The Lead-lag relations in the Commodity Futures Returns: A Machine Learning Approach
55 Pages Posted: 8 Mar 2020 Last revised: 9 Nov 2022
Date Written: January 11, 2022
Abstract
This paper uses machine learning tools to study the lead-lag relations in commodity futures returns. We use LASSO to select the predictors because the number of predictors is large relative to the number of observations. We find significant full-sample and out-of-sample predictability. In the full sample, we find that LASSO can identify a sparse set of predictors that either come from economically linked commodities or are likely driven by excessive speculative trading. The out-of-sample forecasts based on LASSO generate statistically and economically large gains. When we use more complex machine learning models such as regression trees and neural networks to forecast commodity futures returns, the out-of-sample performance is worse than LASSO portfolios, suggesting that nonlinearities and interactions do not appear substantial in the data. Finally, we find that index trading due to financialization drives the excess comovement among commodity futures.
Keywords: Commodity Futures, LASSO, Machine Learning, Predictability, Serial Dependence, Financialization, Comovement
JEL Classification: C22, C58, G11, G12, G14
Suggested Citation: Suggested Citation