Chapter 4
How to evaluate an algo trading strategy
no fantasy numbers. this is how you read a real track record.
These are the same metrics institutional allocators use to evaluate strategies. After this chapter you'll read a track record the way they do — not reacting to the headline return, but understanding what it actually cost to generate it.
Return without context is nearly meaningless. A 30% gain looks completely different depending on how much risk was taken to get there. These metrics give return its context — read them as a set, not in isolation.
Performance metrics
How do you measure an algo trading strategy's performance?
Performance metrics answer one question: how much return did the strategy generate, and how consistently? The four metrics below measure return, reliability, and the relationship between wins and losses. Read them together — any one in isolation can mislead.
CAGR Compound Annual Growth Rate
Annualized return, accounting for compounding. More meaningful than total return for comparing strategies with different track record lengths. A strategy up 50% over 3 years has a lower CAGR than one up 50% over 18 months.
Win Rate
The percentage of trades that were profitable. A high win rate is not necessarily better — it depends on the size of wins versus losses. A strategy winning 40% of trades can still be profitable if winners are significantly larger than losers.
Avg Win / Avg Loss
The average size of winning trades versus losing trades. Read alongside win rate: a 55% win rate with avg win = 2× avg loss is a strong profile. A 70% win rate with avg win = 0.5× avg loss is profitable — but barely. The ratio of win size to loss size matters as much as the win rate itself.
Profit Factor
Total gross profit divided by total gross loss. A profit factor above 1.0 means the strategy made money overall. Above 1.5 is generally considered solid; below 1.0 means it lost money overall, regardless of win rate.
A strategy that earned 20% by taking enormous risk is a very different thing from one that earned 20% with controlled drawdowns.
Risk metrics
What risk metrics matter when evaluating an algo strategy?
Risk metrics answer a different question: what did it cost to generate those returns? A strategy's return only means something once you know its max drawdown, its volatility, and how it behaves when conditions turn against it.
Max Drawdown
Max drawdown is the largest peak-to-trough decline in the strategy's history. If you had started running it at the peak, this is how far down you would have been before recovering. The most psychologically important number — if you couldn't sit through this, the strategy isn't right for you regardless of its returns.
Sharpe Ratio
Return divided by volatility. Measures how much return the strategy generated per unit of risk. As a general benchmark, above 1.0 is acceptable and above 2.0 is strong — though these thresholds vary by strategy type and timeframe. Very high Sharpes (above 3.0) are rare but achievable on live records.
Sortino Ratio
Similar to Sharpe, but only penalises downside volatility — it ignores upside volatility, which investors generally don't mind. Often considered a more practical risk measure than Sharpe for strategies that have asymmetric return profiles.
Calmar Ratio
CAGR divided by the max drawdown. Measures return relative to the worst loss experienced. A Calmar of 1.0 means the annual return equals the max drawdown. Higher is better. Useful for comparing strategies where drawdown tolerance is the primary concern.
Strategy specifications
What strategy specifications should you check before running?
Beyond performance, the specs section describes how a strategy actually operates — its mechanics, constraints, and practical requirements. These details affect whether a strategy fits your account size, your risk tolerance, and how you want to run it. Read them before committing — they matter as much as the return metrics.
Min Capital
The minimum account size required to run the strategy as designed. Running with less capital than the minimum affects position sizing and may significantly alter actual returns versus the track record.
Frequency / Rebalancing
How often the strategy trades and rebalances its positions. An end-of-day strategy behaves very differently from an intraday one — different execution risk, different tax implications, different monitoring requirements.
Position Sizing
How the strategy allocates capital across its positions — equal weight, volatility-weighted, conviction-based, etc. This has a significant effect on drawdown profile and concentration risk.
Capacity
The maximum capital the strategy can absorb before its own trading starts to affect market prices and degrade performance. When capacity is nearly full, capital added later may experience different execution than the published track record.
