Homeใ€€>ใ€€EA & MT5 Knowledge Baseใ€€>ใ€€Reading Backtest Metrics

BacktestPerformance MetricsIntermediate

How to Read Backtest Performance Metrics โ€” Interpreting Report Numbers Correctly

Last updated: 2026-05-20 | Estimated reading time: 15 min

Backtest reports are filled with numbers, and it's easy to feel overwhelmed at first when trying to determine which ones actually matter. Relying solely on total profit can lead you to mistake a dangerous EA for a good one. This article explains what the key metrics mean and what healthy benchmarks look like.

Don't Judge by Total Profit Alone

The first thing you notice in a backtest report is the net profit โ€” but evaluating an EA on that alone is dangerous. A large net profit could be masking an account that was cut in half at some point, or it might simply be the result of oversized lots.

Evaluating an EA requires looking at "how much it earned" and "how much risk it took" together. It helps to read the report metrics through three lenses: profitability, risk, and stability.

Metrics for Profitability

Total Net Profit

Gross profit minus gross loss โ€” the bottom-line P&L. It's the final score, but cannot be evaluated in isolation.

Profit Factor (PF)

Gross profit รท gross loss. 1.0 = breakeven, above 1.0 = profitable. A healthy range is 1.1โ€“1.5. Above 3.0, suspect curve-fitting.

Expected Payoff

Average P&L per trade. A positive value means each trade has a positive expected value. It should be positive after accounting for costs.

Recovery Factor

Total net profit รท maximum drawdown. Shows how much was earned relative to the drawdown taken. Higher is more efficient.

Metrics for Risk

Maximal Drawdown

The largest peak-to-trough decline in account balance (shown as both % and dollar amount). This is the decline you'll need to endure in live trading.

Relative Drawdown

Drawdown expressed as a percentage of account balance. Closest to the pain you feel in live trading. A common benchmark is under 20%.

Consecutive Losses

The maximum number of consecutive losing trades. In live trading, assume the actual streak could exceed this โ€” and size your money management accordingly.

Loss from Consecutive Losses

The total loss during the worst losing streak, not just a single trade. Verify that your account can absorb this amount.

Metrics for Stability

Total Trades

A measure of statistical reliability. Fewer than 100 trades is borderline; fewer than 300 means results could easily be due to chance.

Win Rate

Percentage of winning trades. Meaningless on its own โ€” must be read alongside risk/reward ratio. A 40% win rate with an RR of 1:2 still produces a positive expected value.

Sharpe Ratio

Return efficiency relative to risk (volatility). Higher means more consistent performance. A common benchmark is around 1.0.

Equity Curve (Balance Graph)

Not a number, but arguably the most important indicator. Too smooth suggests curve-fitting; a staircase pattern indicates stability; sudden drops reveal where the risks are concentrated.

Healthy Benchmarks for Each Metric

These are healthy benchmarks for an EA based on a backtest of five years or more. If the numbers look too good, treat that as a red flag for over-optimization.

MetricHealthy RangeWarning Sign
Profit Factor1.1โ€“2.0Above 3.0 (suspect curve-fitting)
Relative Drawdown10โ€“25%Above 40% (excessive risk)
Recovery Factor2.0 or aboveBelow 1.0 (poor efficiency)
Total Trades100 or moreBelow 50 (insufficient reliability)
Sharpe Ratio0.5 or aboveNegative (return does not justify the risk)
Don't pass or fail an EA on a single metric โ€” combine multiple metrics and make an overall judgment. And even when backtest numbers look good, always confirm that results can be replicated through walk-forward analysis and forward testing.

๐Ÿ”ฌ Verify Whether the Numbers Are Real

Good backtest metrics mean nothing if the EA is over-optimized โ€” performance won't hold up in live trading. Use walk-forward analysis to verify whether the edge is genuine.

Read About Walk-Forward Analysis โ†’

Frequently Asked Questions

Q: What profit factor is good enough?

A range of 1.1โ€“2.0 is healthy for a backtest spanning five years or more. Below 1.0 is immediately disqualifying (negative expected value), but conversely, anything above 3.0 should raise strong suspicion of curve-fitting. Genuine edges tend to show modest numbers.

Q: Is a higher win rate always better?

No. Win rate is meaningless on its own. A 40% win rate where winning trades are twice the size of losing ones (RR 1:2) still has a positive expected value. Conversely, a 90% win rate can produce a net loss if the occasional losing trade is enormous. Always pair win rate with risk/reward ratio.

Q: How much drawdown is acceptable?

A relative drawdown of 10โ€“25% is a common benchmark. EAs exceeding 40% become psychologically and financially difficult to endure in live trading. Ask yourself honestly whether you could stay rational through a drawdown of that magnitude.

Q: How many trades are needed for reliability?

At least 100 trades, ideally 300 or more. With fewer trades, good results are more likely to be due to chance. For lower-frequency EAs such as H4 or D1 timeframes, use a longer backtest period to accumulate sufficient trade count.

Q: What is the recovery factor?

Total net profit divided by maximum drawdown โ€” it shows how efficiently the EA earned relative to the risk it took. A value of 2.0 or above is desirable; below 1.0 means the drawdown is disproportionately large relative to the profit generated.