Early Beta — Research Tool
Bullpen IQ
Reliever fatigue, availability, and late-inning run-risk research for props and game markets.
Not official picks. Research data only.
How to Use This Data
Props
High-fatigue bullpens mean weaker late-inning arms enter the game. Look for hitter props (hits, total bases) in games where the opposing bullpen scores high on fatigue and low on quality.
Game Totals
Fatigued bullpens drive late-inning run scoring. When both bullpens show elevated fatigue and risk, the full-game over becomes more attractive relative to the market line.
Team Totals
Opponent team total gets a boost when facing fatigued relief corps. If one bullpen is exhausted while the other is fresh, the team facing the tired pen has late-inning upside.
F5 vs Full Game
Big gap between starter quality and bullpen fatigue signals that the first five innings may play differently from the full game. Use this to compare F5 lines vs full-game lines for value.
Related Tools
Game Markets
Poisson-compound model probabilities across 7 market types.
Team Totals
Per-team run projections for SGP and team total analysis.
Game Totals
Combined run total probabilities against posted over/under lines.
F5 Markets
First 5 innings starter-vs-starter before the bullpen enters.
SlateIQ
Props research board with line hit rates, model edge, and player drill-downs.
Props Board
Full props slate with lines, projections, and model signals.
Methodology
Scoring
Fatigue scores weight recent usage (appearances, pitch counts, back-to-back outings) over a rolling 3-day window. Quality scores reflect recent reliever performance (ERA, K rate, walk rate). Risk is a composite of high fatigue and low quality, normalized to 0-100.
Data Sources
Boxscores from the last 3 days, active MLB rosters, daily lineup data, and the outs_recorded model (full_model variant) for per-starter projected outs. Reliever roles are assigned by team-level usage rank — the most-used short-outing arm is tagged closer, the next is setup, two-plus-IP arms are tagged long relief, and everyone else is middle.
Limitations
Early-season sample sizes are small. No pitch-type data or formal leverage index is included. Role assignments (closer, setup, middle) are still heuristic — they use rank instead of bullpen depth charts. K/9 and BB/9 are computed only when the reliever has at least 1.0 IP of recent work; otherwise they show "--". Data refreshes during the day but may lag roster moves or late scratches.