Reproduce “Hierarchical Graph Learning for Calendar Spread Strategies in Commodity Futures Markets” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Yoonsik Hong, Diego Klabjan

Researchers at Northwestern University developed a hierarchical graph learning framework to predict commodity futures prices for calendar spread strategies, explicitly modeling maturity-dependent interrelationships among contracts. This approach demonstrated higher risk-adjusted returns and substantially lower risk compared to traditional strategies and existing machine learning methods in empirical tests.

Key insights

  • Calendar Spread strategies offer superior risk-adjusted returns and significantly lower risk (variance, delta exposure) in commodity futures markets, a theoretical advantage supported by empirical conditions.
  • Explicitly incorporating maturity-dependent interrelationships among commodity futures contracts through a hierarchical graph structure substantially improves predictive accuracy and trading profitability for statistical arbitrage.
  • Both inter-commodity relationships (across different commodities at the same TTM) and intra-commodity relationships (across different TTMs within the same commodity) are critical for extracting robust predictive information for trading strategies.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.