A backtest showing an 85% win rate can produce a losing strategy in live markets. A backtest with a 35% win rate can generate strong, consistent returns over several years. The difference comes down to five metrics that most traders fail to analyze correctly when validating a strategy. This guide explains each metric with its formula, concrete reference thresholds, and worked examples.
Why win rate alone tells you nothing
Win rate is the first metric most traders look at — and almost always the wrong starting point. An 80% win rate with a 0.2:1 reward-to-risk ratio produces a negative expected value.
Example: (0.80 x $20) minus (0.20 x $100) = $16 minus $20 = -$4 per trade.
The strategy loses on average $4 per trade, yet the win rate looks excellent. This is why every metric in this guide must be read in combination with the others. For a full walkthrough on generating these numbers from your own strategies, see our guide on how to backtest a trading strategy.
1. Expectancy: the foundational metric
Expectancy measures the average gain per unit of risk engaged. It is the metric most directly linked to long-term profitability.
Formula:
Expectancy = (Win Rate x Average Win) - (Loss Rate x Average Loss)
Worked example:
- Win rate: 45%
- Average win per trade: $300
- Loss rate: 55%
- Average loss per trade: $180
Expectancy = (0.45 x $300) - (0.55 x $180) = $135 - $99 = $36 per trade
Using R (risk multiple) as the unit: if you risk $100 per trade, $36 per trade = 0.36 R expectancy. Thresholds below are expressed in R.
| Feature | Backtrex |
|---|
Expectancy and prop firm targets
If your expectancy is 0.3 R per trade at 3 trades per day, the projected expectancy over 20 trading days is 0.3 x 3 x 20 = 18 R. At 1% risk per trade on a $10,000 account, that represents an 18% expected profit over the challenge period — comfortably above FTMO's 10% target. This is a statistical projection; individual results vary. See our guide on backtesting with prop firm rules to simulate these constraints precisely.
2. Profit factor: the quick decision rule
Profit factor is the ratio of total gross gains to total gross losses across the entire backtest. It gives a fast, single-number signal that is easy to compare across strategies.
Formula:
Profit Factor = Total gross gains / Total gross losses (absolute value)
Example: 80 trades, gross gains $24,000, gross losses $13,000 Profit Factor = 24,000 / 13,000 = 1.85
| Feature | Backtrex |
|---|
Profit factor needs enough trades
A high profit factor on 15 trades carries no statistical weight. Variance is too high to distinguish a genuine edge from a lucky run. The minimum recommended threshold is 100 trades across representative market conditions — covering trending, ranging, and high-volatility phases. FX Replay lists this as the industry standard minimum for evaluating backtest KPIs.
3. Sharpe ratio: risk-adjusted return
The Sharpe ratio measures excess return per unit of risk. It allows you to compare two strategies with similar returns but very different volatility profiles.
Annualized formula:
Sharpe Ratio = (Average trade return / Standard deviation of returns) x square root of N
Where N = 252 (trading days in a year).
| Feature | Backtrex |
|---|
Most backtesting platforms calculate the Sharpe ratio automatically. With Backtrex, it appears directly in every backtest report generated.
4. Maximum drawdown: the survival metric
Maximum drawdown (MDD) measures the largest peak-to-trough decline in account equity across the full tested period. It is the survival metric: it determines whether your strategy can outlast its losing sequences long enough to reach its statistical expectancy.
Formula:
MDD = (Trough value - Peak value preceding it) / Peak value x 100
Practical thresholds by strategy type:
| Feature | Backtrex |
|---|
Static drawdown vs trailing drawdown
Most futures prop firms — including Apex Trader Funding and Topstep — use trailing drawdown, not static drawdown. Your backtest MDD should be calculated on the same basis to be comparable to the challenge rules. Our guide on trailing drawdown in prop firms covers how to simulate this constraint correctly in your backtest.
