What is expectancy in trading?

Expectancy is the average amount you earn per trade, expressed in R (multiples of the amount you risk). The formula is expectancy = (win rate × average win) − (loss rate × average loss). A positive expectancy means you make money over many trades; a negative one means you pay to play, however often you win. Expectancy — not win rate — is the number that decides whether a strategy is profitable.

Can a high win rate lose money?

Yes. A 90% win rate that takes 0.15R wins against full 1R losses has an expectancy of about −0.015R per trade after costs — a net loser, because one loss erases nearly seven wins. In a simulation of six profiles, the three highest win rates (90%, 75%, 50%) were all net losers and the three lowest (52%, 40%, 30%) were all winners. Win rate on its own tells you nothing about profitability; you have to know the size of the wins it buys.

What win rate do I need to be profitable?

It depends entirely on your reward-to-risk ratio. The break-even win rate is 1 ÷ (1 + reward:risk). At 1:1 you need 50%; at 2:1 you need 33.3%; at 3:1 you need 25%; at 0.5:1 you need 66.7%. Any win rate above its break-even line is profitable before costs; any win rate below it loses. A win rate is only meaningful next to the reward-to-risk it was earned at.

Does a prop firm evaluation care about my win rate?

No. An evaluation grades whether your expectancy survives a drawdown floor, not how often you win. A high-win-rate style produces a smooth equity curve that hides a fat tail of clustered losses, and the drawdown limit is waiting there. In a 100-trade evaluation with a 5% floor, a 90%-win-rate scalper with negative expectancy failed about 25% of the time and ended −1.4% on average, while the reference trader (52% win rate, positive expectancy) failed about 13% of the time and ended +27%. Feeling right is not the same as being paid, or surviving.

PropSurvival · The number that lies · Seeded Monte Carlo

Win rate is the most
misleading number in trading.

It's the first thing traders quote and the last thing that decides whether they make money. You can win nine trades out of ten and still go broke; you can lose more often than you win and compound for years.

The finding

The trader who wins nine times in ten fails 24.8% of evaluations.

Binding constraint — average loss size, not win frequency

Scroll for the number

Run your expectancy free — (Win% × Avg Win) − (Loss% × Avg Loss) →

01 · The illusion

Does winning more often mean making money?

No. Read the win rates. The three highest here — 90%, 75%, 50% — all lose money. The three lowest — 52%, 40%, 30% — all win.

Six traders, sorted by profit. The win-rate column beside them is in no order at all: it climbs, drops, jumps. That's the whole lesson in one picture — win rate does not sort winners from losers. A trader who wins 9 times in 10 can bleed out if the wins are small and the losses are whole; a trader who's wrong 70% of the time can thrive if the few wins are large.

The teachable fact: the top-3 win rates all lose; the bottom-3 all win

Read down the win rates. Do they sort the winners?

They sort nothing. The 9-in-10 scalper — the highest win rate here, and the bar the whole argument turns on — settles at −0.015R a trade, because one full-R loss erases nearly seven of its 0.15R wins. The 30% trend follower is the third-best earner. Which side of the rule a bar falls on is set by the size of the wins, not by how often they arrive.

expectancy per trade (R) · losses fixed at −1R · net of 0.05R cost

How high must the win rate be, at each reward-to-risk?

The bar you must clear falls as the wins get bigger: 50% at 1:1, 33% at 2:1, 25% at 3:1. This curve is the only thing that makes a win rate readable — quoted on its own, without the ratio it was earned at, the number cannot be scored.

break-even win rate = 1 ÷ (1 + reward:risk) · above the line pays, below it bleeds

02 · The number

So what actually decides it?

Expectancy = (win rate × average win) − (loss rate × average loss). The average R you earn per trade. Positive, you compound. Negative, you pay to play.

A win rate is only meaningful next to the size of the wins it buys. Rearranged, the break-even win rate is 1 ÷ (1 + reward:risk): at 1:1 you need 50%, at 2:1 just 33%, at 3:1 only 25%. That's why a 30%-win-rate trend follower prints money at 3.5:1, and a 90%-win-rate scalper bleeds at 0.15:1 — one full-R loss erases nearly seven of those tiny wins. And note the coin-flipper on the ladder: a perfectly fair 1:1 game at 50% still loses, because the 0.05R cost per trade is a tax you pay whether or not you have an edge.

