All projects

feedstock

A data analytics and forecasting platform designed to help industrial businesses understand commodity price fluctuations and their impact on production costs. The system processes historical market data for materials such as copper, aluminum, gold, platinum, and energy, using 30,302 daily observations across 12 financial instruments. Built with Python, TimescaleDB, and React, it transforms raw market information into normalized datasets, historical trends, and predictive models. Forecasting performance is evaluated through walk-forward backtesting against baseline models, providing transparent and reproducible out-of-sample error measurements. The platform also validates instrument metadata against original data sources to prevent incorrect financial references from affecting analytical results. Its focus is decision support and cost-risk analysis rather than automated trading.

Python · TimescaleDB · Forecasting · Backtesting · React