Statistical arbitrage - optimization
Trade two co-moving names when their spread deviates; relationship breaks are the tail risk.
Overview
Statistical arbitrage - optimization sits in the Stocks chapter of the systematic catalog. On QUSXFI we treat it as a testable hypothesis: specify entries, exits, sizing, and costs—then ask whether edge survives out-of-sample scrutiny.
Discretionary traders often arrive at similar ideas intuitively; the quantitative version forces you to write the rule before you see the next bar. That discipline is what makes results reproducible—or exposes them as luck.
Based on the research catalog 151 Trading Strategies (Kakushadze & Serur, 2018), section 3.18. Educational summary—not a replication of the full formal definition.
Relative-Value Logic
Statistical arbitrage - optimization trades co-moving names when a spread deviates—convergence is the thesis, relationship stability is the assumption.
Before backtesting Statistical arbitrage - optimization, write the economic hypothesis in one sentence a risk manager would accept or reject.
Implementation and Research Process
Estimate hedge ratios for Statistical arbitrage - optimization with rolling windows; static betas lie during relationship breaks.
Decompose Statistical arbitrage - optimization into signal, portfolio construction, and execution modules—each must be path-independent given the same historical tape.
Stress Statistical arbitrage - optimization costs at 2× baseline; many Stocks edges live or die on slippage alone.
Risk: What Breaks This Strategy
Pairs on Statistical arbitrage - optimization assume a stable relationship; mergers, index rebalances, or idiosyncratic fraud break cointegration without warning.
Spread convergence is not guaranteed on your horizon—carry and financing on the short leg eat edge while you wait.
Stop rules on ratio trades are harder than on directional beta; correlation spikes in crises.
Common Mistakes to Avoid
- Changing Statistical arbitrage - optimization parameters after each losing week—implicit discretion destroys reproducibility.
- Waiting for Statistical arbitrage - optimization convergence without a hard stop when fundamentals diverge.
- Deploying Statistical arbitrage - optimization live before paper trading through at least one adverse Stocks month.
- Stacking Statistical arbitrage - optimization with correlated sidebar strategies without netting exposures.
How to Study This Strategy
- Define Statistical arbitrage - optimization entry z-score, exit, and hard stop on the spread—no 'wait and see.'
- Archive the pair if Statistical arbitrage - optimization structural test fails—do not re-enable without fresh evidence.
- Paper Statistical arbitrage - optimization through a relationship scare (headline, merger rumor) without overriding rules.
- Attribute Statistical arbitrage - optimization P&L to beta, spread, and financing separately.
- Select one pair for Statistical arbitrage - optimization; prove cointegration in-sample and monitor out-of-sample.
Key Takeaways
- Statistical arbitrage - optimization bets on spread mean-reversion between co-moving names—relationship breaks are tail events, not noise.
- Cointegration on Statistical arbitrage - optimization requires live monitoring; one merger rumor can invalidate years of history.
- Convergence is not scheduled on your horizon—financing on the short leg eats edge while you wait.
- Stop rules on ratios are harder than on beta—define exit before entry.
- Archive broken pairs from Statistical arbitrage - optimization with a written reason; do not re-enable without fresh tests.
Learning Tip
For Statistical arbitrage - optimization, keep a 'broken pairs' graveyard with dates and reasons—resist reviving dead relationships without new tests.
Explore related strategies in the sidebar or return to the full catalog.