Tool 01
Bayesian Draft Updater
Update P(A-grade) with each observed game line. Binomial likelihood, sequential posterior. Sensitivity table shows how much the conclusion depends on the rate assumptions.
Posterior
Prior P(A)
30.0%
Final P(A)
39.1%
+9.1pp vs prior
Cumulative LR
1.50
Total AB: 4
Posterior trajectory
Per-game updates
| # | AB | XBH | LR | P(A) before | P(A) after |
|---|---|---|---|---|---|
| 1 | 4 | 2 | 1.50 | 30.0% | 39.1% |
Sensitivity to rate assumptions
| Scenario | Rate A | Rate non-A | Final P(A) | Δ vs base |
|---|---|---|---|---|
| Baseline | 10.0% | 8.0% | 39.1% | — |
| +2pp A | 12.0% | 8.0% | 46.9% | +7.8pp |
| −2pp A | 8.0% | 8.0% | 30.0% | -9.1pp |
| +5pp non-A | 10.0% | 13.0% | 21.3% | -17.7pp |
| −5pp non-A | 10.0% | 3.0% | 80.4% | +41.3pp |
Limitations
- Search is name-only (no direct MLBAM ID entry). Promoted players show current affiliated level only — no multi-level aggregation.
- Assumes XBH rates are stable per grade. They aren't — handle with care.
- Binomial assumes independent ABs; reality has streaks, lineup context, etc.
- Rate A and rate non-A are within 3pp. The model has weak power to discriminate.
- Total AB is small; final posterior is sensitive to small input changes.
Analysis