Decoding Systematic Relative Investing: A Pairs Approach

Abstract

We propose a novel theory that brings to light three fundamental performance drivers of zero-cost systematic investment strategies:

(1) high (positive) own-asset signal-return predictability;

(2) low (or negative) cross-asset signal correlation; and

(3) low (or negative) cross-asset signal-return predictability.

We develop these insights in the context of long-short pair strategies used as portfolio building blocks. We test our approach empirically using momentum signals for major asset classes, though our method can generalize to any signal. Our investable pairs-based portfolio harvests over double the average returns of a conventional rank-based portfolio over the last 20 years.

Keywords: cross-sectional strategies, trading strategies, long-short strategies, relative value, momentum, portfolio construction, alpha, outperformance, anomalies, market timing, asset pricing, zero-cost portfolios, pair trades

JEL Classification: G12, G13