Why a 70% win rate can still lose money

The arithmetic, in one line

Expectancy per trade equals the win rate multiplied by the average win, minus the loss rate multiplied by the average loss. If that number is positive, the strategy makes money over many trades. If it is negative, no win rate will save it.

Written out: E = (W x AvgWin) - ((1 - W) x AvgLoss), where W is the win rate as a fraction. Everything in this article follows from that single expression.

Two strategies with the same 70% win rate

Consider two systems, both winning 70% of the time over a hundred trades. The first wins 1 unit when it wins and loses 1 unit when it loses. The second wins 1 unit when it wins and loses 3 units when it loses — a very common shape in strategies that take profits quickly and let losers run.

Why the second shape is so common

Strategy B describes a behaviour almost everyone has experienced: take the profit as soon as it appears, and hold the loser in the hope it comes back. The result is a high hit rate and a fat left tail. The account feels good for weeks and then gives back months in a handful of trades.

There is a second reason it survives in backtests. If those large losses happen rarely, a short test period may not contain any of them. The strategy looks robust because its worst case has not been sampled yet.

Now subtract costs

Every trade pays the spread and the commission, and losing trades pay it too. Costs enter the expectancy as a subtraction from both sides, which means a strategy with many small trades needs a meaningfully larger edge than the same edge harvested slowly.

This is why a strategy that looks profitable in a zero-cost backtest can be unprofitable in reality without anything else changing. The edge did not disappear — the costs caught up with it.

Payoff ratio is the other half

The payoff ratio is the average win divided by the average loss. Pair it with the win rate and you have the whole story. A 40% win rate with a payoff ratio of 2.0 is a healthy trend-following shape; a 70% win rate with a payoff ratio of 0.4 is a losing shape that feels like a winning one.

Neither number means anything alone. Any performance report that shows a win rate without the average win and average loss is telling you half a sentence.

How to check a strategy you are looking at

• Find the average win and the average loss, or the profit factor, on the report. If they are missing, ask for them.
• Compute expectancy yourself and see whether it survives your real costs.
• Look at the largest single loss against the average win. A ratio above five deserves an explanation.
• Check the trade count. Thirty trades is not enough to distinguish a 60% win rate from a 45% one.
• Look for the losing streaks, not the losing trades. Ten losses in a row is what actually breaks accounts.

The takeaway

Stop shopping for win rates. Look for expectancy that survives costs, with a loss distribution you can sit through. A strategy that wins 45% of the time and makes money is not worse than one that wins 70% and loses — it is simply less flattering to look at, and considerably more likely to still be trading next year.