How to read metrics (the right order)
Most traders look at profit first. That’s backwards. A better order:- Trade count & exposure (is the sample meaningful?)
- Drawdown (can you survive it?)
- Profit factor / expectancy (is it edge or noise?)
- Sharpe/Sortino (quality of returns)
- CAGR / total return (scale of returns)
Metrics are only meaningful together. One number rarely tells the truth.
Trade activity metrics
Trade count
How many trades the strategy took in the selected period. Why it matters:- very low trade count = high uncertainty
- very high trade count = noise sensitivity
Win rate
Percent of trades that are profitable. Interpretation:- high win rate can still lose money (small wins, big losses)
- lower win rate can still be profitable (big wins, controlled losses)
Exposure time
Percent of time the strategy is in a position. Interpretation:- high exposure = higher market risk
- low exposure = fewer opportunities, often more selective
Profitability metrics
Total return / Total PnL
The total profit/loss over the backtest period. Interpretation:- depends heavily on position sizing assumptions
- compare alongside drawdowns
Profit factor
Gross profit divided by gross loss. Rules of thumb:- < 1.0 = losing strategy
- ~1.1–1.3 = weak edge (often fragile)
- 1.5+ = stronger edge (still needs validation)
Expectancy
Average profit per trade (often expressed in R or %). Interpretation:- tells you if trades are positive on average
- can be positive even with low win rate
Risk metrics
Max drawdown
The largest peak-to-trough decline in equity during the backtest. Interpretation:- the “pain” you must tolerate
- compare to total return and exposure
Drawdown duration
How long the strategy stayed below its previous equity peak. Interpretation:- long durations can be psychologically hard
- often indicates regime dependence
Largest loss / worst trade
The biggest single losing trade. Interpretation:- helps detect tail risk
- check if it aligns with your stop loss design
Return-quality metrics
Sharpe ratio
Return adjusted by volatility. Interpretation:- higher is better
- sensitive to outliers and assumptions
Sortino ratio
Like Sharpe, but penalizes only downside volatility. Interpretation:- often more relevant for strategies with asymmetric returns
Calmar ratio
CAGR divided by max drawdown. Interpretation:- balances return against worst drawdown
- useful for comparing strategies with different risk profiles
Growth metrics
CAGR
Compound annual growth rate. Interpretation:- assumes reinvestment and stable behavior
- can be misleading on short backtests
Trade-level metrics
Average trade
Average profit/loss per trade.Average win / average loss
Mean size of winning vs losing trades. Interpretation:- helps explain why win rate is (or isn’t) profitable
- a healthy profile often has controlled losses and meaningful wins
Avg / max trade duration
How long trades last. Interpretation:- affects opportunity cost and exposure
- impacts suitability for your style (fast vs slow systems)
Strategy quality metrics (advanced)
SQN (System Quality Number)
A measure that combines trade expectancy and variability. Interpretation:- higher values imply more consistent outcomes
- unreliable with low trade counts
Kelly criterion
A sizing heuristic based on edge and variance. Interpretation:- can be unstable
- useful as a signal of edge strength, not as a literal sizing rule
Common misreads
High win rate but losing total return
High win rate but losing total return
Usually caused by small wins and occasional large losses (poor risk control).
Great CAGR on a short window
Great CAGR on a short window
Often overfit or regime-specific. Extend the backtest range.
Profit factor looks good but drawdown is huge
Profit factor looks good but drawdown is huge
Edge may exist, but risk is likely unacceptable.
Sharpe/Sortino look bad but return is positive
Sharpe/Sortino look bad but return is positive
Best practices
- Compare strategies using the same backtest range
- Start by validating trade history (not metrics)
- Prefer robustness over maximal returns
- Always measure risk first
Trade history & logs
Metrics are summaries - trades are the truth.
What to read next
How backtesting works
Understand the simulation model.
Data coverage
Ensure your backtest window is valid.
Common pitfalls
Avoid misinterpretation traps.
Improving a strategy
Iterate without overfitting.
Metrics are compressed reality.
Always verify them against the trade list.
Always verify them against the trade list.