Free during beta —to track favorites + alerts

THE ONE · FANTASY

Projection
accuracy

Every fantasy site sells you projections. We grade ours — out-of-sample, against what actually happened, and against the season-average baseline you'd use otherwise.

Across 16,564 graded player-weeks (20232025), our weekly projections rank players better than a season-to-date average on 4 of 4 positions.

QBn=1,991
Ranking (corr)
0.481oursvs 0.475 base
we win
Error (MAE)
6.62oursvs 6.41 base
baseline edges
RBn=4,266
Ranking (corr)
0.606oursvs 0.598 base
we win
Error (MAE)
4.74oursvs 4.75 base
we win
WRn=6,843
Ranking (corr)
0.545oursvs 0.535 base
we win
Error (MAE)
4.77oursvs 4.84 base
we win
TEn=3,464
Ranking (corr)
0.517oursvs 0.510 base
we win
Error (MAE)
3.64oursvs 3.80 base
we win

How this is graded. No leakage — every week is projected only from data available before that week, then compared to the realized PPR score. The baseline is each player's season-to-date average (the honest “what you knew going in”). Ranking (correlation with what actually happened) is what start/sit and trade calls hinge on; we show mean error too, even where the simple average is competitive. The board refreshes as the season adds games.