Research into three statistical arbitrage strategies for Polymarket BTC binary options (5-minute Up/Down markets).
Buy underpriced binary options after the market overshoots fair value following BTC moves. Fair probability derived from Deribit IV via Black-Scholes.
| Metric | Value |
|---|---|
| Trades | 574 |
| Win Rate | 54.7% (CI: 50.6%–58.8%) |
| PnL | +$13,338 |
| Sharpe | 7.9 |
| p-value | 0.012 |
Follow the consensus direction when 4+ tracked traders agree on a BTC market outcome.
| Metric | Value |
|---|---|
| Markets (4+ traders) | 52 |
| 5m Accuracy | 75.0% |
| 15m Accuracy | 57.7% |
| p-value (5m) | 0.0002 |
Buy both sides of a market when prices sum to less than $1.00, redeem for risk-free profit. Confirmed by academic research — $40M extracted from Polymarket historically (Saguillo et al. 2025).
| Metric | Value |
|---|---|
| Trades | 1,414 |
| Win Rate | 98.9% (CI: 98.3%–99.4%) |
| Sharpe | 111.8 |
| p-value | < 0.0001 |
backtests/ Backtest engines and results
contrarian/ Overreaction capture (v1-v3)
convergence/ Multi-trader consensus
merge_arb/ Merge arbitrage
copy_trade/ Real-time trade poller for 11 tracked traders
data/ Binance, Deribit, Polymarket API clients
models/ Black-Scholes binary pricing, Kelly sizing
scripts/ Trader analysis and data fetching
live_backtest*.py Paper trading harnesses (v1-v3)
LEARNINGS.md Full research notes and results