Model
Report Card
The live track record of the published Model's Picks — graded the morning after each slate. Records assume a flat 1-unit stake at the price we recorded. This is how the actual picksdid; the model's overall calibration (every projection, not just the bets) is below.
Loading track record…
Every projection — not just the ones we bet
Projection Calibration
How well our projections matched reality, scored each day against final stats — across all projected players, an unbiased read on the model itself. Lower Brier is better; skill is how much we beat the naive always-predict-the-average baseline. Calibration shows whether a projected probability matches the rate things actually happened.
Loading accuracy…