Game Win Probability
Rules v1 liveConservative probabilities from transparent team, starter, and availability rules.
Prediction Intelligence
Explainable MLB game forecasts using team form, expected starters, injury context, and market-aware comparisons.
Model tracking is still early. Accuracy metrics become more meaningful as more games are scored.
Predictions Scored
1124
All tracked model versions
Overall Accuracy
56%
Rules v1: 56%
High Confidence Accuracy
57%
794 scored games
Last Scored Game
Sep 27, 2026
Latest completed prediction
Rules-based v1 forecasts with deliberately conservative probabilities and concise, deterministic reasoning.
Check back soon as the rules-based prediction pipeline processes the next slate of MLB games.
Completed games scored against the forecast stored for each model version.
Performance grouped by the confidence assigned before the game.
Low
53%
177 scored games
Medium
53%
153 scored games
High
57%
794 scored games
| Date | Game | Predicted Winner | Actual Winner | Result | Confidence | Model |
|---|---|---|---|---|---|---|
| Sep 27, 2026 | HOU at ATH | HOU | HOU | Correct | High | rules_based_v1 v0.2.0 |
| Sep 27, 2026 | CHC at BOS | CHC | CHC | Correct | Low | rules_based_v1 v0.2.0 |
| Sep 27, 2026 | COL at CWS | CWS | CWS | Correct | High | rules_based_v1 v0.2.0 |
| Sep 27, 2026 | PIT at DET | DET | PIT | Incorrect | High | rules_based_v1 v0.2.0 |
| Sep 27, 2026 | CLE at KC | CLE | KC | Incorrect | High | rules_based_v1 v0.2.0 |
| Sep 27, 2026 | ATL at MIA | ATL | MIA | Incorrect | Medium | rules_based_v1 v0.2.0 |
| Sep 27, 2026 | STL at MIL | MIL | MIL | Correct | High | rules_based_v1 v0.2.0 |
| Sep 27, 2026 | TEX at MIN | MIN | MIN | Correct | Low | rules_based_v1 v0.2.0 |
| Sep 27, 2026 | TB at PHI | TB | PHI | Incorrect | High | rules_based_v1 v0.2.0 |
| Sep 27, 2026 | ARI at SD | SD | SD | Correct | Low | rules_based_v1 v0.2.0 |
What is active in rules-based v1 and what remains on the model roadmap.
Conservative probabilities from transparent team, starter, and availability rules.
Expected run production for both clubs with transparent uncertainty ranges.
Probable starter ERA and WHIP contribute when both starters are available.
Active injury counts provide a small, capped availability adjustment.
AXScout probabilities are compared with the latest available moneyline.
Completed forecasts are scored by model version with confidence-level accuracy.
Team results against starter archetypes provide a small, sample-aware adjustment when available.
The warehouse-backed context used by the current prediction rules.