Seasonal Trading in Commodity Futures: Evidence from Regression and Singular Spectrum Signals

RESUMO

Commodity futures are shaped by harvest cycles, weather shocks, storage conditions, and seasonal demand, but whether recurring patterns yield robust out-of-sample trading profits remains unclear. This article compares dummy-variable regression, Singular Spectrum Analysis, and low-rank SSA within a unified trading framework using monthly delivery-avoidance returns for 15 liquid commodity futures from 2016 to 2024 with transaction costs and Maximum Entropy Bootstrap assessment. Results show volatility normalisation reduces average DVR performance while weakening SSA-based portfolios, with no model establishing robust outperformance across market subperiods.