01Climate-Factor Quant Research

Weather itself doesn't trade, but coffee, soybeans and corn do — the impact of El Niño, droughts and hurricanes on growing regions ends up written into prices.

The hard part isn't finding correlation, it's making it auditable: climate data comes from several public sources with different definitions, resolutions and revision histories, so factors computed straight from it drift with every source update.

Engineered climate phenomena into backtestable factors (temperature anomaly, ENSO, growing degree days, precipitation) and built the full chain from data ingestion to paper trading.

I set the acceptance criteria and the data governance rules; the AI team executed. Every source's fetch time, version and missing-value handling is persisted, so any backtest traces back to the exact snapshot it used. A factor without a statable physical mechanism never enters the pool.

An institutional-grade systematic research report; an automated paper-trading run every month; every research figure recomputable.