EdgeWatchFXby ErgodicLabs

The RecordMethod

Win rate vs expectancy: why a 90% win rate can lose money

A 90% win rate sounds like an edge. It is only one half of the arithmetic. Expectancy, payoff and the break-even win rate decide whether a strategy made money per trade, and a short worked example shows how a 40% win rate can beat a 90% one.

3 min read

In short: win rate tells you how often a strategy won. Expectancy tells you how much it made per trade on average, once the size of the wins and the losses is counted. A strategy is profitable only when its win rate is above its break-even win rate, which depends on how large its losses are against its wins.

The formula

Expectancy per trade is:

expectancy = win rate x average win - loss rate x average loss

It can be written in money, in percent of the account, or in R, the amount risked per trade. A positive figure means the average trade made money; a negative one means it lost.

A 90% win rate that loses

A strategy wins 90% of its trades. The average win is 0.5% of the account and the average loss is 5.5%.

Expectancy = 0.9 x 0.5% – 0.1 x 5.5% = 0.45% – 0.55% = -0.10% per trade.

Nine winners in a row add 4.5%. The tenth trade takes back 5.5%. Over many trades the account drifts down, while the statement shows a win rate that looks strong.

A 40% win rate that pays

Another strategy wins only 40% of its trades. The average win is 2.5% and the average loss is 1.0%.

Expectancy = 0.4 x 2.5% – 0.6 x 1.0% = 1.0% – 0.6% = +0.40% per trade.

It loses more often than it wins, and a run of five or six losses in a row is ordinary for it. Its average trade is still positive, and by a wider margin than the first strategy’s is negative.

Figure Strategy A Strategy B
Win rate 90% 40%
Average win 0.5% 2.5%
Average loss 5.5% 1.0%
Payoff, or risk reward (win / loss) 0.09 2.5
Break-even win rate 91.7% 28.6%
Expectancy per trade -0.10% +0.40%

Risk reward ratio and the break-even win rate

Every strategy has a win rate at which it makes nothing. It depends only on the payoff, the average win divided by the average loss, which traders often call the risk reward ratio:

break-even win rate = average loss / (average win + average loss)

Strategy A needs to win 5.5 / 6.0 = 91.7% of its trades to break even. It wins 90%, so it loses money. Strategy B needs 1.0 / 3.5 = 28.6%. It wins 40%, so it has room to spare.

The useful question is not how often it won. It is how far its win rate sat above the win rate it needed.

That gap is what to watch. A strategy that wins 90% when it needs 91.7% has no margin at all; a small change in spread, slippage or market conditions moves it further below. A strategy that wins 40% when it needs 28.6% can absorb a lot before it stops paying.

Why very high win rates need more trades

The average loss is the number that decides a high-win-rate strategy, and it is measured on very few trades. In 50 trades at a 90% win rate there are about five losses. The size of a typical loss rests on those five, and one larger loss changes the expectancy completely.

The win rate itself is uncertain too. With 50 trades, a measured 90% has a 95% range of roughly 79% to 96%. For Strategy A, whose break-even rate is 91.7%, most of that range lies below break-even. A record with almost no losing trades has not yet shown what a loss costs. See how many trades before a track record means something.

How win rates get inflated

Some strategies have high win rates by construction:

  • Grids close each basket at a small profit and keep the losing ones open. If every position in a basket counts as a trade, one basket becomes many winners. Counted as one bet, the win rate is much lower. See why a smooth balance curve can hide a grid.
  • Wide or no stops turn many would-be losers into slow winners, at the cost of rare, large losses.
  • Holding losers until they come back keeps them out of the closed record until one does not come back. See losers held longer than winners.

In each case the win rate goes up and the average loss, when it finally arrives, goes up with it.

What to read instead of win rate alone

  • Expectancy per trade, in percent of the account or in R.
  • The break-even win rate, and how far the real win rate sits above it.
  • The number of losing trades the average loss is measured on.
  • Profit factor, gross profit divided by gross loss. Strategy A’s is 0.45 / 0.55 = 0.82; anything under 1 lost money.

How EdgeWatchFX reads it

The Edge quality test reads expectancy against the average loss and the win rate against its break-even rate, not the win rate alone. A grid basket counts as one trade in every statistic. The score also checks whether the record has seen enough losing trades to show what a loss costs, and how a full stop-loss compares with the usual win. The free statement check reads a file for the patterns that inflate a win rate: averaging down, sizing up after losses, missing stops and holding times.

Figures in this note describe recorded results. None of it is a forecast or investment advice.

More from The Record

What changes when an EA goes live

A backtest answers whether a strategy could have worked. A live account answers a different question: is it still behaving like itself? Here is how to tell a…

Method2 min read