Statistical Arbitrage¶
Difficulty intermediate
Overview¶
Statistical arbitrage (stat arb) uses quantitative models to identify and exploit pricing inefficiencies across multiple securities simultaneously. Unlike pairs trading (2 assets), stat arb typically involves portfolios of 20-100+ securities.
Core Methodology¶
Mean-Reverting Portfolios¶
Find a portfolio of assets whose combined value is mean-reverting:
Portfolio Value = Σ wᵢ × Assetᵢ
Find weights w that minimize variance of portfolio returns
while maintaining market neutrality
where:
wᵢweight on asset i (long positive, short negative) ·Assetᵢprice or return of asset i ·Σsum over the basket. does: constructs a synthetic mean-reverting spread from many assets. Constrained optimization: minimize portfolio variance subject to net-beta = 0 — the foundation of multi-asset stat arb.
Cointegration-Based Portfolios¶
Johansen Test¶
Tests for multiple cointegrating relationships across a basket of assets — an extension of the two-variable Engle-Granger procedure to N variables.
- Estimates the number of independent linear combinations of the basket that are stationary (the cointegration rank
r). - Trace and max-eigenvalue statistics are compared to critical values to choose
r. - Each cointegrating vector defines a mean-reverting spread; the portfolio weights for trading are read directly from those eigenvectors.
- Requires sufficient history (typically 250+ trading days) and ongoing re-estimation as relationships drift.
Risk Management¶
| Risk | Mitigation |
|---|---|
| Model breakdown | Monitor cointegration, have stops |
| Liquidity | Trade liquid names only |
| Execution cost | Use VWAP/TWAP, minimize market impact |
| Crowding | Many funds run similar strategies |
| Regime change | Adaptive models, regime detection |
Performance Expectations¶
| Metric | Typical Range |
|---|---|
| Sharpe Ratio | 1.5-3.0 |
| Max Drawdown | 3-10% |
| Win Rate | 55-65% |
| Annualized Return | 8-15% |
| Market Beta | Near zero |
Practical Guidelines¶
- Transaction Costs Kill — Stat arb margins are thin
- Execution Matters — Slippage can eliminate edge
- Monitor Continuously — Relationships break down
- Diversify — Many uncorrelated stat arb strategies
- Capacity Limits — Strategy degrades with more capital
- Infrastructure Required — Needs sophisticated systems
Next Steps¶
- Pairs Trading — Simpler 2-asset version
- Market Making — Spread-based strategies
- Factor Investing — Factor-based approaches