Win rate alone tells you nothing about profitability, and neither does risk-reward. What decides whether you make money is expectancy — the combination of how often you win, how much you win, and how much you lose, minus costs. A trader winning 80% of the time can bleed out, and a trader winning 25% of the time can compound steadily; the formula below explains both. This post covers the expectancy formula, breakeven win rates by risk-reward ratio, how costs move that line, the two traps on either side of the math, and why 20 trades prove nothing.
The Expectancy Formula, in Plain English
Expectancy is your average profit or loss per trade over many trades:
E = (Win% × Average Win) − (Loss% × Average Loss)
If you win 40% of the time, your average winner is $200, and your average loser is $100:
E = (0.40 × $200) − (0.60 × $100) = $80 − $60 = +$20 per trade
Every trade you take is "worth" $20, even the losers. That's the mental shift expectancy forces: individual trades are noise; the distribution is the business.
It's cleaner to work in R multiples, where 1R is the amount you risk per trade. If your winners average +2R and your losers average −1R at a 40% win rate:
E = (0.40 × 2R) − (0.60 × 1R) = +0.2R per trade
Risk $100 per trade across 100 trades and this system earns about $2,000 — before costs, which we'll get to.
Two things follow immediately: a high win rate cannot save a bad risk-reward ratio (the two multiply — neither stands alone), and average win and loss are what count — not your planned stop, but what actually happens, including the day you froze and let a loser run.
The Breakeven Win Rate for Every Risk-Reward Ratio
Set E = 0 and solve to get the breakeven win rate for any risk-reward ratio:
Breakeven win rate = 1 ÷ (1 + R) — where R is your reward-to-risk multiple.
| Risk-reward (R:R) | Breakeven win rate (before costs) |
|---|---|
| 0.5R (risk $100 to make $50) | 66.7% |
| 1R | 50% |
| 1.5R | 40% |
| 2R | 33.3% |
| 2.5R | 28.6% |
| 3R | 25% |
| 4R | 20% |
Two conclusions fall out of that table. First, systems that risk more than they make need remarkably high win rates just to break even — at 0.5R you must win two out of three trades before earning a cent. Second, a 3R system is profitable at a win rate that feels like constant failure — you can be wrong 70% of the time and still make money. You can test any combination of win rate, ratio, and trade count in seconds with the free win-rate calculator instead of doing the algebra by hand.
But notice the caveat in the table header: before costs. That caveat is where most live accounts diverge from the spreadsheet.
Costs Move the Line: A Worked Example on NQ and MES
Every futures round trip pays commission (broker + exchange fees, both sides) and usually eats slippage (fills one tick worse than intended, especially on market orders and stops). Both are invisible in most backtests and relentless live.
Work it through on NQ, where one tick is 0.25 points = $5.00. Say you trade a 1:1 setup with a 20-tick stop and a 20-tick target — $100 risk, $100 reward, 50% breakeven on paper. At the time of writing, a typical retail all-in round turn on NQ runs about $4.20 (check your broker's schedule — fees vary). Add one tick of slippage per side — realistic for market entries and stop exits:
| NQ (tick = $5.00) | MES (tick = $1.25) | |
|---|---|---|
| Stop / target | 20 ticks / 20 ticks | 20 ticks / 20 ticks |
| Gross risk / reward | $100 / $100 | $25 / $25 |
| Commission (round turn, approx.) | $4.20 ≈ 0.8 ticks | $1.30 ≈ 1.0 tick |
| Slippage (1 tick per side) | $10.00 | $2.50 |
| Total cost per trade | ~$14.20 ≈ 2.8 ticks | ~$3.80 ≈ 3.0 ticks |
| Net average win / loss | $85.80 / $114.20 | $21.20 / $28.80 |
| Breakeven win rate after costs | 57.1% | 57.6% |
With costs, breakeven = net loss ÷ (net win + net loss). Your "50/50 coin flip" now requires 57% wins just to break even — costs moved the line by seven full points. And the tighter your stops, the worse it gets: on a 10-tick scalp the same $14.20 is 28% of your risk, and the breakeven win rate on a 1:1 setup climbs past 63%.
Note the micro column carefully. MES cuts your dollar risk to a quarter of ES-class sizing, which is exactly what it's for — but commission is proportionally heavier per tick (about 1.0 tick on MES vs 0.8 on NQ), so micros are a position-sizing tool, not a cost hack. The same dynamic applies to MNQ vs NQ, covered in detail in NQ vs MNQ: which contract to trade. To translate any stop distance into dollars across contracts, the futures calculator does the tick math for you.
The High Win Rate Trap: Winning 80% and Still Losing
The most seductive equity curve is the one that goes up almost every day — then gives it all back in an afternoon. It's the classic failure mode of undisciplined scalping: take profits fast, let losers "come back."
Concrete version, on NQ. A trader wins 80% of trades for an average of +5 ticks ($25). The 20% of losers average −30 ticks ($150) because there's no hard stop. Per 100 trades:
- Wins: 80 × $25 = +$2,000
- Losses: 20 × $150 = −$3,000
- Commissions: 100 × $4.20 = −$420
- Net: −$1,420, an expectancy of −$14.20 per trade — at an 80% win rate.
In R terms: E = (0.80 × 1) − (0.20 × 6) = −0.4R whenever your average loser is six times your average winner. Eight green trades out of ten feels like skill; the formula says slow-motion blowup. One −6R day erases fifteen winners.
