Challenge Run Value (cRV): quantifying strategic value in MLB's ABS challenge era
We introduce Challenge Run Value (cRV), a pitch-level metric for valuing Automated Ball-Strike (ABS) challenges in Major League Baseball. Unlike framing metrics, which measure physical receiving skill under human umpiring, cRV treats the challenge as a scarce strategic resource whose value depends on when it is spent, whether it succeeds, and which future options it forgoes. The model combines run-expectancy deltas from count- and plate-appearance-ending state changes with a time-decaying, scarcity-weighted penalty for failed challenges, calibrated against an empirically derived RE288 run-expectancy surface built from public Statcast data. A central methodological contribution is the valuation of terminal outcomes: challenges that convert a called strike three into ball four (or the reverse) end the plate appearance and must be priced from the resulting base/out state and runs scored rather than zeroed. We implement a reproducible five-module pipeline (ingestion, event isolation, alternate-state generation, RE288 mapping, and scoring) with game-clustered bootstrap uncertainty, leverage stratification, and sensitivity analysis. Applying the model to the 1,971 plate-appearance ending challenges recoverable from the 2026 regular season through August 12, we estimate a mean cRV of 0.159 runs per challenge (95% CI [0.147, 0.172]); catchers generate roughly two-and-a-half times the per-challenge value of batters (0.231 versus 0.088). Zeroing terminal states would instead drive the aggregate negative (-0.016), inverting the sign of the result. The metric gives teams a principled basis for deciding when to spend a scarce challenge and for separating high-leverage conversion skill from reckless early usage.
Authors
- Jacob Goldstein
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-08-13
- DOI
- https://doi.org/10.5281/zenodo.21925849
- Primary Topic
- Sports Analytics and Performance
- Type
- preprint