Skip to main content
Algo Strategy Analyzer
Backtesting & Metrics

Algorithmic Trading Metrics: Complete Guide to Evaluate Your Strategy

Sharpe Ratio, Profit Factor, Win Rate, Drawdown, the equity curve... What each metric means, how to interpret it, and what values are realistic vs suspicious.

Rubén Villahermosa Rubén Villahermosa
January 25, 2026 25 min read

Your backtest shows numbers everywhere: Sharpe Ratio 1.8, Profit Factor 2.3, Win Rate 58%, Max Drawdown 15%... Are they good? Bad? What do they really mean?

Metrics are the language of algorithmic trading. Without understanding them, you're flying blind. With this article you'll learn to interpret each fundamental metric, what values are realistic and which are warning signs of problems in your backtest. If you want to dive deeper into risk-adjusted metrics after this article, we have a dedicated advanced metrics guide.

"You can't improve what you can't measure." — Peter Drucker

Calculate your strategy's metrics

Upload your backtest and automatically get more than 50 professional metrics, including equity curve analysis.

Analyze my strategy →
TOP 8

The 8 fundamental metrics of algorithmic trading

There are dozens of metrics, but these 8 are the essentials you must master before evaluating any strategy:

Metric What it measures Category
Net Profit Total profit in money Profitability
Profit Factor Ratio gross gains/losses Efficiency
Win Rate Percentage of winning trades Profitability
Average Profit Average gain per trade Profitability
Maximum Drawdown Maximum loss from peak Risk
Sharpe Ratio Return adjusted by total volatility Risk adjusted
Expectancy Expected average gain per trade Expectancy
Equity Curve Stability and consistency over time Behavior

Let's look at each one in detail, starting with the simplest.

PROFIT $

Profitability Metrics

These metrics answer the most basic question: does the strategy make money?

Net Profit and Return

Net Profit is simply how much money you made or lost in total. It's the most intuitive metric but also the most misleading if looked at alone.

⚠️

The problem with isolated Net Profit

A strategy with $50,000 profit may seem better than one with $10,000. But if the first risked $500,000 and the second only $20,000, the percentage return tells a different story.

Return (percentage return) normalizes profit relative to initial capital:

Return (%) = (Net Profit / Initial Capital) × 100

CAGR: the return that actually matters

CAGR (Compound Annual Growth Rate) is the return metric you should use instead of average annual return. The difference is critical.

CAGR = (Final Value / Initial Value)^(1/years) - 1

⚠️

The average annual return trap

Imagine you invest $10,000. Year one you gain 100% ($20,000). Year two you lose 50% ($10,000). Your average annual return is +25%, but your CAGR is 0% — you made nothing. The greater the volatility, the larger the gap. According to Vanguard Total Stock Market ETF data, the average annual return was 10.01% but the actual CAGR was only 8.13%.

Rule of thumb: Always use CAGR to compare strategies, especially if they operate over different time periods. Brokers and advisors prefer to show average annual return because it looks more impressive. CAGR gives you reality.

Win Rate: the hit percentage

Win Rate (hit rate) is the percentage of trades that end in profit.

Win Rate = (Winning Trades / Total Trades) × 100

Important: A high Win Rate does NOT mean a good strategy. You can win 90% of your trades but if when you lose, you lose big, you'll still be negative. That's why Win Rate must always be evaluated together with the risk/reward ratio. Here's a revealing fact: a 40% Win Rate with a 1:3 R:R can be more profitable than a 60% Win Rate with a 1:0.5 R:R -- the mathematical relationship between both metrics matters more than the percentage in isolation.

Additionally, the statistical significance of Win Rate depends on the number of trades. According to the Central Limit Theorem, with only 20 trades, a 65% Win Rate has a p-value above 0.2, meaning there's more than a 20% probability that result is due to pure chance. For metrics to be reliable, you need at least 30 trades for initial inference and 100+ for robust results.

Average Profit per Trade

Average Profit per Trade (Average Trade) is Net Profit divided by total number of trades. It's a critical metric, especially for intraday strategies.

