Monolith Research
In plain terms
Cold weather drives heating demand, and heating demand draws down natural-gas storage. This model turns public weather data into a forecast of the weekly US gas-storage figure before the official number is released.
How it works
Temperature readings across the country are converted into population-weighted heating- and cooling-degree-days, then mapped to the change in gas inventory by a walk-forward-fitted relationship — so the model is only ever judged on data it has not seen.
What it’s tested against
Evaluated walk-forward out-of-sample, with a placebo (the same machinery on scrambled inputs) to confirm it reads signal rather than fitting noise. It is a genuine nowcast, not a deployed trade — the public weather it uses is largely already in the price, which is exactly the distinction the gate is built to surface.
Data
Open-Meteo temperature data; EIA weekly natural-gas storage.
Funding-grade research artefact — a clean forecasting result, kept as evidence rather than a live trade.