Win rate alone is a vanity metric. A 70% win rate can still bleed capital if losers are larger than winners. Expectancy combines win rate, average win, and average loss into one number: expected R per trade. That is the metric professionals use before scaling size.
This guide walks through the expectancy formula in R-multiples, a full worked example, how to segment by setup, and how often to recalculate so you know whether your edge is real — not just lucky.
The expectancy formula (in R)
When trades are logged in R-multiples, expectancy is the average R across all trades in your sample. The simplest form:
The equivalent form is useful when you want to see which lever to pull: raise win rate, increase average win R, or shrink average loss R. All three paths can improve expectancy, but only measurement shows which one your process actually needs.
Worked example
Suppose your last 10 trades in R were: +2, +1, −1, +0.5, −1, +3, −1, +1, −1, +2. Sum = +5.5R over 10 trades → expectancy = +0.55R per trade.
Using the alternate formula: 6 wins averaging +1.58R and 4 losses at −1R → (0.6 × 1.58) + (0.4 × −1) = 0.948 − 0.4 = +0.55R. Both methods match.
That means if you keep the same process, you expect to make 0.55R per trade on average. On 100 trades at 1R = $100 risk, that is roughly +$5,500 expected — before costs, slippage, and the reality that variance will not land exactly on the mean.
What expectancy tells you
- Positive expectancy (+0.2R and above is often workable) → edge may be real
- Near zero → breakeven system; commissions may erase gains
- Negative → process loses over time regardless of lucky streaks
- Rising expectancy over months → improvement in execution or setup selection
- Falling expectancy → rule drift, regime change, or size stress — investigate before adding risk
Expectancy vs profit factor
Profit factor = gross winning R ÷ gross losing R (absolute value). A profit factor above 1.0 means winners outweigh losers in total. Expectancy and profit factor usually agree on direction but answer different questions: expectancy is per-trade (better for sizing psychology); profit factor is ratio-based (common in system comparison). Track both if you like, but prioritize expectancy for weekly review.
Segment expectancy by setup
Overall expectancy hides weak setups. Tag trades by pattern and calculate expectancy per tag. You might find +0.8R on “breakout continuation” and −0.3R on “counter-trend fade” — that is actionable: do more of what works, pause what does not, and never scale the blended average blindly.
Require a minimum sample before killing a tag — often 20–30 trades. Below that, label the result “insufficient data” instead of overreacting to variance.
How often to recalculate
Review expectancy weekly on a rolling 20–50 trade window for active strategies, and monthly on the full journal. Sudden drops often trace to rule breaks, oversizing after wins, or a market regime that no longer fits your setup rules.
Use the trader diary to find when expectancy shifted: was it a single bad session, a week of overtrading, or a slow drift across a month? Timing the change narrows the fix.
Common expectancy mistakes
- Calculating expectancy in dollars while size varies every trade
- Excluding breakeven trades to inflate win rate and expectancy
- Mixing paper and live trades in one average
- Scaling up after 10 good trades when sample size is still tiny
- Ignoring tag-level expectancy and trusting the headline number
How Traderizz helps
Traderizz surfaces expectancy on your overview dashboard automatically from logged RR — no spreadsheet formulas after each session. Filter by journal, date range, or tag to see whether expectancy holds for the setups you actually trade.
Pair overview expectancy with the trader diary: when the number dips, drill into the sessions that drove it. Jarvis can answer questions about your journal data so you spend review time on decisions, not hunting through rows.
When you change one rule — stricter entries, wider targets, smaller size — run a before/after expectancy comparison on the next 20 trades. Short A/B windows keep feedback loops tight without waiting a full quarter.