{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,6]],"date-time":"2026-10-06T15:52:10Z","timestamp":1791301930916,"version":"4.1.0"},"posted":{"date-parts":[[2026]]},"group-title":"SSRN","reference-count":0,"publisher":"Elsevier BV","license":[{"start":{"date-parts":[[2026,1,1]],"date-time":"2026-01-01T00:00:00Z","timestamp":1767225600000},"content-version":"unspecified","delay-in-days":0,"URL":"https:\/\/www.uspto.gov\/ip-policy\/copyright-policy\/copyright-basics"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"abstract":"<jats:p>We investigate the out-of-sample predictability of monthly end-of-month WTI crude oil futures returns in a high-dimensional setting. We propose a two-stage framework combining preliminary variable screening based on SIS-type screening procedures (SIRS and DC-SIS) with six penalized regression (Lasso, aLasso, Elastic Net, and SCAD) and machine learning (random forest and gradient boosting machines) estimators applied at horizons h in {1, 3, 6, 12} months over a rolling window of 118 months. Using a dataset of 80 predictors covering the period February 1990 to December 2025, we show that SIRS combined with the adaptive Lasso is the only approach to achieve statistically significant MSE reductions against both the AR(1) and the Zero-Change benchmarks at h = 1. The SIRS and DC-SIS procedures select fundamentally different predictor sets, where uncertainty variables dominate under SIRS, while DC-SIS retains commodity market and real activity variables. Variable importance exhibits a horizon-dependent rotation from uncertainty at short horizons to real activity at medium horizons and commodity fundamentals at long horizons.<\/jats:p>","DOI":"10.2139\/ssrn.7571022","type":"posted-content","created":{"date-parts":[[2026,10,6]],"date-time":"2026-10-06T14:54:21Z","timestamp":1791298461000},"source":"Crossref","is-referenced-by-count":0,"title":["Forecasting WTI crude oil futures returns in high dimension: Penalized regressions, machine learning, and the role of uncertainty"],"prefix":"10.2139","author":[{"ORCID":"https:\/\/orcid.org\/0000-0001-8967-7584","authenticated-orcid":true,"given":"Jean-Baptiste","family":"Bonnier","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Amelie","family":"Charles","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-2796-4873","authenticated-orcid":true,"given":"Olivier","family":"Darn\u00e9","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"78","container-title":[],"original-title":[],"deposited":{"date-parts":[[2026,10,6]],"date-time":"2026-10-06T14:54:21Z","timestamp":1791298461000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.ssrn.com\/abstract=7571022"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026]]},"references-count":0,"URL":"https:\/\/doi.org\/10.2139\/ssrn.7571022","relation":{},"subject":[],"published":{"date-parts":[[2026]]},"subtype":"preprint"}}