You can journal every trade, respect every stop, size perfectly, and stay ice-cold under pressure — and still lose money forever. Discipline executes a system faithfully. It cannot fix a system that loses. That’s the trap: a negative-expectancy strategy fails you slowly enough that you blame yourself instead of the math.
What expectancy is and why it decides your long-run result
Expectancy is the average amount you win or lose per trade, over many trades. Expressed in R (units of your initial risk), it answers one question: if I take this trade a thousand times, what do I net per attempt?
- Positive expectancy: each trade is, on average, worth more than zero. Volume and time work for you.
- Negative expectancy: each trade is, on average, worth less than zero. Volume and time work against you.
Nothing else about trading matters more. Position sizing controls how fast your account moves; expectancy controls which direction. A brilliantly risk-managed negative-expectancy system just bleeds slowly instead of quickly. If you want the full derivation, start with what trading expectancy is — but the intuition above is enough to see why this is the master variable.
Win rate vs average R: the two-variable trap
Expectancy is built from two numbers that traders love to consider separately and must consider together:
- Win rate — the fraction of trades that win.
- Average R — the average size of a win versus a loss, in units of risk.
The simple form: Expectancy = (Win% × Avg Win in R) − (Loss% × Avg Loss in R).
The trap is optimizing one variable while ignoring the other. Traders chase a high win rate because winning feels like edge. But a 70% win rate paired with wins that are half the size of losses is a losing system. And a 35% win rate with wins three times the size of losses is a strong one. You cannot judge a strategy from either number alone — only from how they combine. Thinking in R-multiples forces both into the same equation, which is the whole point.
How a high win rate can still be negative expectancy
This is the single most expensive misunderstanding in trading, so make it concrete.
Suppose you win 8 out of 10 trades. Feels unstoppable. But your winners average +0.4R and your two losers average −2R each (you let losers run, or your stop was wide relative to your target). Run the math:
(0.8 × 0.4) − (0.2 × 2.0) = 0.32 − 0.40 = −0.08R per trade.
Eighty percent wins, and you lose 0.08R every single trade on average. Over a funded account’s typical volume, that’s a steady drain masked by a scoreboard that keeps flashing “win.” This pattern — many small wins funding a few large losses — is how disciplined traders quietly go broke while feeling successful. The equity curve looks fine for a while, then a single cluster of the big losses erases weeks of the small wins.
The lesson: never let a comforting win rate substitute for measuring the whole equation.
Measuring expectancy from your own trade history
Textbook examples are clean; your trading isn’t. The only expectancy that matters is yours, computed from your filled trades — not a backtest, not a vibe.
- Export or pull your closed trades with their actual entries, exits, and stops.
- Convert each result to R (profit or loss ÷ the risk you took on that trade).
- Average across all of them. That average is your realized expectancy.
Run your numbers through an expectancy calculator to get the figure without arithmetic errors. The friction, of course, is having clean data in the first place — which is exactly where most manual journals fall apart. Shibiki’s auto-journaling records every fill automatically, so your expectancy is computed from what actually happened rather than from the trades you remembered to log. Its live edge health then tracks that expectancy per strategy in real time, so a system sliding negative shows up as a trend, not a surprise at the end of the month.
Sample size and confidence before trusting a number
A positive expectancy over 12 trades is not evidence — it’s noise wearing a nice hat. Small samples produce wild swings in both directions; you can look brilliant or broken for reasons that have nothing to do with your edge.
The question is never just “is my expectancy positive?” but “how confident can I be that it’s positive?” That’s a function of sample size and consistency of results. This is why Shibiki wraps a Wilson confidence interval around your win rate: at 15 trades the interval is wide, and honesty forces you to say “I don’t know yet.” At 150 trades it tightens, and only then does a positive number start to mean something. A confidence interval turns “I’m probably fine” into a defensible statement — or exposes that you’re trading on a hunch.
Fixing or dropping a negative-edge system
Once you’ve measured a genuinely negative edge on a real sample, you have exactly two honest options:
- Fix the equation. Attack whichever variable is dragging you down. Cutting losers faster raises average R. Tightening entry criteria can raise win rate. But change one thing and re-measure on a fresh sample — don’t rebuild everything and lose the ability to attribute the improvement.
- Drop it. Some systems are negative because the underlying premise has no edge. No amount of discipline rescues them. Retiring a strategy is not failure; funding it forever is. Compare how dedicated review tools like Edgewonk frame the same decision — the throughline is always: measure, then act on the measurement.
The mistake isn’t having a negative-expectancy system. Everyone builds a few. The mistake is not knowing, and paying for the ignorance one small, comfortable win at a time.
Related: Trading expectancy · Expectancy calculator · R-multiple