MLB Prediction Model v6 — 14 signals
Weighted composite model combining Vegas odds, pitcher matchups, team strength, and advanced Statcast metrics. Predictions run hourly. Trades execute on Kalshi across 8 market types.
Primary signal — 60% accuracy when available
+4.1pt gap between correct vs wrong picks
Early season noise — improves with sample
Added v6 — now stored for backtesting
MLB ~54% home advantage
Team batting — needs more games
IL status from ESPN injury feed
+4.2pt gap — strong team strength signal
Season record with Bayesian prior
Opening vs current moneyline
Bayesian smoothed matchup history
Temp, wind, humidity — dome parks neutral
Win/loss momentum (capped at 10)
Runs scored vs allowed
Confidence Tiers & Bet Sizing
Kalshi Market Coverage
* Ready = code deployed, awaiting Kalshi market liquidity
Model Calibration (since Mar 27)
0
Resolved
0%
Win Rate
14
Signals
8
Market Types