What a win rate tells you, and what it hides

A win rate is the share of trades that made money. If a strategy closed 100 trades and 80 of them ended above the price paid, its win rate is 80%. It is the most quoted number in trading, and it is one of the easiest to misread.

This note explains how to work one out, why a high win rate is not the same as a good result, and which figures to read beside it. It uses our own published numbers, because they show the gap clearly.

How to calculate a win rate

Count the closed trades. Count the ones that made money. Divide the second number by the first and multiply by 100. That is all a win rate is.

Two choices change the answer, so check them whenever you see the figure quoted. The first is what counts as a trade. We count only closed trades: a position that is still open, and perhaps down, is not in the figure, because its result is not known yet. Some tools count open positions at their current price, which gives a different number. The second is what counts as a win. We count a trade as a win if it sold above the price paid, with no trading costs. Add commission, spreads or tax, and some small wins become losses.

Why a high win rate can mislead

A win rate says how often a strategy was right. It says nothing about how much it made when it was right, how much it lost when it was wrong, or how long the money sat idle between trades.

A strategy that takes many small profits and holds its losers until they recover can show a very high win rate. So can one that sells quickly whenever it is ahead. Neither tells you whether the money would have done better simply held in the same stock or fund for the whole period.

Time out of the market matters most. A rule that buys after a fall and sells after a bounce spends long stretches in cash. If the stock rises strongly while the rule waits, the rule can win most of its trades and still end far behind someone who bought once and held.

What our own numbers show

Our published cards report two figures side by side for this reason: the share of closed trades that made money, and the share of backtests that beat buying and holding the same ticker over the same dates.

On SPY, the S&P 500 fund, 344 of 392 closed trades from the pullback strategy made money, a win rate of 87.8%. Yet only 3 of 553 backtests beat holding SPY over the same dates, which is 0.5%. On NVDA, 329 of 401 closed trades made money (82.0%), and again 3 of 553 backtests beat holding (0.5%). On AAPL, 502 of 644 closed trades made money (78.0%), and none of the 553 backtests beat holding.

So a strategy can be right four times out of five and still finish behind simply holding the same ticker. That is not a quirk of one ticker. Across the published cards, most strategies we test do not beat holding.

Figures are from the StrategyLens published cards, updated 29 October 2026. They are hypothetical backtests with no trading costs, before tax, and they will change as the cards are refreshed.

The figures to read beside a win rate

Beat buy-and-hold rate. The share of tests in which the strategy finished ahead of buying and holding the same thing over the same dates. This is the comparison most people actually care about.

Average win and average loss. A 90% win rate with tiny wins and a few large losses can lose money overall. Expectancy combines them: the win rate times the average win, minus the loss rate times the average loss.

Return per year (CAGR). How fast the money grew, compounded. It lets you compare a strategy with holding on the same scale.

Maximum drawdown. The largest fall from a peak. It tells you what you would have had to sit through.

Number of trades and years tested. A win rate from 12 trades says little. One from several hundred trades over decades says more, though it still describes the past.

A quick check you can do yourself

When you see a high win rate quoted, ask three questions. Is it closed trades only? Does it include costs? And what did buying and holding the same thing over the same dates do? If the answer to the last question is missing, the win rate on its own does not tell you whether the strategy was worth following.

You can see both figures for each ticker on our ticker pages, and the method behind them on the methodology page.

Win rate, expectancy and the size of each trade

Expectancy is the figure that joins a win rate to the size of the wins and losses. It is the average amount a trade makes, counting the losers. Here is an illustration, not a real strategy: a rule wins 80% of the time, its average win is £50 and its average loss is £250. Expectancy is 0.8 × £50 minus 0.2 × £250, which is £40 minus £50, or a loss of £10 a trade. A rule that is right four times out of five loses money on average.

Now flip it. A rule wins only 40% of the time, but its average win is £300 and its average loss is £100. Expectancy is 0.4 × £300 minus 0.6 × £100, which is £120 minus £60, or a gain of £60 a trade. Some trend-following approaches look like this: wrong more often than right, with occasional large wins.

Neither number tells you about time in the market. A rule with a positive expectancy can still trail buy-and-hold if it trades rarely and sits in cash while prices rise, which is why the comparison with holding is the last and most important check.

This is research, not financial advice. Past results, real or simulated, do not tell you what will happen next.

How do you calculate a win rate?

Divide the number of closed trades that made money by the total number of closed trades, then multiply by 100. Check whether open positions and trading costs are included, because both change the answer.

Is a high win rate good?

Not on its own. A strategy can win most of its trades and still end behind buying and holding the same stock, because it sits in cash while prices rise. Read the win rate beside the beat buy-and-hold rate and the return per year.

Why do StrategyLens pages show two figures?

One is the share of closed trades that made money. The other is the share of backtests that beat buying and holding the same ticker over the same dates. They answer different questions, so both are shown.