Average Profit = Net Profit / Number of Trades

🚨

Critical for intraday strategies

Average profit must exceed your execution costs. If your average profit is $20 per trade but you pay $8 commission + $7 slippage, your real profit is only $5. Many strategies seem profitable in backtest but are unviable due to costs.

Practical example:

  • Strategy A: Average profit $150/trade, 50 trades/year = viable
  • Strategy B: Average profit $15/trade, 500 trades/year = at risk if costs > $10

Algo Strategy Analyzer includes stress tests that simulate cost increases to verify if your strategy remains viable.

Profit Factor: gains vs losses efficiency

Profit Factor is not just a profitability metric — it's an efficiency metric that relates what you win to what you lose.

Profit Factor (definition): Ratio between total gross profit and total gross loss. A PF of 2.0 means for every $1 lost, the strategy gains $2.

Profit Factor = Gross Profit / Gross Loss

Institutional traders consider a Profit Factor above 1.75 as the minimum recommended for a viable strategy. However, a PF above 4.0 is a clear warning sign indicating probable overfitting or a sample size that's too small. According to a study by uTrade Algos, 95% of traders who use backtesting with proper metrics report significant improvements in their performance, demonstrating the importance of measuring correctly.

Profit Factor Interpretation
< 1.0 Losing - Loses more than it gains
1.0 - 1.5 Marginal - Profitable but fragile to costs
1.5 - 2.5 Good - Typical range of robust strategies
> 3.0 Suspicious - Probable overfitting or small sample
RISK

Risk Metrics

Profitability metrics alone don't tell the whole story. How much risk did you take to get that return?

Maximum Drawdown: the most important metric

Maximum Drawdown (Max DD) is, in my opinion, the most important metric of all. It measures the maximum loss from a peak to the deepest valley.

Maximum Drawdown (definition): The largest percentage drop from a historical maximum to the subsequent minimum before a new maximum. It's the strategy's "worst moment". We dive deep into this critical metric in our complete drawdown guide.

Max DD = (Peak - Valley) / Peak × 100

💀

Drawdown is the "killer" of traders

A 50% drawdown requires a 100% gain to recover. A 75% DD needs 300%. Live drawdown is usually 1.5-2x greater than in backtest due to slippage, unmodeled costs and real market conditions.

Drawdown Gain needed to recover Psychological impact
10% 11% Tolerable
20% 25% Uncomfortable
30% 43% Hard to maintain
50% 100% Devastating
75% 300% Practical ruin

Maximum Drawdown and Time Underwater

PEAK VALLEY (-20%) RECOVERY DD 20% Time Underwater Equity curve Underwater zone Recovery period

Drawdown measures how much it falls; Time Underwater measures how long it takes to recover the previous peak

Time Underwater: the forgotten metric

Maximum Drawdown tells you how much you can lose, but Time Underwater tells you how long you'll suffer. Both metrics are equally important for a trader's mental health.

⏱️

What is Time Underwater?

It's the period from when equity reaches a maximum until it surpasses it again. A strategy can have a "small" DD of 15%, but if it takes 18 months to recover, few traders would psychologically endure it.

Real example: A 30% Drawdown lasting 2 months is easier to trade than a 15% DD lasting 14 months. Depth matters, but duration can be the factor that breaks your discipline.

Sharpe Ratio: the consensus problem

The Sharpe Ratio, created by Nobel laureate William F. Sharpe, is the most cited metric in the industry. But it has a serious problem: there's no consensus on how to calculate it for trading.

Sharpe Ratio = (Return - Risk-Free Rate) / Standard Deviation

⚠️

The annualization problem

The financial community annualizes Sharpe by multiplying by √252 (trading days). But this formula assumes returns are independent (no serial correlation). In real trading, this can overestimate the Sharpe by up to 65% according to CAIA studies.

