Expectancy Explained

Win rate is the most prominently displayed and least informative statistic in trading, and expectancy is what it should be replaced with.

The calculation

Multiply the win rate by the average win, multiply the loss rate by the average loss, and subtract the second from the first.

A strategy winning 40% of the time with average wins of $300 and average losses of $100 produces 0.40 times 300, minus 0.60 times 100 — a positive expectancy of $60 per trade.

A strategy winning 70% of the time with average wins of $100 and average losses of $300 produces 0.70 times 100, minus 0.30 times 300 — a negative expectancy of minus $20 per trade, despite winning more than twice as often as it loses.

Those two examples are the entire argument. The second strategy looks far better on every leaderboard and loses money.

Why high win rates are easy to manufacture

This is the part that matters for evaluating anyone else's record.

A high win rate can be produced deliberately by taking small profits quickly and letting losses run. Close winners at a small gain and most trades are winners. Hold losers hoping for recovery and few of them are realised as losses yet.

The result is a record that looks excellent right up until the accumulated open losses are realised. Nothing about the win rate warned of it, because win rate cannot see position size or holding behaviour.

The pattern is common enough in copy trading marketplaces that it is worth checking explicitly. Evidence from a 2025 study of crypto copy trading found leaders showing win rates above 57% who still delivered losses to followers, because average losses exceeded average wins.

If a provider or platform shows win rate but not average win and average loss, the omission is informative.

What expectancy still does not tell you

Positive expectancy is necessary and it is not sufficient, and the gaps are worth naming.

It says nothing about the path. Two strategies with identical expectancy can have completely different drawdown profiles. One grinds steadily; the other produces the same average through violent swings you may not survive holding.

It is computed from a sample. Expectancy from fifty trades is a weak estimate. Expectancy from a period covering one market regime describes that regime.

It hides tail risk. A strategy with a long record of small gains and one catastrophic loss not yet in the sample shows positive expectancy right up until it does not. Selling options produces exactly this shape.

It assumes costs are included. An expectancy computed before commissions, spread, and slippage is not describing an outcome anyone experienced. For high-frequency strategies the difference is frequently the entire edge.

Expectancy and position sizing

The connection people miss.

Expectancy per trade multiplied by the number of trades gives expected total return, which makes frequency look like a lever — trade more, earn more, if expectancy is positive.

That reasoning holds only if each trade is sized so a losing sequence is survivable. Positive expectancy over many trades is worthless if a run of losses ends the account before the average asserts itself, and losing runs are longer than intuition suggests even for good strategies.

This is why sizing is upstream of expectancy rather than downstream of it. The divide-by-20 rule caps any single position at available trading capital divided by twenty precisely so that the sample has room to play out.

Measuring your own

Straightforward and rarely done.

Log every trade with its result. Compute win rate, average win, and average loss separately, and derive expectancy from all three rather than reading win rate alone.

Then break it down. By time of day, by instrument, by market condition. A strategy is frequently profitable in some segments and not others, and an aggregate figure conceals that in a way that a breakdown makes immediately actionable.

Include costs. Spread, commissions, and measured slippage all belong inside the numbers, because they are inside the outcome.

What this means for automation

Two points.

Report expectancy, not win rate. A dashboard displaying win rate prominently is training its user to evaluate on the wrong number. Average win and average loss belong alongside it.

Beware optimising for it. Expectancy computed on historical data and then maximised by parameter tuning is a fitted result, and the tuning that improves it in-sample frequently degrades it forward. The more parameters involved, the more this applies.

Computing these statistics across a trade history is periodic work rather than order-path work, and it belongs on worker thread pools so analysis cannot delay execution.

The honest limits

Expectancy is an estimate from a sample, and small samples produce unreliable estimates that look precise because they are expressed as a number.

It describes the average and says nothing about the distribution, which is what determines whether you can hold the strategy through its bad stretches.

And a positive figure is not a guarantee of anything. It is a description of what happened in a particular period under particular conditions. Position sizing is what makes surviving long enough for an edge to assert itself possible, and it works on infrastructure you control whether or not the expectancy estimate was right.

Frequently asked questions

What is expectancy in trading? The average amount a trade is expected to make or lose, computed from win rate together with average win and average loss.

Why is win rate misleading? It describes frequency, not profitability. A strategy winning 70% of the time loses money if losers are three times the size of winners.

How is expectancy calculated? Win rate times average win, minus loss rate times average loss.

Does positive expectancy mean a strategy is safe? No. It says nothing about drawdown path, sample size, tail risk, or whether costs were included.

Should I optimise for expectancy? Carefully. Maximising it on historical data is a fitted result, and heavy tuning tends to degrade forward performance.

Disclaimer: This article is educational content about trading mechanics and software. It is not investment advice, financial advice, tax advice, legal advice, or a recommendation to buy or sell any security, nor a recommendation of any strategy, position structure, or order type. Any instruments, figures, or examples are used solely to illustrate mechanics. Options and futures trading involve substantial risk of loss and are not suitable for all investors; selling options can produce losses substantially greater than the premium received. Please read Characteristics and Risks of Standardized Options before trading options. Automated trading carries additional risks including software defects, connectivity failures, broker API changes, and outages that may prevent orders from being placed, modified, or cancelled. Past performance does not indicate future results, and no configuration, structure, position-sizing rule, or risk setting can guarantee a profit or prevent a loss.

StaxInvesting LLC sells self-hosted trading software. It is not a broker-dealer, investment adviser, or financial institution, and it does not manage accounts, hold member funds, place trades on behalf of members, or access member brokerage accounts. Members run the software in their own cloud environment, connect their own brokerage accounts under their own credentials, and are solely responsible for their configuration, their credential security, and every trade executed in their account. Broker order handling, approval levels, and available features vary; verify against your broker's current documentation. Consult a qualified financial adviser and tax professional regarding your individual circumstances.