Fourth & Forecast
Loading results

MODEL ACCOUNTABILITY

Predictions versus reality.

Every result is measured against the immutable model and market record frozen at kickoff.

GRADED SAMPLECollecting predictions
Read the sample firstEarly results are directional. Calibration and accuracy become meaningful only with a larger sample.Jump to challenger performance ↓

BENCHMARKS

Model versus market

Winner accuracy and Brier score

CALIBRATION

Confidence versus outcomes

Predicted home win rate

WEEKLY TREND

Accuracy over the season

Fourth & Forecast vs. Kalshi consensus

MODEL LEARNING VIEWS

Where the forecast works—and where it misses

Scenario-level accuracy, Brier score, and margin error

SILENT CHALLENGER

Candidate model performance

Separate frozen predictions · never rewrites V3.2

RESULT LOG

Recently graded games

Right pickWrong pick
Game and gradeFinal resultFrozen model pickClosing market pickMargin miss