Accuracy
Predicted vs Actual Ks
1144 pitcher-starts evaluated
MAE · Placed Bets
1.94
Ks · n=2665
MAE · All Predictions
1.78
Ks · n=15,533
Bias · Placed Bets
+0.29
neutral
Bias · All Predictions
+0.05
neutral
Error Distribution
How far off are predictions, in Ks?
Green bar = perfect predictions · symmetric histogram = no bias · long tails = the model occasionally misses badly
| Placed Bets n=1,295 | All Predictions n=15,533 | |
|---|---|---|
| Hit-rate bands · how often is the rounded prediction within X K? | ||
| Exact (±0 K) | 16.4% | 17.7% |
| Within ±1 K | 46.4% | 49.7% |
| Within ±2 K | 70.0% | 74.0% |
| Within ±3 K | 84.4% | 88.8% |
| > 3 K off (catastrophic) | 15.6% | 11.2% |
| CI coverage · % of actual outcomes inside the predicted interval | ||
| 80% CItarget 80% | 83.6% | 88.0% |
| 90% CItarget 90% | 91.9% | 94.5% |
| 95% CItarget 95% | 96.5% | 97.6% |
| 99% CItarget 99% | 99.7% | 99.7% |
Conformally calibrated intervals
The model is slightly conservative: nominal NB 80% catches 88.0% of actuals due to discrete-quantile overshoot. Pitcher detail charts use tighter calibrated percentiles that hit each target empirically.
80% CI
[10%, 90%]
[15.2%, 84.8%]
90% CI
[5%, 95%]
[8.4%, 91.6%]
95% CI
[2.5%, 97.5%]
[4.6%, 95.4%]
99% CI
[0.5%, 99.5%]
[1.2%, 98.8%]
MAE Over Time
Drift sentinel
Flat = consistent model · rising = retrain alarm
Scatter
One dot per pitcher-start
Dashed diagonal = perfect prediction. Hover any point to inspect.