Prediction scores

Every prediction we made for a night whose setlist we now know, scored against what actually got played.

These numbers look low, and they should. Phish are deliberately unpredictable — they rotate songs on purpose and enjoy wrong-footing exactly this kind of model, and 100 would mean calling an entire setlist. Read Skill, not the raw score: it measures how much of the headroom above a baseline that just guesses the most-played songs the model actually captured. Beating that baseline at all is the achievement.

Scores shown here are ones already computed and cached. Scoring a night rebuilds a full baseline prediction to measure Skill against, so this page never computes them on the fly — “score it” does one date on request.