Model
WTI Crude Oil Signal
Predicts whether retail gas prices will be higher in one week by tracking WTI crude oil spot price trends and momentum via Alpha Vantage.
Decision surface
The model's buy probability across its two most influential features. Every point was computed by calling model.predict_proba on the actual .joblib. Drag to rotate.
What the model pays attention to
Approach
I used daily WTI crude oil spot prices from Alpha Vantage covering 1986-02-12 to 2026-03-25 to build a forward-looking binary classifier. My background in aerospace and defense procurement informed the feature design: the model tracks current WTI price, 7-day and 30-day rolling averages, 7-day price change, and a momentum indicator (normalized spread between short- and long-term averages). The label is 1 if WTI is higher 7 days out (buy gas today before prices rise) and 0 otherwise. I held out the most recent 20% of observations as a time-ordered test set to avoid lookahead bias. GradientBoostingClassifier outperformed Random Forest and Logistic Regression in cross-validation, achieving 0.53 test accuracy. The momentum feature was the strongest predictor, consistent with how crude-to-retail price transmission works: WTI trends tend to persist over the 1-2 week lag before they reach the pump.
Features
Built by
M.S. AI in Business candidate at Arizona State University's W. P. Carey School of Business (expected December 2026) with five years of supply chain and procurement experience in aerospace and defense. Previously managed $2.3M in avionic systems procurement and international defense contracts at MD Helicopters. Focused on the intersection of AI, predictive analytics, and supply chain operations.
Trained on
1986-02-12 to 2026-03-25