The profit factor is the ratio between total gross gains and total gross losses in a trading strategy: a ratio above 1.5 is the minimum recommended threshold before any live deployment. Yet this single figure is not enough. A backtesting report contains roughly ten interdependent metrics: profit factor, expectancy, maximum drawdown, Sharpe ratio, win rate, recovery factor. Knowing how to read them together, and in the right order, is the skill that separates a trader publishing flattering results from one who deploys a robust strategy. This guide walks through each category of metrics, explains their meaning, and identifies the most common interpretation mistakes.
Why Backtesting Metrics Matter
The Risk of Over-Interpretation
According to ESMA, between 74% and 89% of retail accounts lose money trading CFDs and Forex. One of the structural causes is deploying strategies validated on insufficient criteria: an attractive win rate, a positive net profit, and nothing else. Over-interpreting two or three favorable metrics while ignoring drawdown or expectancy leads to inevitable losses as soon as the market changes regime.
A backtesting report is not a profitability certificate. It is an analytical tool that requires critical reading across all its indicators. For a deeper look at common errors, check our guide on backtesting mistakes to avoid.
What to Look at First
The recommended reading order follows a filtering logic:
- Profit factor: is the strategy globally profitable?
- Maximum drawdown: is the risk compatible with your capital and psychology?
- Expectancy: does each trade generate a positive expected gain?
- Sharpe ratio: is the performance proportional to the risk taken?
- Win rate and consecutive losses: is consistency sufficient to hold in live conditions?
If one metric fails this filter, the others do not matter. There is no point analyzing in depth a strategy whose maximum drawdown exceeds 40% of capital, regardless of how smooth its equity curve looks.
Automatic metric calculation
Backtrex calculates all these metrics in real time during backtesting, without any export to Excel. Warning signals (excessive drawdown, low profit factor, potential overfitting) are highlighted directly in the report. Explore all features on the features page.
Profitability Metrics
Profit Factor: Definition and Minimum Threshold
The profit factor is the ratio between the sum of all gross gains and the sum of all gross losses:
Profit factor = Total gross gains / Total gross losses
A profit factor below 1.0 means the strategy loses more than it earns. The reference thresholds used in quantitative finance are:
| Profit factor | Interpretation | Recommendation |
|---|---|---|
| < 1.0 | Losing strategy | Reject |
| 1.0 - 1.3 | Marginally positive | Insufficient for live trading |
| 1.3 - 1.5 | Acceptable with caution | Validate on more data |
| 1.5 - 2.0 | Good strategy | Recommended threshold for deployment |
| > 2.0 | Excellent, but suspect overfitting | Check out-of-sample validation |
A profit factor above 2.0 on historical data should raise suspicion: it is often a sign of overfitting, meaning the strategy has been over-fitted to the past. For a deeper analysis of this metric, see our article on expectancy, profit factor and Sharpe ratio.
Expectancy: Average Gain per Trade
Expectancy measures the average expected gain for each unit of risk committed. It is the metric most directly tied to long-term profitability.
Expectancy = (Win rate x Average gain per trade) - (Loss rate x Average loss per trade)
A positive expectancy confirms that the strategy generates value over a large number of trades. A negative expectancy means the strategy will destroy capital, even with an apparent win rate of 70%.
Concrete example: a strategy with 40% win rate, average gain of $250 and average loss of $120 gives an expectancy of (0.40 x 250) - (0.60 x 120) = 100 - 72 = $28 per trade. Each trade returns on average $28, regardless of its individual outcome.
Net Profit vs Gross Profit
Gross profit is the total gains before deducting commissions and slippage. Net profit is what remains after these costs. The gap between the two reveals the real cost of the strategy and its impact on profitability.
A high-frequency scalping strategy may show an attractive gross profit and a slightly negative net profit after fees. This metric is particularly important for strategies with a high trade count (several hundred over the tested period) where brokerage fees accumulate quickly.
Always use net profit
Comparing strategies solely on gross profit distorts the comparison. Net profit, which incorporates real commissions and estimated slippage, is the only relevant profitability measure for an honest evaluation.
Risk Metrics
Maximum Drawdown: Absolute and Relative
Maximum drawdown (MDD) measures the largest loss suffered from a capital peak to the subsequent trough. It is the most directly readable risk metric for a trader.
Absolute drawdown: expressed in monetary value (e.g., -$1,200 drop on $10,000 capital). Relative drawdown: expressed as a percentage of peak capital (e.g., -12%).
Prop firms such as FTMO impose strict drawdown limits (typically 5% daily, 10% maximum). A backtest whose maximum drawdown exceeds these thresholds means the strategy would have triggered a cut during the tested period. See our guide on backtesting prop firm rules for practical implications.
A maximum drawdown below 15% over multi-year data is a sign of robustness. Above 25%, the strategy becomes psychologically difficult to sustain in live conditions, even if profitability metrics are positive.
Sharpe Ratio and Sortino Ratio
The Sharpe ratio measures excess return (above the risk-free rate) per unit of total volatility. According to Investopedia, the reference thresholds are: a Sharpe below 1.0 is substandard, between 1.0 and 2.0 is acceptable, and above 2.0 is considered very good.
Sharpe ratio = (Average return - Risk-free rate) / Standard deviation of returns
The limitation of the Sharpe ratio is that it penalizes upside volatility as much as downside volatility. For directional strategies whose profits are asymmetric, the Sortino ratio is more relevant.
Sortino ratio = (Average return - Risk-free rate) / Standard deviation of negative returns
The Sortino only penalizes the volatility of losses. A strategy that generates large occasional gains (such as trend following) will be better valued by the Sortino than by the Sharpe. For a detailed comparison, see our guide on backtesting performance ratios: Sharpe, Sortino and Calmar.
