Expectancy: the only metric that actually matters
You can win 70% of your trades and lose money. You can win 30% and be very profitable. Win rate alone, without its partner metric, tells you absolutely nothing about whether your system works.
Ask a trader how they're doing and they'll almost always answer with their win rate: "I win 65% of my trades." Sounds good. Feels like good news. And yet that number, on its own, says nothing about whether that trader makes or loses money long term — because win rate is only half the equation, and the missing half can completely flip the conclusion.
Why win rate alone can lie
Imagine a trader who wins 70% of their trades. Each win is $50. Each loss, in the remaining 30%, is $200. Over a hundred trades: 70 wins of $50 ($3,500) minus 30 losses of $200 ($6,000). That trader, with an enviable 70% win rate, loses $2,500 every hundred trades. The number that sounded like success was hiding a losing system.
Now imagine the opposite case: a trader who wins only 30% of their trades, but each win is $400 and each loss is $100. Over a hundred trades: 30 wins of $400 ($12,000) minus 70 losses of $100 ($7,000). That trader, with a win rate that sounds mediocre, makes $5,000 every hundred trades. The number that sounded like failure was hiding a very profitable system.
The formula that combines both halves
Expectancy = (Win rate × average win) − (Loss rate × average loss). The result is what you can expect, on average, to make or lose per trade you take. A positive number means the system, sustained over time with consistent position sizing, is profitable. A negative number means that no matter how many winning trades it has along the way, the system loses money long-term.
This formula is why two traders can look at the same hundred-trade history and reach opposite conclusions: one looks only at the win rate and feels good or bad based on that single number; the other calculates expectancy and knows, with mathematical precision, whether the system has a real edge.
Many of the most profitable systematic and quant fund strategies have win rates of 35% to 45% — numbers a retail trader, looking only at that figure, would dismiss as "a bad system." What makes those systems profitable is a payoff ratio (average win divided by average loss) high enough to more than compensate for the majority of losing trades.
This also explains why chasing a higher win rate, without looking at payoff, can make a system worse instead of better — cutting winners short raises win rate and sinks payoff at the same time.
How to calculate yours, with your own journal
Take your last fifty or hundred trades. Calculate what percentage you won (win rate) and what percentage you lost (loss rate, simply 100% minus win rate). Calculate the average, in dollars or in R, of your winning trades, and separately the average of your losing trades. Apply the formula. That number — positive or negative — is the most honest answer that exists to the question "does my system work?", far more honest than any subjective feeling about how your week went.
Why this changes how you should react to each trade
If your expectancy is positive and calculated over a large enough sample, an individual losing trade isn't a sign that something's wrong — it's, literally, part of what your own system predicts will happen some of the time. Reacting to that individual loss by changing the system, sizing up to "recover," or abandoning the strategy after a short losing streak, is reacting to short-term noise instead of trusting the long-term signal your own expectancy already gave you.
Expectancy doesn't tell you what's going to happen on the next trade. It tells you what's going to happen, on average, if you keep taking that same type of trade a hundred or a thousand more times. That's the timescale trading is actually played on.
The problem is almost no trader calculates their real expectancy — most operate guided by how the day, the week, or the last trade felt, which is exactly the noisiest and least informative timescale there is. Calculating it once, with real data, is often enough to completely change which decisions actually make sense and which are just emotional reaction dressed up as analysis.