Polymarket Whale Copy-Trading: Research Report

Based on a comprehensive parameter sweep across 2 strategy variants, no whale copy-trading strategy on Polymarket produced statistically significant positive returns after Bonferroni correction. The best-performing strategy was 'immediate' with a Sharpe ratio of 0.00 and total PnL of $267,291.22, but this result was not distinguishable from random market assignment at the 5% significance level.

The 95% bootstrap confidence interval for the best strategy's Sharpe ratio is [0.381, 0.622]. Extrapolating from 5.7 months of simulated data, the estimated annual return is $566,695.16 ($47,224.60/month). These estimates use simple linear extrapolation and should be interpreted with caution.

Caveats: (1) Returns are based on historical backtesting with simulated execution; actual trading may differ due to liquidity constraints and market impact. (2) Slippage is modeled as a fixed spread, which may underestimate costs in illiquid markets. (3) The data period is limited and may not represent future market conditions. (4) Detection delay is simulated but real-world whale detection latency may vary. (5) No transaction fees or gas costs are included in PnL calculations.

This report evaluates four whale copy-trading strategies on Polymarket prediction markets using a systematic parameter sweep with statistical significance testing.

Strategies Tested

1. Immediate Follow -- Copy whale trades as soon as detected
2. Confirmation Follow -- Wait for price confirmation before entering
3. Size-Weighted Follow -- Weight position sizes by whale confidence score
4. Contrarian Fade -- Take the opposite side of whale trades

Parameter Grid

Statistical Testing

Each strategy's best configuration (by Sharpe ratio) is tested for statistical significance using a bootstrap permutation test. The null hypothesis is that the strategy performs no better than random market assignment. Market IDs are shuffled to generate a null distribution of total PnL values. Bonferroni correction is applied to account for multiple strategy comparisons (alpha / n_strategies).

immediate

size_weighted

The table below shows the permutation test results for each strategy's best configuration. A strategy is considered statistically significant if its p-value falls below the Bonferroni-corrected threshold (alpha / number of strategies tested).

Interpretation: A p-value below the corrected threshold indicates that the strategy's performance is unlikely to be due to chance alone. However, statistical significance does not guarantee future profitability.

Distribution of total PnL across all parameter configurations and cumulative PnL curves.

Best Strategy: immediate

Sharpe Ratio 95% CI: [0.381, 0.622]

Estimated Annual Return: $566,695.16 ($47,224.60/month over 5.7 months)

Note: Annual return is a simple linear extrapolation from backtest data. It does not account for compounding, market regime changes, or capacity constraints.

Caveats

• Returns are from historical backtesting with simulated execution
• Slippage is modeled as a fixed spread; actual costs may differ
• Data period is limited and may not represent future conditions
• Detection delay is simulated; real-world latency may vary
• No transaction fees or gas costs included

Top 20 parameter configurations by rank (Sharpe ratio descending):