Why a 90% Win Rate Is Misleading

What a 90% win rate actually measures — and what it hides

A claimed 90% win rate means that nine out of every ten trades closed in profit. That is all it means. It says nothing about how large those profits were, how large the losses were when they occurred, whether the sample was large enough to be statistically meaningful, or whether trading costs were included in the calculation. The number is presented as though it answers the question 'Is this signal strategy good?' It does not answer that question — it answers only the narrower and far less useful question 'How often did trades close positive?'

The reason win rate dominates signal marketing is that most people interpret high percentages as strong evidence of quality. If nine out of ten trades win, surely the system is profitable? This intuition breaks down the moment you ask what the losses look like. A strategy that wins $10 on nine trades and loses $1,000 on the tenth has a misleading 90% win rate and a cumulative loss of $910. The win rate tells you one thing; the economics of the strategy tell you another.

Expectancy — the average amount a strategy gains or loses per trade — is what the win rate headline leaves out entirely. A single high percentage, stripped of context about wins, losses, sample size, and costs, is a marketing figure, not an analytical one.

How providers engineer a 90% win rate through trade structure

The most common method for manufacturing a very high win rate is structural: set take-profit targets that are a fraction of the size of the stop-loss, or operate with no meaningful stop-loss at all. The math is straightforward. If a strategy sets a take-profit target at 0.5% and a stop-loss at 5%, price only needs to move 0.5% in the right direction before the trade closes as a winner — a much smaller barrier to clear than the 5% move needed to hit the stop. In trending or even moderately directional markets, this structure will produce a high win rate by letting trades close small and quickly on any favourable tick.

The arithmetic reveals the problem. To break even at a reward-to-risk ratio of 0.1:1 — where the target is one-tenth the size of the stop — a strategy needs a win rate above 91% just to avoid losing money over time, before fees. At a 0.05:1 ratio, the required win rate exceeds 95%. These are mathematical constraints, not rules of thumb. Any strategy reporting a 90% win rate at very tight targets is operating at or near break-even on the trade structure alone — and a single streak of losses, or a brief period where the market moves against the trend before recovering, wipes out many small wins at once.

The alternative structure — large or absent stop-losses — achieves the same outcome by a different route. Without a defined stop-loss, losing trades are never formally closed; they sit open until the market either recovers (claiming a win) or causes such a large drawdown that the position is force-closed or the account is margined out. All interim losses are invisible to the win-rate calculation because the trade has not technically closed. This is not a technicality — it is the mechanism that allows some providers to claim extraordinary win rates on positions that are quietly sitting in deep drawdown.

Cherry-picked windows, ignored costs, and martingale strategies

Three further techniques inflate headline win rates beyond the structural mechanics above. The first is the cherry-picked time window. Crypto markets spend extended periods trending — in those periods, almost any strategy that bets on continuation will have an elevated win rate. A provider who measures their results only during the strongest trending phase of a bull market will show a win rate that says very little about what to expect across a full market cycle including consolidations, false breakouts, and reversals.

The second technique is ignoring fees and slippage. Exchange fees typically run 0.02% to 0.10% per trade, with round-trip costs of 0.15% to 0.20% for a single in-and-out. On a strategy targeting 0.5% gains, this fee load consumes 30–40% of the gross profit per trade. A win rate that is marginally positive on a gross basis becomes marginally negative on a net basis once realistic transaction costs are included. Providers rarely present win rates on a fee-adjusted net basis.

The third technique is a form of position sizing known as martingale: doubling the size of a trade after each loss, so that a single winner recoups all prior losses and records a net profit. This structure produces a very high win rate in the short term — you are always eventually recording a win before counting the trade — but at the cost of position sizes that grow exponentially through losing streaks. A run of six consecutive losses requires the seventh position to be 64 times the size of the first just to break even. The win rate number looks favourable until the account runs out of capacity to double further.

Why one large loss erases many small wins: the arithmetic of asymmetry

There is a mathematical reason that oversized losses are particularly corrosive beyond the simple arithmetic of summing gains and losses. Losses and gains are not symmetric in their effect on capital.

