Guide · Statistics
The short answer
Less than traders think. A 35% win rate at 1:3 beats a 65% win rate at 1:1, and both beat a high win rate with no stop.
Win rate is the most quoted statistic in trading and the least informative one in isolation. It answers how often, and says nothing about how much — and a strategy is paid in the second currency.
A 90% win rate is trivially achievable: take a one-pip profit and let losers run. A 20% win rate can be excellent if the winners are large enough. Neither number, alone, distinguishes the two.
The pairing is the payoff ratio, and together they produce expectancy. The break-even line between them is exactly the table in what is a good risk-reward ratio:
| Win rate | Payoff 1:1 | Payoff 1:2 | Payoff 1:3 |
|---|---|---|---|
| 30% | losing | losing | break-even |
| 40% | losing | winning | winning |
| 50% | break-even | winning | winning |
| 60% | winning | winning | winning |
Read a row across: 40% is a losing strategy at 1:1 and a comfortable one at 1:2. The win rate did not change.
Win rate has one genuine effect, and it is psychological rather than mathematical: it sets how long the losing streaks are. A 30% win-rate strategy with strong expectancy will still deliver runs of eight and ten losses regularly, and most people abandon a strategy during one.
So a lower win rate demands either a smaller position size, a longer commitment, or both. That is a real cost, and it is the honest reason to prefer a higher win rate at equal expectancy.
Worth knowing
If you are choosing between two strategies with the same expectancy, take the higher win rate — not because it earns more, but because you are more likely to still be trading it in a year.
A win rate estimated from a small sample is barely an estimate. The standard error on a proportion is roughly the square root of p(1−p)/n, and at a true rate of 50% that gives a 95% interval of about ±1/√n.
| Trades | Approx. 95% interval | Width |
|---|---|---|
| 20 | 28% – 72% | 44 points |
| 50 | 36% – 64% | 28 points |
| 100 | 40% – 60% | 20 points |
| 500 | 45.6% – 54.4% | 9 points |
| 1,000 | 46.9% – 53.1% | 6 points |
Twenty trades cannot distinguish a 30% strategy from a 70% one. That is not a quibble — it means the win rate from your first month contains almost no information, and any decision made on it is being made on noise.
This is the strongest practical argument for judging a strategy on expectancy in R and on its rule, rather than on its recent hit rate.
A win rate is also a statement about how the losses arrive, and that shapes whether a strategy is holdable more than the expectancy does.
| Win rate | Typical longest run | Not unusual |
|---|---|---|
| 30% | 8 losses | 12 |
| 40% | 6 losses | 10 |
| 50% | 5 losses | 8 |
| 60% | 4 losses | 7 |
These are properties of the distribution, not signs that anything has broken. A 40% strategy producing ten consecutive losses somewhere in two hundred trades is behaving exactly as a 40% strategy should.
Two errors pull in opposite directions during a losing streak, and most traders make both in sequence.
Worth knowing
Decide in advance what would constitute evidence that a strategy has stopped working — a number of trades, or a departure from the interval — and write it down before the streak, because during one the threshold moves.
Because the streaks are a property of the win rate, the choice between two strategies with equal expectancy is partly a choice about what you can hold through.
A 30% win-rate trend system and a 65% mean-reversion system can have identical expectancy and require entirely different temperaments. The first spends most of its time losing and makes its return in a handful of trades; the second wins routinely and gives it back in occasional large losses.
Worth knowing
The failure mode differs too. Low win rates are abandoned during streaks; high win rates are over-sized because they feel safe, until the loss that pays for all the wins arrives.
On its own, no. A 90% win rate is trivially achievable by taking tiny profits and letting losers run. Win rate only means something paired with the payoff ratio, and together they produce expectancy.
Several hundred. At 100 trades the 95% interval on a 50% win rate runs roughly 40% to 60% — wide enough to contain both a winning and a losing strategy.