Is a 35% win rate bad?

On its own it is neither good nor bad. Over the sample measured to 20 August 2026 it came with an average winner of 4.11% against an average loser of 1.38%, which is a positive expectancy. The same win rate with those figures reversed would be badly unprofitable.

What is expectancy in trading?

Expectancy is the average result you should expect per trade: (win rate x average win) minus (loss rate x average loss). It is the single figure that decides whether a strategy is worth running, and it is the one almost nobody advertises.

Why does BlofinX publish an unflattering win rate?

Because a number you can check is worth more than a number designed to impress. A published record with a sample size and a confidence interval can be audited by a reader; a bare percentage cannot.

Does a positive expectancy mean the strategy will make money?

No. It means the measured sample had a positive average result. A 50-day sample with a wide confidence interval is evidence, not a forecast, and past results do not carry any guarantee about future ones.

Why a 35% Win Rate Can Still Make Money

Almost every crypto signal service advertises a win rate, and almost none of them publish anything else. That is not an accident: win rate is the one statistic you can make look good without making any money. This guide works through what the number actually tells you, using our own — which is 35.29%, and which we publish for exactly that reason.

Expectancy calculatorEnter your own win rate and average win and loss, and see the break-even bar the payoff sets.

The number on its own is not a claim about profit

A strategy that wins 35% of the time and a strategy that wins 70% of the time cannot be compared until you know what a win and a loss are each worth. If your winners return 4% and your losers cost 1.4%, a 35% hit rate is profitable. If your winners return 1% and your losers cost 4%, a 70% hit rate is not. The win rate is one of three inputs, and it is the only one most services will show you.

Expectancy = (Win% x Average_Win) - (Loss% x Average_Loss)

// Worked with BlofinX's measured figures, 20 Aug 2026, n=85 decisive:
// (0.3529 x 4.114%) - (0.6471 x 1.383%)
// = 1.452% - 0.895%
// = +0.58% expected per forecast

Our measured figures, as at 20 August 2026

Across 121 published forecasts, 112 had resolved: 30 wins, 55 losses, and 27 that expired without reaching either their target or their invalidation level. Of the 85 that resolved decisively, 35.29% were wins. The average winner returned 4.11% and the average loser cost 1.38%, giving a realised reward-to-risk of 2.97 and a profit factor of 1.65. The largest peak-to-trough drawdown over the period was 32.93%. Every one of those numbers is live at /track-record, and the current values there supersede the ones printed here.

Why the losers matter more than the winners

The reason a low win rate can work is entirely in the size of the losses. Our average loss is roughly a third of our average win, which is a consequence of publishing an invalidation level with every forecast and treating it as a decision rather than a suggestion. Remove that discipline and the same 35% hit rate becomes a losing strategy immediately — the arithmetic is unforgiving in both directions.

The expired forecasts are part of the record

27 of our 112 resolved forecasts reached neither target nor invalidation before their horizon elapsed. They are neither wins nor losses, and a service that quietly drops them from the denominator can inflate a headline win rate substantially. We report them separately and compute the win rate over decisive outcomes only, which is the more conservative of the two conventions.

What the sample size does and does not support

50 days and 85 decisive outcomes is a moderate sample, not a strong one. The 95% confidence interval around that 35.29% runs from roughly 26.0% to 45.9%. That is a wide band, and it means the honest statement is 'somewhere in that range', not the point estimate. Any service quoting a win rate without a sample size and an interval is quoting a number that cannot be checked.

How to apply this to anyone else's numbers

Ask three questions of any published record. Over how many resolved trades? What were the average win and average loss? Were undecided trades included or dropped? If a service cannot answer all three, the win rate they are advertising is not evidence of anything. Our answers are 85, 4.11% and 1.38%, and expired trades are excluded from the rate and reported separately.

Summary

Win rate is the most quoted and least informative statistic in retail trading. Read it beside average win, average loss and sample size, or do not read it at all. Ours is 35.29% over 85 decisive forecasts as at 20 August 2026 — modest, published, and checkable at /track-record.

What a 35% Win Rate Actually Means (And Why Ours Is One)