An illustrative example: starting with $10,000, a 10% loss brings the account to $9,000. To recover from $9,000 back to $10,000 requires an 11.1% gain — slightly more than the loss that caused the drawdown. A 20% loss from $10,000 leaves $8,000; recovering requires a 25% gain. A 50% loss requires a 100% gain just to break even. Each time capital falls, the percentage gain required to recover grows faster than the loss that caused the problem.

In the context of a 90% win rate strategy with a large stop-loss or no stop at all, this asymmetry means that a run of four or five consecutive losses — a normal and expected occurrence in any trading system over a large enough sample — can generate a drawdown that requires a disproportionately long winning streak to recover from. The high win rate does not eliminate the asymmetry; it defers the reckoning while the losses accumulate unrealised.

Expectancy: the honest measure of an edge

The metric that resolves all of the above is expectancy: the average amount a strategy gains or loses per trade over a large sample. The formula is: Expectancy = (Win Rate × Average Win) − (Loss Rate × Average Loss).

Two illustrative examples show why win rate alone misleads. Provider A claims a 90% win rate. The average winner is $50 (small take-profit), and when a loss does occur, the average loss is $500 (large stop or no stop). Expectancy = (0.90 × $50) − (0.10 × $500) = $45 − $50 = −$5 per trade. Despite winning nine out of ten, the strategy loses money on average. Over 100 trades, the expected loss is $500 — before fees.

Provider B reports only a 50% win rate — a figure that might cause a beginner to dismiss it, yet this is not automatically a losing rate. The average winner is $200, and the average loss is $80. Expectancy = (0.50 × $200) − (0.50 × $80) = $100 − $40 = +$60 per trade. Despite winning only half its trades, this strategy has a positive expectancy of $60 per trade. Over 100 trades, the expected gain is $6,000 — before fees. The win rate tells you nothing useful until you know the average size of wins and losses alongside it.

What realistic win rates look like across different trading styles

Across professional trading literature and practitioner discussion, profitable discretionary traders typically operate in the 40–60% win rate range, not at the 90%+ levels claimed in most signal marketing. Systematic trend-following strategies — those that stay in the direction of momentum and accept many small losses while waiting for large trending moves — often post win rates below 50% and remain profitable over time because their winners are significantly larger than their losers. Mean-reversion or range-trading strategies can achieve higher win rates in the 60–70% range, but individual wins are typically smaller relative to the losses that occur when the range breaks.

For crypto signal providers specifically, a believable and sustainable win rate — one consistent with honest risk-reward structures across diverse market conditions — falls approximately in the 50–70% range. Any claimed rate above 75% across a sample of 100 or more trades spanning multiple market phases warrants substantial scrutiny. The question to ask is always: at what average reward-to-risk ratio, and over what type of market conditions, was this rate measured?

This does not mean all providers claiming higher rates are dishonest — an unusually strong trending period can produce elevated rates for any momentum strategy for a limited time. But what is structurally implausible is a 90%+ win rate sustained across a large sample of trades in varying market conditions at any reward-to-risk ratio that would produce positive expectancy. The mathematics do not support it. Only risk capital you can afford to lose when evaluating any signal strategy — regardless of the win rate advertised — and weight claimed win rates accordingly.

How to pressure-test any win-rate claim

When you encounter a 90% win rate claim, the correct response is to treat it as the beginning of an inquiry, not a conclusion. A credible provider can answer several specific questions immediately and completely.

Ask for the full trade record, not a summary or screenshot. The record should include every trade over the stated period — wins and losses — with the entry price, exit price, entry timestamp, exit timestamp, the trading pair, and whether the result includes exchange fees. A provider who can supply this has a verifiable record. A provider who responds with a screenshot of winners, a curated selection of highlighted trades, or a summary statistic without an underlying dataset does not have a verifiable record.

Ask what the average reward-to-risk ratio was across the record. A 90% win rate at a 0.1:1 ratio is a losing strategy. A 60% win rate at a 1:2 ratio is a strong one. These numbers belong together. Ask for the sample size and the date range, and ask whether results were live (placed in real time with verifiable timestamps) or backtested. Backtested results routinely overstate live performance due to overfitting, look-ahead bias, and optimistic fill assumptions. The combination of verifiable timestamps, complete win-and-loss record, stated R:R, and net-of-fees reporting is what separates a serious provider from a marketing claim.