The result is simulated—and replayable.

What decides a game

Each plate appearance is sampled from fictional batter and pitcher rates through a seeded random stream. The engine then applies versioned base-running and out rules, including contextual singles and doubles, sacrifice flies, productive outs, and eligible double plays.

Natural-language recaps may describe recorded plays. They never choose an outcome, statistic, strategy recommendation, or winner.

Replay authority

The ordered play-by-play event log is authoritative. Scores, box scores, standings, recaps, championships, and manager history are projections of those events.

Every game stores its seed, simulation version, and fictional player-universe version. Older games retain their original rules and are not recalculated when a newer model ships.

What the model includes

The current alpha operates at the plate-appearance level with nine active hitters, a three-starter rotation, and one reliever. It models the core hit, walk, strikeout, hit-by-pitch, home-run, out, advancement, double-play, and sacrifice-fly paths needed for credible short-season play.

What remains outside the model

Errors, unearned runs, fielder's choices, steals, pickoffs, pitch-by-pitch sequences, active park and platoon effects, pitcher fatigue, and a larger bullpen are not represented in this version. Sprint does not claim to reproduce every professional-baseball mechanism.

Calibration evidence

The run environment and event frequencies are tested across thousands of deterministic games and a separate 20,000-game holdout. Public professional-baseball team aggregates are used only as a credibility reference for broad rates.

No professional player name, club identity, logo, photo, uniform, or row-level professional-player record appears in Sprint's fictional universe or runtime.

Fictional player data

The inaugural player pool comes from a versioned, anonymous, blended offline statistical artifact. Source names, identifiers, and clubs were removed; fictional profiles were independently tested and stored locally before use.

Public scouting lines are evidence for drafting, not a promise of a future result. Hidden exact ratings and matchup probabilities are not published or exported.