Calmar Ratio: Risk-Adjusted Return
The Calmar ratio compares annualized return to the maximum absolute drawdown. It is the preferred metric of hedge fund managers for evaluating risk efficiency over a long period.
Calmar ratio = Annualized return / Maximum drawdown
A Calmar ratio above 1.0 means the strategy generated at least as much annual return as its worst loss episode. A ratio above 3.0 is considered excellent in the quantitative industry.
Consistency Metrics
Win Rate and Why It Is Not Enough
Win rate is the percentage of winning trades. It is the most often cited metric and the most often misinterpreted. A win rate of 80% with a gain/loss ratio of 0.2 produces a negative expectancy. A win rate of 35% with a ratio of 3.5 produces a very positive expectancy.
Win rate must always be read in combination with the average gain to average loss ratio. These two values together define the real expectancy of the strategy.
Consecutive Losses: The Resilience Test
The maximum number of consecutive losses is as much a psychological and capital indicator as a statistical one. If your backtest shows a maximum streak of 12 consecutive losses, you must be able to hold that sequence in live trading without changing your sizing or abandoning the strategy.
This metric is particularly important for traders managing funded accounts with daily drawdown rules. A run of consecutive losses over a short period can exhaust the daily drawdown before the strategy has had time to recover.
Calculating consecutive loss probability
To estimate the probability of a streak of N consecutive losses, use the formula (1 - win rate)^N. With a win rate of 55% and N = 8: (0.45)^8 = 0.17%. That seems low, but over 5,000 trades, this sequence will statistically occur several times.
Recovery Factor
The recovery factor measures the strategy's capacity to recover its losses. It is calculated as the ratio between net profit and the maximum absolute drawdown.
Recovery factor = Net profit / Maximum absolute drawdown
A recovery factor above 3.0 indicates that the strategy generates at least three times its maximum losses in net gains. A recovery factor below 1.0 means cumulative losses over the worst episodes exceed total gains.
Overall Interpretation and Common Traps
Overfitting: When Numbers Lie
Overfitting is the phenomenon whereby a strategy is excessively fitted to historical data and fails to generalize to new data. A backtesting report subject to overfitting displays remarkable metrics (profit factor > 2.5, drawdown < 5%, Sharpe > 3.0) that collapse as soon as the strategy goes live.
Warning signs of overfitting in a backtest report:
- Very high profit factor over a short period (less than 2 years of data)
- Low trade count (fewer than 100 trades over the period): statistics are not significant
- Metrics that improve systematically with each parameter adjustment
- Lack of robustness across sub-periods (the strategy performs on 2022-2023 but fails on 2020-2021)
For more detail, see our complete guide on how to backtest without overfitting.
Out-of-Sample Validation
Out-of-sample (OOS) validation is the most important test for evaluating a strategy's robustness. The principle: calibrate parameters on 70% of the data (in-sample), then verify that the metrics remain consistent on the remaining 30% (out-of-sample) without recalibration.
A robust strategy sees its metrics slightly degraded on the OOS period (slightly lower profit factor, slightly higher drawdown) but remains profitable. An over-optimized strategy sees its metrics collapse in OOS.
Backtesting vs Live Trading
A backtest produces results under ideal conditions: no unexpected slippage, no execution error, no emotion. Live trading introduces friction that backtesting does not always capture fully. Understanding these differences and how to anticipate them is essential before deployment.
Build in a sufficient sample: test over at least 3 to 5 years of data and across different market regimes (trends, ranges, crises). A strategy that only performs in a strong trending environment is not robust.
Important Risk Warning
Conclusion
Reading a backtesting report correctly means analyzing an interdependent system of metrics. Profit factor confirms overall profitability, expectancy validates each trade individually, maximum drawdown defines the actual risk level, and adjusted ratios (Sharpe, Sortino, Calmar) allow comparing strategies with different profiles. All these metrics are calculated automatically by Backtrex, with built-in overfitting alerts, without any Excel export. Check the pricing page to start backtesting.
A profit factor of 1.5 is the minimum recommended threshold before any live deployment. Below 1.3, the strategy is too fragile to absorb live fees and slippage. Above 2.0, systematically check for overfitting by testing the strategy on an out-of-sample period not used during parameter calibration.
Expectancy is calculated with the formula: (Win rate x Average gain per trade) - (Loss rate x Average loss per trade). A positive expectancy means that each trade generates an average gain, regardless of its individual outcome. An expectancy of $20 over 500 trades represents $10,000 in expected profits, before fees.
The Sharpe ratio penalizes all volatility, whether upside or downside. The Sortino ratio only penalizes negative volatility (losses). For directional strategies with asymmetric gains, such as trend following or SMC strategies, the Sortino is more representative of the actual risk/return quality.
A maximum drawdown below 15% on multi-year data is generally considered robust. For prop firms (FTMO, My Forex Funds), the threshold is often set at 10% maximum on the account. A drawdown above 25% makes the strategy psychologically difficult to sustain in live trading, even if profitability metrics are positive.
Warning signs include: a profit factor above 2.5 over a short period (less than 2 years), fewer than 100 trades over the period, metrics that improve with every parameter adjustment, and performance that collapses on sub-periods not used during calibration. The solution is out-of-sample validation: reserve 30% of the data and use it only once for the final verification.
The recovery factor is the ratio between total net profit and the maximum absolute drawdown. It measures the strategy's capacity to generate gains that exceed its worst loss episodes. A recovery factor above 3.0 is considered excellent. A ratio below 1.0 means total gains have not offset the maximum losses incurred.
Backtesting statistics require a minimum of 100 trades to be statistically significant, and ideally more than 300 trades over at least 3 years of data. Below 100 trades, profit factor and expectancy can be heavily influenced by a few extreme trades and do not reflect expected long-term performance.