Crypto Signal Accuracy and Win Rates
Break even win rates, how 90 percent accuracy claims are built, losing streak odds and how many trades a record needs.
Rayner6 min read19 September 2026Crypto signal accuracy is the number most channels lead with and the one that says least. “92 percent accuracy” sounds decisive, but without the size of the average win and loss it tells you almost nothing about whether following the signals makes money. It is also the easiest number to inflate, simply by choosing what counts as a win.
This guide shows the win rate you actually need at different reward to risk ratios, the accounting choices behind very high accuracy claims, how long losing streaks get even for good providers, and how many trades it takes before a win rate means anything. The forex version is in forex signal accuracy.
Crypto signal accuracy, the share of signals that win, means little on its own. What matters is win rate combined with the average win and loss in R. A provider with 40 percent winners at 2R makes money, while one with 80 percent winners can lose it. Claims of 90 percent or more usually rely on counting partial targets as wins or removing losses.
The win rate you actually need
| Average win | Average loss | Break even win rate |
|---|---|---|
| 0.5R | 1R | 67 percent |
| 1R | 1R | 50 percent |
| 1.5R | 1R | 40 percent |
| 2R | 1R | 33 percent |
| 3R | 1R | 25 percent |
Break even win rate = average loss ÷ (average win + average loss). Fees push every figure up a little. A provider whose win rate is only a few points above break even is barely profitable once costs and slippage are included.
How 90 percent accuracy is built
| Accounting choice | Effect on the win rate |
|---|---|
| TP1 hit counts as a win, even if the rest returns to entry or the stop | Very large. TP1 is often close to entry |
| Break even exits not counted as losses | Removes many near misses from the loss column |
| Signals that never triggered excluded, including ones that ran to the stop first | Quietly removes losses |
| Stops moved after entry, then the trade counted as closed at break even | Turns losses into draws |
| Losing signals deleted | Anything is possible |
| Very wide stops with very close targets | High win rate, large average loss |
None of these is visible from a headline number. The only defence is rebuilding the record yourself, as explained in how to judge crypto signal providers and how to verify signal results.
Win rate and R together
The measure that combines both is expectancy, the average result per trade in R.
Expectancy = (win rate × average win) minus (loss rate × average loss)
| Provider | Win rate | Average win | Expectancy per trade | Over 100 trades |
|---|---|---|---|---|
| A | 85% | 0.2R | 0.17 minus 0.15 = +0.02R | +2R, less than fees |
| B | 60% | 1R | 0.60 minus 0.40 = +0.20R | +20R |
| C | 40% | 2R | 0.80 minus 0.60 = +0.20R | +20R |
| D | 70% | 0.3R | 0.21 minus 0.30 = minus 0.09R | minus 9R |
All losses are 1R in this table. Provider A has the most impressive accuracy and almost no edge. Provider D looks better than C on accuracy and loses money.
How long losing streaks get
Streaks are normal even for a profitable provider. These are the chances of at least one streak of the given length within 100 trades, assuming each trade is independent.
| Win rate | Streak length | Chance within 100 trades |
|---|---|---|
| 60% | 5 losses in a row | about 46 percent |
| 50% | 6 losses in a row | about 55 percent |
| 40% | 8 losses in a row | about 49 percent |
| 33% | 10 losses in a row | about 45 percent |
A provider trading 2R targets at 33 percent can be sound and still post ten losses in a row. At 1 percent risk per trade that is roughly a 10 percent drawdown. Size so that you can sit through it, as covered in how much to risk per signal.
How many trades a record needs
| Trades in the sample | Measured win rate could be off by, roughly |
|---|---|
| 20 | 22 percentage points |
| 50 | 14 points |
| 100 | 10 points |
| 200 | 7 points |
Figures are the approximate 95 percent margin for a win rate near 50 percent. A provider showing 60 percent over 20 trades might really be a 40 percent provider. That is why a week of good results means very little, and why 50 trades is the minimum before a record says much.
Roughly, at 95 percent confidence. Over 200 trades it narrows to about 7 points.
What to ask any provider
- 01How is a win definedDoes TP1 then back to entry count as a win, a draw or a partial?
- 02Are untriggered signals listedIncluding those that hit the stop before the entry filled?
- 03What is the average win and loss in RWithout these, the win rate is decoration.
- 04Over how many tradesFewer than 50 says little.
- 05Before or after feesFrequent signals lose more to fees than the win rate shows.
Our own signals are scored in R with staged targets, and the ones that hit their stop stay on the crypto signals page.
Keep reading
- Best crypto signal providers the one hour audit
- How to read a crypto signal how staged targets are scored
- Crypto signal scams fake results and deleted losses
- Forex signal accuracy the same maths in currencies
Frequently Asked Questions
What is a good crypto signal accuracy?
It depends on the reward to risk. With 2R targets, anything reliably above about 40 percent is good. With 0.5R targets, even 70 percent can lose money. Always look at the average win and loss alongside it.
Are 90 percent accurate crypto signals real?
Accuracy that high usually comes from counting partial targets as wins, excluding break even exits or untriggered signals, or deleting losses. Rebuild the record yourself before believing it.
How do I calculate a break even win rate?
Divide the average loss by the sum of the average win and the average loss. With 2R wins and 1R losses, that is 1 divided by 3, about 33 percent.
What is expectancy in trading?
The average result per trade: win rate times average win, minus loss rate times average loss. A positive expectancy after fees is what makes a signal service worth following.
How many trades do I need to judge a signal provider?
At least 50, and ideally 100 or more. Over 20 trades a measured win rate can be more than 20 points away from the true figure.
Is a long losing streak a sign the provider is bad?
Not necessarily. A sound provider with a 40 percent win rate has about an even chance of eight losses in a row within 100 trades. Judge on net R over a large sample.
Do fees affect signal accuracy?
They do not change the win rate, but they reduce every win and increase every loss. Providers with many small targets are hit hardest.
