TEST THE CLAIM
Expectancy needs a distribution, not a win rate.
Expectancy combines the probability and average size of gains and losses. A positive historical average is a sample estimate, not a promise, and its uncertainty depends on dispersion, sample size and how independent the observations really are.
Before you begin
Learning objectives
- Explain: Calculate the whole payoff
- Explain: Precision depends on more than count
- Explain: Write the probability question first
Calculate the whole payoff
A simple trade expectancy is win probability multiplied by average win, plus loss probability multiplied by average loss, after modelled costs. A hypothetical 40-trade sample with an 18/22 win-loss split, +1.4R average win and -0.8R average loss estimates +0.19R per trade. The arithmetic does not establish stability.
Precision depends on more than count
Larger samples generally narrow uncertainty around a mean, but clustered trades from one regime are not the same as independent evidence across conditions. Report dispersion, time coverage and the number of distinct market environments alongside the average.
Write the probability question first
Decide what change matters, what confidence or tolerance is required and which data are eligible before inspecting the result.
- Separate research and held-out periods
- Include costs and rejected variants
- Show loss size and tail observations
- Do not convert a confidence interval into a probability of future profit
Common mistakes
- Reading the result without the stated scope and assumptions.
- Changing the rule after seeing an outcome while still calling the data unseen evidence.
Trader Checklist
- Separate research and held-out periods
- Include costs and rejected variants
- Show loss size and tail observations
- Do not convert a confidence interval into a probability of future profit
Practice exercise
Choose one of your own trading examples and write one page of rules, evidence and stopping conditions using the principles in "Expectancy needs a distribution, not a win rate.".
What evidence would overturn the conclusion
The conclusion should be overturned or narrowed if a key assumption cannot be reproduced inside the supported scope, the control cannot be executed, or new out-of-sample evidence repeatedly contradicts it.