Expected Value vs Variance, In Plain English
Positive EV doesn't mean you'll make money this year. It means you should keep doing it.
Expected value (EV) is the average outcome over many trials. Variance is how much individual outcomes scatter around that average. A coin flip at +110 is positive EV, and it can lose ten times in a row without telling you anything about whether the edge is real.
Two failure modes follow from confusing the two:
Failure mode 1: you take a positive-EV position, lose three times, and abandon the strategy. The EV was real. Your sample was tiny. You walked away with a sound process and a bad result.
Failure mode 2: you take a negative-EV position, win three times, and double the size. The wins were variance. The expected value was still negative. You are now sized up on a losing strategy.
The defense against both is process discipline. Track the thesis, not the outcome. Track many decisions, not a few. ROW's 'confidence accuracy' metric is built to surface this: it compares your stated confidence on entries to the realized hit rate over time. If you say 70% and hit 55%, you are systematically overconfident, independent of whether you are up or down this month.