Take ten trades of +2, −1, −1, +1.5, −1, +0.5, −1, +2, −1 and −0.4 R. They sum to +0.6R, so expectancy is +0.06R per trade. The same number follows from win rate times average win minus loss rate times average loss: 0.4 × 1.5 − 0.6 × 0.9.
It answers the question a win rate cannot: what is a typical trade worth? A method that wins seven in ten times with small wins and large losses can have a negative expectancy, while one that wins one in three with large wins and small losses can have a positive one.
Expectancy is an estimate from a sample, not a fixed property. With few trades the figure is noisy, and the spread of individual results can be much larger than the average. The estimate needs a decent number of trades before it deserves trust.
It is most useful for comparing parts of your own record. Expectancy by setup, by session, by time since the last loss or by emotional state shows where the average comes from and which parts are dragging it down.
A positive expectancy is necessary for a method to make money over many trades, but it is not sufficient: it says nothing about how deep the drawdowns run or whether the position size is survivable.