This trap is lethal in prop-firm evaluations, where one oversized loser breaches a daily loss limit or trailing drawdown and ends the account — a recurring pattern in why traders fail prop firm challenges. Scalping itself isn't the problem (scalping futures strategies covers viable approaches) — but a scalp only works when losers are capped as tightly as winners are taken.
The Opposite Trap: 3R Systems and 8-Loss Streaks
Flip the table and there's a mirror-image trap. A 3R system breaks even at 25% and makes good money at 35% — but a 35% win rate means you lose 65% of your trades — and losses don't arrive politely spaced. They cluster.
The streak math is simple. The probability that any given trade starts a run of k straight losses is:
P = (1 − W)^k
At a 35% win rate, the chance of 8 straight losses starting from any particular trade is 0.65⁸ ≈ 3.2%. Sounds rare — until you remember that a 100-trade sample gives you nearly a hundred starting points. Run the full runs math (or a quick simulation) and a streak of 8 or more losses shows up in roughly two-thirds of all 100-trade samples, and in about 97% of 300-trade samples. It is not a tail risk. It is a scheduled feature.
Typical longest losing streak you should expect somewhere in a 100-trade sample (simulated, rounded):
| Win rate | Expected longest losing streak per 100 trades |
|---|---|
| 70% | 3–4 |
| 60% | 4–5 |
| 50% | 6 |
| 40% | 8 |
| 35% | 9 |
| 25% | 13 |
This is why traders abandon perfectly profitable low-win-rate systems: they hit the statistically guaranteed streak, conclude the edge is gone, and quit — usually right before the math reverts. A classic example is the opening range breakout — small losses, occasional large wins; the opening range breakout NQ strategy post shows what those loss clusters look like in practice. The streak table is also a sizing constraint: nine straight losses at 1% risk is a 9% drawdown, which is survivable — at 3% risk it's 27%, which usually isn't, and under a prop firm's trailing drawdown it's account-ending well before that (see trailing drawdown explained: EOD vs intraday).
Why 20 Trades Prove Nothing
Here's the uncomfortable part: win rate measured over a small sample is mostly noise. The standard error of a measured win rate is √(W(1−W)/N), which means the 95% confidence range around your observed win rate is roughly:
| Trades | Margin of error on win rate |
|---|---|
| 20 | ±22 points |
| 100 | ±10 points |
| 400 | ±5 points |
After 20 trades, an observed 50% win rate is statistically consistent with anything from roughly 28% to 72% — the difference between a losing system and a great one. This is the single biggest reason to backtest before trading an idea live: you need trade count, and the calendar only gives you a handful of setups per week. A disciplined process for building that sample is laid out in how to backtest a trading strategy, and if you're choosing tooling, the futures trading simulator guide covers what actually matters in a replay platform.
As a working rule: 100 trades minimum per setup before trusting the win rate, more if your win rate is low — streak-heavy distributions need bigger samples for the average to stabilize.
How to Measure Your Real Numbers
Expectancy only helps if the inputs are your numbers, not aspirational ones. That means logging, per setup: win rate, average win, average loss, and costs — then computing E per trade and per R.
The fastest way to build that dataset is market replay: stream historical sessions bar by bar with the right edge hidden, trade them as if live, and let the platform do the bookkeeping. TestMax computes win rate, expectancy, and per-setup breakdowns automatically on every replay session, and because replay runs at 1x–50x speed, a month of trading days fits into an evening or two — which is how you get to a 100-trade sample in weeks instead of a year. If you're comparing options first, start with the roundup of the best free backtesting platforms. And if your goal is a funded account, measure your numbers under the same constraints you'll trade under — the prop firm simulator guide covers practicing with trailing drawdown and daily loss limits enforced.
Whatever tool you use, separate setups. A blended "my win rate is 52%" across three trade types hides the one setup with negative expectancy taxing the other two.
The Honest Part: Backtest Expectancy Decays Live
Assume your live results will be worse than your backtest, because they almost always are. The decay has boring, structural causes:
- Fills: replay and backtests fill you at your price; live markets sometimes don't, and the misses cluster on your best trades.
- Selection bias: you tested the idea because it looked good on the data you tested it on. Fresh data is rarely as kind.
- Regime change: an edge measured in one volatility regime meets a different one.
- Execution gap: the backtest never hesitated on an entry, moved a stop, or revenge-traded. You will, occasionally, even after training.
The practical response isn't despair — it's a margin of safety. Don't trade a system whose backtest expectancy is barely positive after realistic costs; a +0.05R edge on paper is a losing system live. Demand an edge big enough that it survives a haircut, model costs pessimistically (assume the extra tick of slippage), and track live expectancy against backtest expectancy from day one, so you see the gap instead of guessing. Replay practice narrows the execution part of that gap — it can't narrow the market's part.
Put Your Own Numbers on the Table
Everything above reduces to a short checklist you can run this week:
- Pick one setup and define entry, stop, and target mechanically enough to count.
- Trade 100 samples of it in replay across a few months of sessions.
- Record the four inputs: win rate, average win, average loss, cost per round trip in ticks.
- Compute E — per trade and per R. If it's negative or marginal after costs, the setup needs work, not more size.
- Check the streak table and size so the expected worst streak is survivable.
You can run the whole loop on TestMax's futures backtesting replay — the free plan includes three months of futures data with no credit card — enough to put a real expectancy number on a setup before any live dollar is at risk. The formula doesn't care how the trades feel. That's exactly why it works.