Strategy Evaluation Metrics for Prediction Market Backtests
Key performance metrics for evaluating backtested prediction market strategies — Sharpe ratio, drawdown, win rate, and more.
Why Metrics Matter
A profitable backtest is not enough — you need to evaluate how profits are generated and whether the strategy is robust. The right metrics reveal risk-adjusted performance, consistency, and whether your strategy will survive real-world conditions on Polymarket prediction markets.
Essential Performance Metrics
| Metric | Formula / Description | Good Threshold |
|---|---|---|
| Sharpe Ratio | (Mean Return - Risk-Free Rate) / Std Dev of Returns | > 1.5 for prediction markets |
| Sortino Ratio | (Mean Return - Target) / Downside Deviation | > 2.0 |
| Max Drawdown | Largest peak-to-trough equity decline | < 20% of capital |
| Profit Factor | Gross Profits / Gross Losses | > 1.5 |
| Win Rate | Winning Trades / Total Trades | Context-dependent |
| Calmar Ratio | Annualized Return / Max Drawdown | > 2.0 |
Sharpe Ratio
The Sharpe ratio measures risk-adjusted returns — how much return you earn per unit of risk. For prediction market strategies backtested on PolyHistorical data, a Sharpe above 1.5 indicates a strong strategy, while below 0.5 suggests the returns do not adequately compensate for the risk taken.
Calculating Sharpe from Backtest Data
- Calculate returns for each period (hourly or daily)
- Compute the mean and standard deviation of returns
- Sharpe = (mean_return - risk_free_rate) / std_dev
- Annualize by multiplying by sqrt(periods_per_year)
Maximum Drawdown
Max drawdown measures the worst peak-to-trough decline in your equity curve. This is arguably the most important risk metric because it determines whether you can psychologically and financially survive the strategy's worst period.
Win Rate vs Payoff Ratio
Win rate alone is misleading. A strategy with a 30% win rate can be highly profitable if winning trades are much larger than losing trades. Conversely, a 90% win rate strategy can be destroyed by rare large losses. Always evaluate win rate together with the average win / average loss ratio.
Strategy Archetypes
- High win rate, low payoff: Spread capture, market making (70%+ win rate, small wins)
- Low win rate, high payoff: Trend following, event trading (30-40% win rate, large wins)
- Balanced: Mean reversion strategies (50-60% win rate, moderate wins)
Metrics Specific to Prediction Markets
Beyond standard trading metrics, prediction market strategies should be evaluated on:
- Resolution accuracy: How often does your strategy correctly predict the outcome?
- Edge per trade: Average profit relative to the bid-ask spread (must exceed spread)
- Liquidity utilization: How much of available depth your strategy actually accesses
- Turnover ratio: Trading frequency relative to available markets
Using PolyHistorical for Evaluation
PolyHistorical's sub-second order book data lets you compute these metrics with realistic execution assumptions. Account for slippage using actual order book depth at each timestamp, and include Polymarket fee calculations in your return series for accurate metric computation.