Red flags to watch for
Red flags to watch for
Most metrics tell you what a strategy did. These are the signals that something may be worth scrutinizing. None of these are automatic reasons to reject a strategy. They are prompts to look more carefully — to ask what explains the number before accepting it at face value.
evaluation red flags
Sharpe above 3.0Rare in live trading. Can indicate overfitting or a strategy not yet tested across varied market conditions. Worth scrutinizing closely, not automatically disqualifying. The right question is: what explains this Sharpe? A high-frequency strategy with very consistent small daily gains can legitimately achieve a high Sharpe. A monthly-rebalancing equity strategy achieving the same number warrants a closer look at the underlying data.
Suspiciously smooth equity curveReal strategies have rough patches. A live record with almost no losing periods may mean the period was unusually favorable, or that the strategy hasn't been stress-tested by difficult conditions yet.
Large max drawdown vs stated SharpeA strong Sharpe alongside a very large max drawdown can signal a strategy that earns steadily but occasionally takes a large loss. Check both numbers together. A strategy with a Sharpe of 1.8 and a max drawdown of -32% is a very different risk proposition to one with a Sharpe of 1.8 and a max drawdown of -8%. The Sharpe tells you about day-to-day smoothness; the max drawdown tells you about tail risk.
Common questions
How to evaluate an algo trading strategy — FAQs
What is a good CAGR for an algo trading strategy?
A good CAGR depends heavily on the strategy type, asset class, and leverage used. As a rough benchmark for a systematic equity strategy, a CAGR of 15–25% with a Sharpe ratio above 1.0 is generally considered strong — though what matters more than the absolute return is its relationship to drawdown. A Calmar ratio above 1.0 (CAGR divided by max drawdown) is a useful combined signal. Return without risk context is nearly meaningless.
What is the Sharpe ratio and why does it matter?
The Sharpe ratio measures return per unit of risk (volatility). A Sharpe of 1.0 means the strategy returned one unit of return for every unit of volatility — generally considered the minimum acceptable threshold. Above 2.0 is strong. The ratio matters because two strategies with the same return but different Sharpe ratios represent very different risk propositions.
What is overfitting in algo trading?
Overfitting occurs when a strategy is tuned too closely to historical data — performing well in backtests but failing in live markets because the patterns it learned were noise rather than signal. It is the most common reason promising backtests disappoint in live trading. The defense is out-of-sample testing: validating on data the strategy was never trained on.
What is a good Sharpe ratio for algo trading?
A Sharpe ratio above 1.0 is generally considered acceptable for a retail algo trading strategy. Above 2.0 is strong. Very high Sharpes (above 3.0) are rare in live trading and warrant scrutiny — they can indicate overfitting or a strategy not tested across diverse market conditions. Sharpe ratio should always be evaluated alongside max drawdown, not in isolation. The metric was originally defined by William F. Sharpe to measure return per unit of risk.
What is CAGR in trading?
CAGR stands for Compound Annual Growth Rate — the annualized return of a strategy, accounting for compounding. It's more meaningful than total return when comparing strategies with different track record lengths. A strategy up 50% over 3 years has a lower CAGR than one up 50% over 18 months. Always read CAGR alongside risk metrics like max drawdown and Sharpe ratio.
What is a good max drawdown for a trading strategy?
There's no universal good number — it depends on your risk tolerance and the strategy's return profile. The key question is: if you started running it at the worst possible moment, could you hold through the full drawdown without stopping the strategy? If not, the drawdown is too large for you regardless of the historical recovery.
What is profit factor in trading?
Profit factor is total gross profit divided by total gross loss. Above 1.0 means the strategy was profitable overall. Above 1.5 is generally considered solid. Profit factor is most useful read alongside win rate — a strategy can have a low win rate but high profit factor if its winners are significantly larger than its losers.
key takeaway
Return, Sharpe, Max Drawdown, Win Rate, Sortino, Calmar, Profit Factor — each metric answers a different question about a strategy. Read together, they give a complete picture. Read in isolation, any single number can mislead.