Statistical Arbitrage (Stat-Arb)

Category: science

A strategy that uses mathematical models to identify mean-reversion opportunities between correlated assets.

Unlike simple cross-exchange arbitrage (which exploits price differences of the *same* asset), Stat-Arb exploits price divergences between *related* assets (e.g., two stocks in the same sector). If the price ratio of Asset A to Asset B deviates from its historical Z-score, the model sells the expensive one and buys the cheap one, betting on a return to the mean.

Common Examples

• Our Stat-Arb model monitors 500 correlated pairs, triggering trades only when the price divergence exceeds three standard deviations from the historical mean.
• Stat-Arb requires a high-fidelity data feed to ensure that the correlation coefficient between assets hasn’t undergone a structural "regime change."