Multi-Factor Portfolio Selection for Statistical Arbitrage

Summary

The paper studies cointegration-based statistical arbitrage and compares ways to find candidate equity portfolios using multiple factors. It evaluates K-means clustering, graphical lasso across the equity pool, and a hybrid that combines the two. The reported results favor clustering over graphical lasso alone, with the hybrid performing better still.

The authors also test whether updating portfolio candidates once during the trading period improves results. In this study, adaptation does not outperform the strategy without relearning. The findings pass a statistical arbitrage test at a statistically significant level and are validated using a separate dataset for formation and trading periods. The summary does not specify the factors, portfolio construction details, transaction costs, or performance measures, so it does not establish how the approaches would compare in other markets or under different implementation assumptions.

Key ideas

• The study uses multiple factors to identify candidate portfolios for cointegration-based statistical arbitrage.
• It compares K-means clustering, graphical lasso, and a hybrid selection approach.
• Clustering performs better on average than applying graphical lasso to the full equity pool.
• Combining clustering with graphical lasso yields stronger reported results than either method alone.
• Recomputing candidate portfolios once during trading does not improve results in this study.
• The reported results pass a statistical arbitrage test and are checked on a separate dataset.

Tags

# A Multi-factor Adaptive Statistical Arbitrage Model


# A Multi-factor Adaptive Statistical Arbitrage Model









This paper examines the implementation of a statistical arbitrage trading strategy based on co-integration relationships where we discover candidate portfolios using multiple factors rather than just price data. The portfolio selection methodologies include K-means clustering, graphical lasso and a combination of the two. Our results show that clustering appears to yield better candidate portfolios on average than naively using graphical lasso over the entire equity pool. A hybrid approach of using the combination of graphical lasso and clustering yields better results still. We also examine the effects of an adaptive approach during the trading period, by re-computing potential portfolios once to account for change in relationships with passage of time. However, the adaptive approach does not produce better results than the one without re-learning. Our results managed to pass the test for the presence of statistical arbitrage test at a statistically significant level. Additionally we were able to validate our findings over a separate dataset for formation and trading periods.

Shown in full with attribution under the source's licence. Licence: abstract CC0

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.