Why each platform calculates Sharpe differently

Inconsistencies include:

  • Risk-free rate: 3-month bonds? 1 year? Zero? Each platform chooses differently.
  • Calculation period: Daily returns, monthly, or per-trade? Very different results.
  • Annualization: Multiplying by √12 (monthly) or √252 (daily) gives different results.
  • Serial correlation: If returns are correlated, the square root of time doesn't apply.
"Calculating Sharpe ratios using daily vs monthly returns can give results 20% different for the same asset. If returns have autocorrelation, the difference can reach 65%." — CAIA Association
Sharpe Ratio Interpretation Context
< 1.0 Suboptimal Ignore after costs
1.0 - 2.0 Good Realistic goal for retail
2.0 - 3.0 Very good Verify calculation and overfitting
> 3.0 Suspicious Probable calculation error or overfitting

Sharpe Ratio: same average return, different volatility

0% Mean return: 8% Low Sharpe (high volatility) High Sharpe (low volatility) Large standard deviation High Sharpe (~2.0) Low Sharpe (~0.5)

Same 8% average return, but the green strategy is more consistent (less dispersion)

According to the studies of David H. Bailey and Marcos López de Prado (author of Advances in Financial Machine Learning), a Sharpe Ratio above 2.0 in backtest typically drops to 1.0-1.5 in live trading, representing a significant deterioration.

In practice, this means that if your backtest shows a Sharpe of 1.5, you could actually be trading with a Sharpe below 1.0 in the real world. That's why many professional managers require a minimum Sharpe of 2.0 in backtest before considering a strategy viable.

💡

Practical recommendation

Don't obsess over the exact Sharpe value. Use it to compare strategies within the same platform, not between different platforms. And consider Sortino as a more robust alternative.

Sortino Ratio: only the risk that matters

The Sortino Ratio, developed by Frank A. Sortino in the 1980s, solves a Sharpe problem: it only penalizes negative volatility (downside), not positive. Traders don't mind gaining more than expected.

Sortino Ratio = (Return - Risk-Free Rate) / Downside Deviation

Rule of thumb: If your Sortino is significantly higher than your Sharpe, your strategy has good positive volatility but controlled downside risk. That's good.

Calmar Ratio: drawdown in the denominator

The Calmar Ratio, introduced by Terry W. Young in 1991, puts maximum drawdown in the denominator. It's especially useful for strategies where the worst moment (the drawdown) is more relevant than average volatility.

Calmar Ratio = Annualized Return / Maximum Drawdown

Ratio What it measures in denominator When to use it
Sharpe Total volatility (standard deviation) General strategy comparison
Sortino Only negative volatility (downside) When downside risk matters more
Calmar Maximum Drawdown When worst-case scenario is critical

A Calmar > 1.0 means your annual return exceeds your worst drawdown. A Calmar > 3.0 is excellent. If you have good Sharpe but poor Calmar, it means although your average volatility is low, your worst moment was catastrophic.

Recovery Factor: your strategy's resilience

The Recovery Factor measures your strategy's ability to bounce back from drawdowns. It's a simple but very revealing metric that many traders overlook.

Recovery Factor = Net Profit / Maximum Drawdown

Example:

  • Strategy A: Profit $100,000, Max DD $12,500 → RF = 8.0 (excellent)
  • Strategy B: Profit $120,000, Max DD $30,000 → RF = 4.0 (good)

Although Strategy B has more absolute profit, Strategy A is more efficient relative to the risk taken.

Recovery Factor Interpretation
< 1.0 Dangerous - Doesn't recover its own losses
1.0 - 2.0 Marginal - Recovers but with little margin
2.0 - 5.0 Good - Typical range of robust strategies
> 5.0 Excellent - High resilience

Key difference from Calmar Ratio: Recovery Factor uses total cumulative net profit, while Calmar annualizes the return. This makes Recovery Factor more useful for evaluating complete backtests, and Calmar for comparing strategies across different time periods.

CURVE 📈

The Equity Curve: what the numbers don't tell

The equity curve is, for me, as important as any numerical metric. It shows how your capital evolves over time and reveals the stability and consistency of the strategy.

