What does win rate actually tell you?
Win rate tells you how often a strategy was right and nothing at all about how much being right was worth, which is why it can be moved almost at will without changing the underlying edge.
The mechanism is simple. Widen the stop and tighten the target and the win rate goes up, because the target is now closer and easier to reach. The losses get bigger in exact compensation. Nothing about the strategy's expectation has changed; only the shape of its outcomes has.
That makes win rate the most reliably marketed number in trading. A vendor quoting an 85% win rate has told you they chose a wide stop, and nothing else.
It does carry one genuinely useful piece of information: what the strategy will feel like to trade. A 30% win rate means long losing streaks are routine, and knowing that in advance is the difference between expecting them and abandoning the system during one.
Is profit factor a better number?
Profit factor — gross profit divided by gross loss — is more informative than win rate because it accounts for size as well as frequency, and it has its own specific failure mode.
A profit factor above 1 means the strategy made more than it lost. Values around 1.2 to 1.6 are typical of real, durable retail strategies once costs are included; much higher figures in a backtest usually mean the same things a very high Sharpe means.
The failure mode is concentration. A profit factor of 2.0 where a single trade contributed most of the gross profit is not a 2.0 strategy — it is a strategy that caught one move. Remove that trade and the figure often collapses toward or below 1.
The check takes one minute and is almost never run: recompute profit factor with the largest winning trade excluded. If the strategy still works, the number describes a process. If it does not, the number describes an event.
Which pair of numbers actually describes a strategy?
Expectancy in R alongside the trade count, with profit factor and win rate as context — because expectancy is the only figure that combines frequency and size into what you actually earn per unit of risk.
The trade count is not optional beside it. Expectancy is an average of a wide distribution, and the same figure over 60 trades and over 600 supports completely different conclusions.
Win rate and profit factor then describe the SHAPE of the outcomes, which is what determines whether you can trade the thing. Two strategies with identical expectancy — one at 65% win rate and one at 28% — demand very different temperaments, and the second one will be abandoned by most people during its first ordinary losing streak.
Reporting all four together costs nothing and removes most of the ways a summary can mislead, including the ways it can mislead the person who built it.
How do costs change these numbers?
Costs hit win rate hardest for short-horizon strategies, because a fixed cost per trade is a larger fraction of a small target — and a strategy can cross from winning to losing without its win rate moving much at all.
Consider a scalping rule targeting 8 pips with a 1.2 pip round-trip cost. That is 15% of the target consumed before anything happens, applied to every trade. The win rate barely shifts; the expectancy shifts enormously.
Profit factor degrades in the same direction but more visibly, since costs reduce the numerator and increase the denominator simultaneously. A gross profit factor of 1.5 can become a net 1.05 on a high-frequency strategy, which is the difference between a business and a hobby.
The practical rule is to compute every metric net, once, and never quote a gross figure — because a gross number quoted alongside a net one will be the one that gets remembered.
There is one more asymmetry worth knowing. Costs do not merely reduce the average result; they reduce the number of strategies that have any result at all, because a marginal edge sits close enough to zero that a realistic cost model moves it to the wrong side. That is why so many strategies look promising gross and vanish net, and why the cost assumption deserves as much scrutiny as the entry rule it is being applied to.