Win Rate Is a Vanity Metric (And What Counts Instead)

In the Lexicon of the Abyss, this is the letter with the most thankless task: it must dethrone a number everyone loves. The win rate — the hit rate — is the most charming metric in trading. It is easy to understand, easy to show and hard to resist. It has only one flaw: it answers the wrong question.

What the number says — and what it does not

The definition is quickly dealt with: the win rate is the share of trades that close in profit. 54 out of 100 trades positive — win rate 54 %.

So the number answers the question: how often was the system right? And here lies the design flaw, for the markets do not pay for being right. What is paid is the difference — what the hits bring in, minus what the misses cost. About both, the win rate is entirely silent. It counts events and does not weigh them.

A numerical example, deliberately stark: a system wins one unit in 90 out of 100 cases and loses fifteen in 10 cases. Win rate: a proud 90 %. Balance: 90 won, 150 lost — a hole of 60 units, profit factor 0.6, a leak without mercy. Conversely, a system with a 35 % hit rate can earn brilliantly if its winners average four times its losers. Being right and earning are two different professions.

The missing half: the payoff ratio

What the win rate lacks has a name: the payoff ratio — the average winning trade divided by the average losing trade. Only this pair together yields the expectancy, the actual answer to whether an approach holds.

The two quantities stand in a natural exchange, almost like communicating vessels: whoever secures profits early and gives losses room buys hit rate at the cost of payoff. Whoever cuts losses fast and lets profits run pays with many small red trades for occasional large green ones — low win rate, high payoff. Neither style is superior in itself. But one of the two feels better, and exactly there the problem begins.

The vanity of the hit rate

For why is this letter titled as it is? Because the win rate’s primary issue is not statistical weakness but psychological function: it flatters. Being right is rewarded more deeply in humans than earning — the ego keeps its own account, and on that account every hit scores a point, no matter how small.

From this grows a quiet corruption more dangerous than any calculation error: you begin to optimise your own behaviour towards the hit rate without noticing. Profits are secured earlier — then the hit is official. Losses get “a little more time” — as long as the position is open, the miss does not count. Each of these individual decisions polishes the win rate and ruins the expectancy. It is the two demons in their finest disguise: not as a storm, but as the ego’s bookkeepers.

The old schools knew this pattern under another name. The strict teachers warned against performing the exercise for the sake of a pretty protocol — whoever wants to impress the diary has stopped observing and started posing. The remedy is the same as it was then: measure what counts, not what shines.

What counts instead

The lexicon’s order of rank, in one line: expectancy before hit rate, ratio before event, sample above all.

Concretely: the profit factor carries the main load, for it weighs all wins against all losses. The maximum drawdown stands beside it, for it measures what the ratio costs when things get serious. Sample size ennobles both — and the win rate? It stays in the toolbox, but as a context number: it describes a system’s character (many small hits or rare large ones), not its quality. Useful as description, worthless as proof.

In this site’s documentation the win rate is therefore shown — but never alone. It stands in the protocol next to profit factor, drawdown and trade count, as one stroke in the portrait, not as the portrait. The machine whose trial runs here was built for expectancy, not for beauty of the quota; on some days it will be wrong more often than right and still do its work.

So the next time a hit rate is handed to you as proof — by a stranger on the net or by your own ego after a good week — ask the one question this letter gives you: and what does a miss cost? By the answer you will recognise whether you are looking at a chronicle or a shop window.