Forecast Ledger (score your calls or they're just opinions)

A probabilistic forecast you never grade teaches you nothing. Superforecasting's core finding: forecasters improve only through scored feedback — Brier scores and calibration. This skill is that loop: every dated call made by superforecasting (or any market call) gets logged with its probability, resolved on its date, and scored.

When to use

• Log — right after emitting any forecast with a probability + a resolution date ("70% SOL tags $60 by Jul 1"). If it has a P and a by-when, it goes in the ledger.
• Resolve — on/after the horizon date, mark hit/miss against what actually happened.
• Score — to review calibration: are your 70%s actually right ~70% of the time? Which lens/source is sharpest?

Don't use for undated opinions or non-probabilistic takes — nothing to score.

The loop