.
- David H. Bailey, Jonathan M. Borwein, Marcos López de Prado and Qiji Jim Zhu, Pseudo-Mathematics and Financial Charlatanism: The Effects of Backtest Overfitting on Out-of-Sample Performance, Notices of the American Mathematical Society 61(5), May 2014, pp. 458–471. DOI 10.1090/noti1105; ams.org PDF (retrieved 28 July 2026) — on why a threshold cleared by one variant out of many is not evidence.
- The two-stage blow-up mechanism: stage one, the concentration of loss into a small number of large events, is arithmetic and follows from the reward-to-risk ratio. Stage two, position size rising during a quiet stretch, has no primary source — no behavioral-finance study is cited for it here. It is a practitioner account of how traders describe the failure, and it is labeled as one on the page.
- The 0.3 / 0.5 / 0.7 style "good win rate" bands that circulate on trading blogs have no primary source and are not used anywhere on this page.
- Tradelyze implementation, reviewed 15 September 2026 — the Win Rate tile on the Best Metrics card, the Win Rate % and PF columns in the Top Trials table, the absence of any expectancy figure in the results, and the per-trade P&L column (repeated as PnL, next to a running cumulative_pnl total) in the file that Download Trades saves. The button sits inside an expanded Top Trials row and reads Download Trades (N of M) when the file holds only the first N of a trial's M trades.