Why we don't publish a win rate

Every crypto signal channel leads with a win rate. Ninety percent. Eighty-five percent. Pick a number. We don't publish one at all, and the reason is not modesty — it's that the number is almost impossible to state honestly.

A win rate can be engineered without improving anything. Move the take-profit close enough to the entry and almost every trade eventually touches it. You will have a spectacular win rate and a losing system, because the rare loss is now many times larger than the common win. The number went up. The account went down.

So we report expectancy in R. R is the amount risked on a trade — the distance between the entry and the stop-loss. A result of +1R means the trade made exactly what it put at risk; -1R means it lost it. Expectancy is the average R across every trade, winners and losers together. Unlike win rate, it cannot be improved by moving the target closer, because moving the target closer also shrinks the reward.

It is a less flattering number, and that is the point. A win rate tells you how often you were right. Expectancy tells you whether being right that often was worth anything.

Three rules decide how our outcomes are counted.

The first is first-touch. A trade is resolved by whichever level the price reaches first — the stop or the target — not by whichever produces the nicer result at the end of the period.

The second is that ties go to the stop. When a single candle covers both the target and the stop, there is no way to know from that candle which was touched first. We record the loss. Resolving it the other way would quietly inflate every result on the record, and the inflation would be invisible.

The third is that costs count. A round-trip exchange fee is charged against every simulated outcome. This matters most exactly where it is easiest to forget: with a tight stop, the distance you risked shrinks while the fee does not, so the fee eats a larger share of every R. A system that looks marginally profitable before fees is frequently unprofitable after them.

We also separate what is validated from what is research. Setups outside the validated cohort are marked SHADOW and tracked on the same record as everything else. They are not hidden, and they are not dressed up as trades to take. If a research idea turns out badly, it stays on the record turning out badly.

None of this makes a signal correct. It makes the reporting of it honest, which is a different and smaller claim — but it is a claim you can check, and that is the only kind worth making.

If a channel only ever shows you green, the question is not whether they are lucky. It is what happened to the red.