5. Win rate and R:R: read them together
Win rate and the reward-to-risk ratio (R:R) are only meaningful when read as a pair. The table below shows the minimum R:R required at each win rate level to maintain a positive expectancy:
| Feature | Backtrex |
|---|
Typical ICT/SMC strategies run at 40-55% win rate with a 1:2 to 1:3 R:R. Classic trend-following strategies run at 35-45% win rate with a 1:3 to 1:5 R:R. Both profiles can produce the same positive expectancy — this is mathematically expected and perfectly normal. For a systematic approach to backtesting SMC strategies, see our guide on Smart Money Concepts trading.
Reading all 5 metrics together
No single metric should be read in isolation. A genuinely robust backtest meets all of these thresholds simultaneously:
If one metric is outside its threshold, that is not an immediate disqualification — but it is a signal to investigate before going live.
Checklist: robust backtest or overfit?
An overfit backtest shows flattering metrics on training data, then collapses in real conditions. The warning signs to watch for:
- Profit factor above 3.0 on fewer than 100 trades: probable overfitting
- Win rate above 70% on scalping: suspect, spreads and slippage will reduce it significantly
- Max drawdown below 2% over 2 years: too clean to be realistic
- No extended drawdown periods anywhere in the history: the backtest avoided difficult market phases
To address each of these traps systematically, see our article on common backtesting mistakes.
Important Risk Warning
Conclusion
Evaluating a backtest is not about looking at total profit or win rate. Expectancy, profit factor, Sharpe ratio, maximum drawdown, and win rate/R:R coherence form a validation system that separates a genuinely robust strategy from an optimization artifact. If you want a tool that automatically calculates these metrics across 5-10 years of historical data, Backtrex generates all of them in every backtest report — no coding required.
Expectancy is the average gain per trade expressed in dollars or risk multiples (R). Formula: (Win Rate x Average Win) - (Loss Rate x Average Loss). A positive expectancy means the strategy is mathematically profitable over a large sample of trades. A negative expectancy means it is a losing strategy regardless of the win rate displayed. It must be calculated on at least 100 trades to carry statistical weight.
A profit factor of 1.5 is the minimum recommended threshold for a strategy traded with real-world costs (spreads, slippage, commissions). Below 1.2, the strategy is too sensitive to costs to survive live trading. Above 2.0, the strategy is considered robust. A profit factor above 3.0 on a small sample of trades is often a sign of overfitting rather than a genuine edge.
Annualized Sharpe ratio = (Average trade return / Standard deviation of returns) x square root of 252. Most backtesting platforms calculate this automatically. A Sharpe ratio above 1.0 is considered good for retail trading; above 2.0 is exceptional and rare in real market conditions. A Sharpe below 0.5 suggests the risk taken is not justified by the returns generated.
FTMO imposes a 10% drawdown limit calculated from the initial account balance. In practice, your in-sample backtest should not exceed 5-6% MDD — because live drawdown is almost always higher than backtest drawdown, and you need a safety margin for unseen market conditions. If your backtest MDD is already at 8-9%, the strategy is likely too risky to pass the challenge.
No. An 80% win rate with a 0.2:1 R:R produces a negative expectancy: (0.80 x 0.2) - (0.20 x 1) = 0.16 - 0.20 = -0.04 R per trade. The strategy loses on average 4% of your risk each trade. Win rate should never be read alone — it must always be combined with the average R:R to calculate the real expectancy.
The minimum recommended is 100 trades. Below this threshold, variance is too high to distinguish a genuine edge from a lucky run. For low-frequency strategies like swing trading, 100 trades may represent 1-3 years of data — which is precisely why backtesting on long historical datasets (5-10 years) is important for reliable results.
Profit factor measures the overall ratio of gross gains to gross losses across all trades. Expectancy measures the average gain per trade, factoring in win rate and average R:R. They are complementary: a profit factor of 2.0 on 10 trades is unreliable due to high variance, while an expectancy of 0.4 R calculated on 200 trades carries real statistical weight. Use both together to validate your backtest statistically.