← Lab
Archived · 2024
NFL Forecasting Archive
Historical model comparison, calibration, and backtesting experiments.
What this project shows
A bounded implementation, not a universal claim.
Preserves evidence of forecasting and probability-quality work. The useful evidence is the combination of public source, explicit stack, and a limitation that stays close to the work.
Boundary
What it does not establish
Archived analysis; not betting advice and not a maintained recommendation system.