PlayCaller's MLB coverage splits cleanly into two things developers usually want separately: a fast-moving injury/news signal feed, and two validated models covering batters and starting pitchers.
Injury and beat reporter signals
GET /v1/intelligence/news and GET /v1/intelligence/injuries cover all three of PlayCaller's intelligence sports — NFL, MLB, and NBA — through the same two endpoints, filterable by sport, severity, team, or player. The feed refreshes every 2 hours from beat-reporter sources, not a generic RSS scrape, and each item includes a severity tag and source attribution.
What's validated: batter confidence and starting-pitcher trend
MLB batter confidence took several attempts to get right. Early versions, scored against a field-wide rank, capped out around an AUC-ROC of 0.58, well short of the internal promotion bar. The version that shipped changed the label to self-relative (does this player beat his own trailing baseline) and added park factors and platoon splits, reaching AUC-ROC 0.625 with a 26.98-percentage-point quartile lift. Starting pitchers get a separate model: a self-relative rolling-form read (today's start vs. the pitcher's own trailing baseline) that clears AUC-ROC 0.636-0.658 across 3 walk-forward folds. That one ships as a qualitative trend read, Trending Up, Steady, or Trending Down, rather than a percentage, because the underlying score wasn't built to be displayed as one.
The rest of the raw data
MLB also has the deepest raw-data footprint of any sport in the API — 9 endpoints covering players, teams, rosters, games, standings, and fantasy scoring leaders/categories.
Ready to build? Start a free 14-day Developer Sandbox — no credit card, API key in under a minute.