A1161
Title: Stationary but not profitable: A critical look at pairs trading
Authors: Davide Pandini - International Federation of Technical Analysts (IFTA) (Italy) [presenting]
Abstract: Pairs trading is a market-neutral statistical arbitrage strategy that targets mean reversion in spreads between economically related assets. Conventional implementations select pairs using distance measures or cointegration tests, yet much of the evidence is drawn from pre-2008 crisis sample periods, a restricted investment universe and optimistic assumptions about costs and parameter stability, leaving its effectiveness in modern equity and equity-commodity markets uncertain. A fully reproducible end-to-end workflow for spread construction, cointegration diagnostics and rule-based trading is developed, evaluated under a strict train/test protocol with explicit transaction-cost modeling on daily data from 2007 till 2025. Results indicate that equity and index spreads provide the most consistent net risk-adjusted performance, whereas commodity-linked spreads seldom exhibit cointegration after 2018 and typically underperform once costs are applied, in contrast with prior findings in the commodity pairs-trading literature. The evidence further shows that cointegration is informative but neither sufficient nor necessary for out-of-sample profitability: some in-sample cointegrated relations deteriorate following abrupt changes in market behavior, while certain non-cointegrated spreads still yield short-horizon convergence returns. The workflow and diagnostics delineate when spread trading adds value as a controlled relative-value and risk-management tool in recent markets.