📊

It's not just about how much you win, but how you win it

Two strategies can have the same Net Profit and Sharpe Ratio, but completely different equity curves. One can be smooth and ascending; another can have steps, plateaus and sudden drops. The first is tradeable; the second can destroy you psychologically.

What to look for in an equity curve

Real Curve vs Problematic Curve (same final result, very different experience)

$150k $125k $100k $75k $50k Long underwater period (6+ months) Abrupt rise (repeatable?) Real curve (natural zigzag, constant progress) Problematic curve Same result Time →

Both curves end at $150k, but the red one would have made you quit due to prolonged stagnation frustration

✅ Positive signs

  • Constant slope: Stable gains over time
  • Brief drawdowns: Recovers quickly from drops
  • Consistency: Similar behavior in different periods
  • No dependency: Doesn't depend on a few large trades

🚫 Warning signs

  • Steps: All profit comes from few trades
  • Long plateaus: Periods without winning or losing
  • Vertical drops: Sudden, unexpected losses
  • Curve at the end: Only wins in the most recent period

Problematic patterns in the curve

📉

"Staircase" curve

Indicates that profit depends on a few very profitable trades. If those trades hadn't occurred, the strategy would be losing. High dependency = high risk.

📊

Curve that only rises at the end

Sign of possible overfitting to the most recent period. The strategy may be optimized for market conditions that have already passed.

📈

"Too perfect" curve

A perfect straight line upward is suspicious. Real strategies have volatility. If it looks too good, there's probably overfitting or backtest errors.

🔍

Analyze your equity curve

In Algo Strategy Analyzer you can visualize your equity curve, identify drawdowns, and analyze the distribution of winning and losing trades.

AVG E

Expectancy Metrics

Expectancy: the expected average gain

Expectancy (mathematical expectation) combines Win Rate with the average size of wins and losses.

Expectancy = (Win Rate × Avg Win) - (Loss Rate × Avg Loss)

Example:

  • Win Rate: 40% (Loss Rate = 60%)
  • Average win: $300
  • Average loss: $100
  • Expectancy = (0.40 × $300) - (0.60 × $100) = $120 - $60 = $60 per trade

If Expectancy is positive, the strategy is profitable long-term, regardless of Win Rate. What matters is not winning more often, but that what you win when you're right compensates for what you lose when you're wrong.

Risk/Reward Ratio and Win Rate: the mathematical connection

Payoff Ratio: the relative size of your wins

The Payoff Ratio (also called Average Win / Average Loss) measures how much you win on average when you're right, compared to how much you lose when you're wrong.

Payoff Ratio = Average Win / Average Loss

Important: The Payoff Ratio by itself tells you nothing about your strategy's profitability. You need to combine it with Win Rate. A strategy with Payoff Ratio of 3.0 but Win Rate of 20% is a loser. That's why Expectancy exists — it unites both metrics.

The Risk/Reward Ratio (R:R) measures how much you risk vs how much you can gain. The formula is:

Risk/Reward = (Take Profit - Entry) / (Entry - Stop Loss)

The key is that R:R and Win Rate are mathematically connected. From this relationship, we can calculate the minimum Win Rate needed to break even for any given R:R:

Breakeven Win Rate = 1 / (1 + R:R)

MATRIX %

Profitability Matrix: Win Rate vs Risk/Reward

This table shows the expectancy for each combination of Win Rate and Risk/Reward ratio. Positive values (green) indicate profitable strategies; negative (red) indicate losses.

