WAR Index Methodology
The WAR Index is the settlement value for every contract on the platform. It is designed for one audience above all others: a counterparty's lawyer. Current version: 1.0.0.
Definition
For a player and MLB regular season:
WARIndex(player, season) = bWAR_batting + bWAR_pitching— Baseball-Reference Wins Above Replacement, batting and pitching, summed across all team stints in the season. Multi-period contracts sum the seasonal values. Three properties follow by construction:
- Two-way players count fully. Ohtani's pitching and hitting both accrue — as did Babe Ruth's in our backtests.
- Trades are a non-event. Stints sum; the index doesn't care what uniform the production came in.
- Negative seasons are real. WAR goes negative for below-replacement play and the index does not floor it. Floors are a product feature (collars), never an index feature.
Why bWAR only, for now
Version 1.0.0 settles exclusively on Baseball-Reference WAR, retrieved programmatically from B-Ref's published WAR data files. A blended index (bWAR + FanGraphs fWAR) is architecturally supported but deliberately disabled: FanGraphs does not permit unlicensed programmatic access, and a settlement index must not depend on scraping around an unwilling data provider. Activating the blend under a licensed feed is a major version change with its own changelog entry — no open contract is ever affected.
What the index is not
No era adjustment, no park factors, no aging curves, no projections. Realized WAR in, index value out. Forward-looking modeling — where to set ladder strikes, how to price a swap's fixed leg — lives in a separate pricing layer that never touches settlement. A trader may disagree with our projections; they can never disagree with settlement.
Settlement mechanics
- Settlement date: T+14 calendar days after the final day of the covered regular season.
- Settlement data: B-Ref WAR files as published on the settlement date, archived with a content hash as the settlement record.
- Corrections: source revisions after settlement never reopen a settled contract.
- Version pinning: every contract names its index version; methodology changes are never retroactive.
- Provisional values: in-season readings are informational only.
Backtest evidence
We reconstructed the index season-by-season over ten Hall-of-Fame-caliber careers spanning 1914–2010 — 215 player-seasons:
- Career index totals reproduce publicly known career bWAR exactly: Ruth 182.6, Bonds 162.8, Mays 156.2, Aaron 143.2.
- For 9 of 10 players, an MVP or Cy Young season falls within their top-3 index seasons.
- All 215 season-over-season transitions pass continuity bounds — no data discontinuities.

The career arcs are historically faithful: Ted Williams's WWII and Korea service gaps, Mays's 1952–53 military years, and Bonds's late-career 2001–04 peak are all visible in the reconstructed index.

Data sufficiency tiers
| Tier | Meaning | Contract implication |
|---|---|---|
| Established | ≥ 502 career PA or ≥ 30 pitching G | Full instrument menu |
| Rookie partial season | Debuted, below thresholds | Prominent sample-size disclosure |
| Pending MLB debut | No MLB record | No instruments listed |
Versioning
Semantic versioning with a public changelog. Patch = data refresh; minor = optional addition with default behavior unchanged; major = any change to the settlement formula or source blend. The full specification, changelog, and backtest code are version-controlled and available to partners and counterparties.