Post by Fabian Diaz Gallardo

Five months ago, this started with a simple question: Does crude oil really drive energy stocks the way everyone assumes? It ended up becoming a three-phase research project that challenged far more than my econometric skills. Coming from Petroleum Engineering and transitioning into Data Science, I expected the hardest part to be learning new models. It wasn't. The hardest part was accepting when the data contradicted intuition. Over the past few months I built a fully reproducible framework combining petroleum engineering, econometrics, quantitative finance, and machine learning to analyze 25 years of energy market data. The biggest lesson wasn't a statistical result. It was realizing that negative findings can be just as valuable as positive ones. Sometimes proving that sophisticated models cannot outperform a naive benchmark tells you more about market efficiency than about the models themselves. The article I'm sharing today brings together the complete journey—from the original question to the final conclusions—and explains why these methods matter for anyone working in energy markets, commodities, or quantitative finance. A few questions for you: • Is the energy sector really one sector? • Can long-run equilibrium exist even when returns remain unpredictable? • What should investors trust more: intuition or statistical evidence? I'd love to hear your thoughts after reading the article. cc: Daniel Yergin, Helima Croft, Anna Mikulska, Morgan Stanley, Deloitte Insights, U.S. Energy Information Administration #DataScience #Econometrics #QuantitativeFinance #EnergyEconomics #PetroleumEngineering #OilAndGas #Python #MachineLearning #CommodityMarkets