WhitmanTrading

How to Measure Your Expectancy

To measure expectancy, express every closed trade as a multiple of what you risked on it, then take the average across all of them including the losses. A positive average means the approach earns per attempt; the number is only meaningful if no trades were left out.

Expectancy answers the only question that matters about an approach: does taking it repeatedly earn anything. Everything else — the win rate, the best trade, the feeling — is downstream of that number.

Before you start

A record of every closed trade, including the ones you would rather forget. A record with holes measures your memory rather than your method, and the holes are systematically the bad ones.

The risk you took on each, so results can be expressed in units of it. Entry minus stop, times size. Without it, a large win on a large position looks like skill.

A decision about the sample period, made before you look at the answer. Choosing the window after seeing the numbers selects the window that flatters you.

The steps

1. List every closed trade in the period

A candlestick chart with a series of completed positions marked.
Every trade, including the ones you left out last time. Illustrative chart - not real market data.

Date, instrument, entry, stop, exit, size. Nothing else is needed yet, and no trade is excluded for being unrepresentative.

2. Compute the risk on each one

The first half of a price series with a risk distance marked.
Entry minus stop, times size. That is one unit. Illustrative chart - not real market data.

The distance from entry to the stop you actually placed, multiplied by the position size. This is one R, and it is fixed at entry rather than at exit.

3. Express each result as a multiple of that risk

A section of the price series with results scaled to risk.
Every result becomes comparable once it is in units of risk. Illustrative chart - not real market data.

A trade that made twice what it risked is 2R. One that lost what it risked is −1R. A trade stopped worse than planned is −1.4R, and it is recorded that way.

4. Average across every trade, losers included

A window of price bars showing mixed outcomes averaged.
The average of everything is the number. Illustrative chart - not real market data.

Add the R multiples and divide by the count. That is expectancy: what the approach earns per attempt, in units of what each attempt risked.

5. Subtract the costs before believing it

The second half of a price series with costs deducted.
Costs come out of every attempt. Illustrative chart - not real market data.

Commission and spread are paid whether the trade won or lost. On a small average they are a large share of it, so a pre-cost expectancy is a different number from a real one.

6. Count the trades and be honest about the sample

A range-bound stretch with a small number of events.
Thirty trades is a story, not a measurement. Illustrative chart - not real market data.

Distinguishing a five-point edge from noise takes roughly 380 observations. A positive expectancy over 30 trades is an encouraging story and it is not yet evidence.

7. Recompute it monthly and keep the history

A long-horizon view with a measure tracked over time.
The trend in the number is more useful than the number. Illustrative chart - not real market data.

One figure tells you where you are. A series of them tells you whether it is improving, which is the question the measurement is actually for.

How to tell it worked

The trade count in your calculation matches the trade count in your broker statement. If it does not, the record has holes and the number describes a selection.

Losers are in the average. Recomputing with only winners should produce a visibly different figure — if it does not, they were already excluded.

The figure is quoted with its sample size. “0.32R over 214 trades” is a measurement; “0.32R” alone is a claim.

And costs were subtracted, so the number describes what reached the account rather than what the chart implied.

What the number cannot tell you

A candlestick chart annotated with the round-trip cost of a switch.
Costs are paid on every attempt regardless. Illustrative chart - not real market data.

It cannot tell you the next trade will work. It is an average over a sample, and on this site’s shared series direction runs average 2.01 bars with a longest run of 11 — streaks are ordinary, and an average says nothing about their order.

On the same series a round trip measures about 2% of the median bar range of 0.493. A strategy with an expectancy of 0.05R is inside that cost band, which means the measurement and the fee are the same size. The figures are in research/series-measurements.json.

A candlestick chart with a volume histogram beneath it.
And backtested fills are kinder than real ones. Illustrative chart - not real market data.

It also cannot see slippage you did not record. A stop filled worse than placed is a larger loss than −1R, and recording it as −1R quietly inflates the average.

Reading the number once you have it

A figure between 0 and 0.1R is inside the noise for most sample sizes. It may be real and it cannot be distinguished from zero without a very large record.

Between 0.1R and 0.3R is a workable edge if the sample is in the hundreds and the costs are already subtracted. Most durable approaches live here rather than higher.

Above 0.5R over a large sample is unusual, and the first thing to check is whether the record is complete rather than whether the strategy is exceptional.

And a negative figure is the most useful result the calculation produces. It says the approach loses per attempt, which no amount of position sizing or discipline corrects — the answer is to stop taking it rather than to take it better.

The original data

Of the 24,971 unique videos in research/search-study-corpus.jsonl, zero have expectancy in the title. Win rate appears in 161 videos at a median of 13,711 views across 130 channels, and trading journals in 29 at 24,870. The counts come from site/corpus_count.py, which deduplicates by video id.

A candlestick series with several gaps, the largest of them marked.
A gap produces a loss larger than one unit of risk. Illustrative chart - not real market data.

Zero videos on expectancy and 161 on win rate. The number that decides whether an approach pays has no coverage at all, while the number that cannot answer that question on its own has the most in the risk category — which is a fair summary of what the subject rewards attention-wise.

A stretch of price bars cut short at a decision point.
Expectancy is positive over 40 trades. Scale up? Illustrative chart - not real market data.

The answer to the question on that chart is that 40 trades cannot separate an edge from a good run. The arithmetic is on the sample size calculator, and it puts the requirement in the hundreds. A positive figure over 40 trades is a reason to keep the record going, not a reason to increase size — and increasing size is what turns a normal losing streak into a serious one.

When it fails

The failure is a record with quiet gaps, and it produces a number that looks respectable. Trades closed in frustration, positions exited without a plan, the week that went badly — none of them get entered, not through dishonesty but because entering them is unpleasant. The resulting expectancy describes a filtered version of the strategy, it is always better than the real one, and the filter removes exactly the trades that would have said something useful.

The second failure is averaging only the winners. That is not expectancy.

A third is quoting it without a sample size. The count is half the measurement.

A fourth is using pre-cost figures. Costs are paid on every attempt.

A fifth is recording every stop as −1R. Slippage makes some of them worse.

And a sixth is recomputing after every trade. The number moves on noise and invites reacting to it.

Trade review is the process that produces the record. Trading journal is where the raw entries live. And win rate is the number this one exists to replace.

What I actually do

The number that changed how I judged things was not my expectancy — it was seeing how much it moved when I added the twelve trades I had quietly not recorded. A record with gaps produces a number that describes a selection rather than a strategy, and the gaps are never random.

— Michael Whitman

This page is educational, not financial advice. Test every idea on your own charts before risking money.