A WAR reconstruction engine for the integrated record. The methodology that makes the comparison makeable, documented openly, the data open, the assumptions auditable.
Gary Ashwill and Kevin Johnson built the Seamheads Negro Leagues Database over fifteen years. They normalized statistics across leagues, parks, and opponent strength. They produced per-season WAR for every documented Negro Leagues player.
Baseball Reference incorporated their work in 2020. FanGraphs incorporated it in 2023. Major League Baseball officially recognized it in 2024.
But the comparison question was never fully answered. Josh Gibson's career WAR is 60.2 (Seamheads). Ty Cobb's is 151.4 (Baseball Reference). The first reading says Cobb was roughly 2.5 times more valuable. The reading is wrong, and the reason it is wrong is that Cobb played more than three times as many league games as Gibson. The seasons are not the same unit.
The standard Hall of Fame methodology, JAWS, averages career WAR with seven-year peak WAR. It was built for the MLB record. It does not handle the structural conditions of Negro Leagues play. Applied without adjustment, it understates Negro Leagues players by exactly the seasonal-length disparity their leagues operated under.
This chapter builds the alternative. Rate-based career and peak metrics. Position-by-position integrated leaderboards. The methodology open, the data open, the assumptions auditable.
Every qualified player in the integrated record. Sorted by Rate JAWS by default. The integration is the point: Negro Leagues and MLB players ranked by the same methodology, in the same table, using the same units.
| # ▲ | Player ▲ | Pos ▲ | Era ▲ | Career WAR ▲ | WAR/600 PA ▲ | Peak Rate ▲ | Std JAWS ▲ | Rate JAWS ▼ | HOF ▲ |
|---|
Each point is a player. Above the diagonal: players who score higher on rate JAWS than standard JAWS. Predominantly Negro Leagues players whose career totals were suppressed by short seasons. Below the diagonal: players whose career totals reflect longevity more than peak production.
Every WAR figure on the platform is a credible interval, not a point estimate. The width of the interval reflects the underlying data completeness. Players from sparsely-documented eras have wider intervals. The platform does not paper over uncertainty. It surfaces it.
A UMAP dimensionality reduction over the multi-component WAR space. Players cluster by similarity in value profile, not by ranking. Gibson clusters with Piazza and Bench. Charleston clusters with Mays. The clustering is not narrative assertion. It is multi-dimensional similarity.
Adjust the methodology. See the leaderboard shift. The platform does not hide its choices. It exposes them. Every parameter is documented, every alternative is auditable.
Six models. Each documented with inputs, outputs, confidence labels, and the reasoning for why this approach and not an alternative. Click any card to expand the full methodology.