Win rate is the most seductive number in trading — and the least meaningful on its own. Ninety small wins feel wonderful, and they mean nothing until you know what the ten losses cost. Sellers of systems lead with win rate precisely because it hides that question, and buyers accept it because winning often feels like the same thing as earning. It isn't.

The number that actually decides whether an account grows is expectancy: what the average trade pays once wins and losses are weighed together.

The formula and the pretty loser

Expectancy = (win rate × average win) − (loss rate × average loss). It is the answer to the only question that compounds: what does one trade of this system earn, on average, over many trades?

Now the pretty loser. A system wins 90% of the time, collecting $10 each win: +$9 per trade on average. It loses 10% of the time at $150 a loss: −$15 per trade. Expectancy: −$6. The account shrinks while the win rate sparkles — and its equity curve looks glorious right up until the losses land. Every martingale and most "signal services" live in this quadrant.

The ugly winner

Flip it. A system wins only 40% of the time, but wins average $300: +$120 per trade. It loses 60% of the time at $100: −$60. Expectancy: +$60 per trade — from a strategy that is wrong more often than it is right, feels terrible to trade, and would be nearly impossible to sell on a screenshot. Trend-following, the oldest continuously profitable strategy family in existence, lives here.

The lesson is not that low win rates are good. It is that win rate and payoff size are two halves of one number, and judging either alone is like judging a fraction by its numerator.

Worked check: 0.90 × 10 − 0.10 × 150 = −6. 0.40 × 300 − 0.60 × 100 = +60. Two minutes of arithmetic, and no screenshot survives it.

Questions to ask any track record

  • What are the average win, average loss, and total trade count? (Expectancy needs all three; fewer than a few hundred trades is anecdote, not evidence.)
  • What is the maximum drawdown, and how long did recovery take?
  • Are losing trades visible in the history — or are they floating, still open, somewhere off the screenshot?
  • Does the equity curve include spread, commission, and slippage, or is it a frictionless backtest?

Next: the drawdown asymmetry every trader underestimates →