Leakage‑aware backtesting framework for financial metric forecasting
It provides a reproducible pipeline that pulls SEC filings, yfinance prices, and FRED macro data, then applies LightGBM with a simplex meta‑learner under strict leakage controls. The backtester enforces point‑in‑time splits and includes regression tests that catch common leakage pitfalls. Designed for quantitative analysts and researchers needing robust, open‑source financial forecasting without proprietary data. Compared to generic ML pipelines, it embeds domain‑specific leakage guards and transparent evaluation, making results trustworthy.
View on GitHub →ritvikmaini/financial-metric-forecasting