Win Rate R:R 0.5 R:R 1.0 R:R 1.5 R:R 2.0 R:R 2.5 R:R 3.0 R:R 4.0 R:R 5.0
20% -0.70 -0.60 -0.50 -0.40 -0.30 -0.20 0.00 +0.20
25% -0.63 -0.50 -0.38 -0.25 -0.13 0.00 +0.25 +0.50
30% -0.55 -0.40 -0.25 -0.10 +0.05 +0.20 +0.50 +0.80
35% -0.48 -0.30 -0.13 +0.05 +0.23 +0.40 +0.75 +1.10
40% -0.40 -0.20 0.00 +0.20 +0.40 +0.60 +1.00 +1.40
45% -0.33 -0.10 +0.13 +0.35 +0.58 +0.80 +1.25 +1.70
50% -0.25 0.00 +0.25 +0.50 +0.75 +1.00 +1.50 +2.00
55% -0.18 +0.10 +0.38 +0.65 +0.93 +1.20 +1.75 +2.30
60% -0.10 +0.20 +0.50 +0.80 +1.10 +1.40 +2.00 +2.60
65% -0.03 +0.30 +0.63 +0.95 +1.28 +1.60 +2.25 +2.90
70% +0.05 +0.40 +0.75 +1.10 +1.45 +1.80 +2.50 +3.20
75% +0.13 +0.50 +0.88 +1.25 +1.63 +2.00 +2.75 +3.50
💡

How to read this table

Find your Win Rate in the left column and your R:R in the top row. The resulting value is the expectancy per unit of risk. For example: with 40% Win Rate you need at least R:R 1.5 to be profitable (breakeven). With 25% Win Rate, you need minimum R:R 3:1.

Analyze your complete strategy

Algo Strategy Analyzer automatically calculates all these metrics, visualizes your equity curve, and includes cost stress tests.

Analyze my strategy →
ALERT !

Realistic vs suspicious values

This table is your quick reference to detect problems in your backtest:

Metric Realistic Suspicious Warning sign
Sharpe Ratio 1.0 - 2.5 > 3.0 Overfitting or calculation error
Profit Factor 1.3 - 2.5 > 4.0 Curve fitting, small sample
Win Rate 35% - 65% > 80% Fragile strategy or martingale
Max Drawdown 15% - 30% < 5% Too good to be true
Annual return 15% - 50% > 100% Excessive leverage
Equity curve Smooth, ascending Perfect, no DD Almost certain overfitting
🚫

The golden rule

If all your backtest metrics are "excellent", you probably have overfitting. Real strategies have flaws. A Sharpe of 1.5 with 20% drawdown is more believable than a Sharpe of 3.0 with 5% drawdown.

ALL

How to interpret metrics together

No single metric tells the whole story. You must look at them together.

The evaluation framework

1

What does the equity curve look like?

Before looking at numbers, look at the curve. If you don't like what you see, the rest doesn't matter.

2

Is the drawdown survivable?

Could you psychologically handle that DD multiplied by 1.5-2x?

3

Is it profitable after costs? (Average profit)

Average profit must exceed commissions + slippage with margin.

4

Is it statistically valid? (N > 100 trades)

With fewer than 100 trades, metrics are not statistically significant.

5

Is it suspicious? (Metrics too good)

If everything is "excellent", check overfitting and validate with Walk Forward.

Conclusion

Metrics are the language for evaluating trading strategies, but no single metric is perfect on its own. Sharpe Ratio has calculation and consensus problems. Win Rate can deceive without the R:R context. Profit Factor can inflate with few trades.

The most important things:

  1. Look at the equity curve before any number
  2. Drawdown is the metric that defines if you'll survive
  3. Average profit must exceed your real costs
  4. Don't obsess over the exact Sharpe — it varies between platforms
  5. Use the Win Rate/R:R matrix to verify mathematical viability
  6. Distrust "perfect" metrics

Now that you understand the metrics, the next step is to learn how to properly validate your strategy using techniques like Walk Forward and Monte Carlo. If you're still designing your system, check out the anatomy of a trading strategy and our guide to create algorithmic strategies. We also recommend reviewing the essential tools and making sure your market data is high quality.

Get professional metrics for free

More than 50 metrics calculated automatically, equity curve analysis, and cost stress tests.

Analyze my strategy for free →
FAQ ?

Frequently Asked Questions

What is a good Sharpe Ratio for a trading strategy?
Why does each platform calculate the Sharpe Ratio differently?
What Win Rate do I need to be profitable?
Why is the equity curve so important?
What is average profit per trade and why does it matter?
What metrics should I look at first in a backtest?
Why is real drawdown usually greater than in backtest?
How do I know if my metrics indicate overfitting?