Accuracy
How our projections compare to what actually happened, week by week, against the raw Sleeper baseline.
Rebuilding past weeks…
How our projections compare to what actually happened, week by week, against the raw Sleeper baseline.
Rebuilding past weeks…
| Week | Players | Model miss | Baseline miss | Δ | In range | Bias |
|---|---|---|---|---|---|---|
| Week 1 | 260 | 5.55 | 5.55 | −0.00 | 67% | -0.10 |
| Week 2 | 255 | 5.37 | 5.26 | +0.11 | 73% | -0.26 |
| Week 3 | 260 | 5.40 | 5.23 | +0.18 | 70% | -0.02 |
| Week 4 | 259 | 4.96 | 4.95 | +0.01 | 71% | +0.07 |
Players projected 5+ points who played. Historical projections rebuild the model using only data available before each week (closing nflverse lines; injury status not available historically). Miss = mean absolute error. Bias > 0 means players outscored projections.
| Week 5 | 240 | 5.42 | 5.53 | −0.11 | 68% | +0.17 |
| Week 6 | 221 | 5.31 | 5.12 | +0.19 | 72% | -0.37 |
| Week 7 | 233 | 5.12 | 5.16 | −0.04 | 78% | -0.51 |
| Week 8 | 224 | 5.48 | 5.37 | +0.10 | 75% | -0.37 |
| Week 9 | 238 | 5.11 | 5.03 | +0.08 | 81% | -0.62 |