Building Triangulated Statistical Arbitrage for a Solo Trader
Summary
The article describes a three-part approach to equity statistical arbitrage for independent traders. First, rank related stock pairs using measures of historical mean-reversion returns and consistency of convergence, then retain economically sensible candidates. Next, translate each pair into views on its constituent tickers and combine those views into a long-short portfolio. Weight signals by pair quality, and account for how many pairs support a ticker and whether their directions agree. The discussion favors trading the strongest ticker signals rather than the noisy middle of the distribution.
The author reports that pair ranking creates the largest performance separation: the best tier has a Sharpe of 1.6 versus 0.42 for the weakest, in before-cost historical analysis. Aggregating mispriced legs improves on trading pairs, while depth and consistency add a smaller gain. An earnings-surprise filter further reduces fades of moves that may reflect real information; a tested volume feature added little to the live portfolio. The evidence is presented without implementation details and includes cost-free comparisons, so live results may differ. Pair universe quality remains the central dependency.
Key ideas
- Pair selection based on realized reversion and convergence consistency is the foundation of the strategy.
- Triangulation converts pair signals into ticker-level views that can be combined into a more capital-efficient portfolio.
- Aggregation should reflect pair quality, signal depth, and agreement across pairs.
- Trading extreme ticker signals can reduce noise compared with acting on every name.
- An earnings-surprise filter can help avoid fading price moves driven by new information.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.