View the break-even win rates

Win rate needed to break even, by reward:risk
Reward : risk	Break-even win rate
0.5 : 1	66.7%
1 : 1	50.0%
1.5 : 1	40.0%
2 : 1	33.3%
3 : 1	25.0%
4 : 1	20.0%

03 · The two traders

Then why does everyone chase win rate?

Because being right feels like skill. The 9-in-10 scalper wins almost every day — and ends an evaluation down 1.4% on average, failing one time in four.

Put both traders through the same 100-trade evaluation. The scalper's outcomes are a narrow band around break-even — smooth, reassuring, and quietly negative: a near-perfect reinforcement schedule that says “right again” nine times in ten, and charges for the feeling. The reference trader hears “wrong” almost every other trade but carries a long right tail: a median that's positive and a +45.8% good run that the scalper's style can never reach. A high win rate buys a comfortable feeling. Expectancy buys the right tail.

Watch a good trader get ended by rules →

Where do 20,000 evaluations end, for each trader?

Both fields bottom out at the same place, near −5%, because both traders can meet the floor. Only the 52% trader's field reaches +45.8%. The two are the same measurement of the same quantity, so they are drawn the same way — what differs is the length of the upside, and that is bought by expectancy, not by being right more often.

evaluation outcome, 10th–90th percentile · 1% risk · 5% floor · 100 trades · 20,000 paths

Which of the two fails the evaluation more often?

The one that wins nine times in ten: 24.8% of its evaluations end at the floor, against 13.5%. Failure is set by the size of the losses relative to the floor and by how they cluster — the count of winning days does not enter the test.

share of 100-trade evaluations failed · same floor · 20,000 paths each

24.8%

Evaluations failed by the trader who wins nine in ten

04 · What the eval grades

The evaluation never asks your win rate.

It asks whether your expectancy survives a floor. The scalper wins more often and fails nearly twice as often — 24.8% versus 13.5%.

A high-win-rate curve looks smooth right up until the cluster of losses that the style always eventually meets — and the drawdown floor is waiting there. This is where the two halves of survival meet: expectancy gets you a positive average, and risk of ruin decides whether you're still trading when the average arrives. You need a positive edge and enough room to outlast its variance. Win rate speaks to neither.

The teachable fact: the higher win rate failed 1.8× more often

05 · The takeaway

Stop counting wins. Start weighing them.

Win rate is a vanity metric; expectancy is the P&L. And the evaluation only grades the second — after variance has had its say.

See your real expectancy — (Win% × Avg Win) − (Loss% × Avg Loss) — and your pass odds, free →

Enter your win rate, average win and average loss in R — or import a trade CSV — and get your expectancy, your risk of ruin and your pass probability under each firm's actual rules. Nothing you enter leaves your device.

Five facts worth keeping

1 · Win rate does not sort winners from losers.: The three highest win rates (90/75/50%) all lose; the three lowest (52/40/30%) all win.
2 · Expectancy is the only verdict.: (win rate × avg win) − (loss rate × avg loss) — the average R per trade. Positive or you pay to play.
3 · Break-even win rate = 1 ÷ (1 + reward:risk).: 50% at 1:1, 33% at 2:1, 25% at 3:1. A win rate means nothing alone.
4 · A high win rate hides a fat tail.: The 90% scalper ends −1.4% on average and fails 1 in 4 evaluations.
5 · The evaluation grades expectancy, not win rate.: Pair a positive edge with enough room to outlast its variance — that's survival.

See whether your edge survives the floor →The full study Prop trading glossary

PropSurvival is independent analytical software — not affiliated with any firm, and not investment advice. Expectancy and break-even figures are exact algebra; the evaluation outcomes are measured from a seeded Monte Carlo model (1% risk per trade, 5% static floor, 100 trades, 20,000 paths per profile, mulberry32 seed 12345, losses fixed at −1R, 0.05R cost). Your own numbers are the only ones that describe you.