MLB Start Score

How the MLB Start Score works

The formula. Each batter's projected fantasy points is estimated as: fp_per_game × Matchup × Park × Availability. For starting pitchers it is fp_per_start × Matchup × Park; for relievers, fp_per_game × Park. Each projected value is then converted to a 0–100 score by percentile rank within the position bucket — so an 85 means the player ranks in roughly the top 15% of scorable players at their position for that game.

Fantasy points formula (Sleeper MLB default). Batters: 1B ×3, 2B ×5, 3B ×8, HR ×10, RBI ×3.5, R ×2, SB ×4, BB ×2, HBP ×2, SO ×(−1). Starting Pitchers: IP ×3, W ×4, ER ×(−2), SO ×1, H ×(−1), BB ×(−1), HBP ×(−1). Relievers add SV ×5 and HLD ×2. This is a labeled assumption; platform-specific formats are future toggles.

Form (fp/game) is the player's season-to-date fantasy point rate — total season fantasy points divided by games played. This is the heaviest term: a player's production rate is the most stable signal available from public box-score data. Minimum 10 plate appearances (batters) or 3.0 IP (pitchers) required for a score.

Matchup is a bounded multiplier (×0.80 – ×1.20). For batters, it is the opposing starting pitcher's WHIP vs. the league-average SP WHIP — a higher opponent WHIP means more baserunner opportunities, which is better for batters; with no probable starter named it is ×1.00. For starting pitchers it is meant to be the opposing lineup's quality, but our source does not supply wRC+, so today it is ×1.00 for every starter and a starter's score moves only with form and park. Relievers receive no matchup adjustment (park factor only) — their game situation is too variable to model with an advance signal.

Park factor is the home park's run factor from FanGraphs 5-year regressed park factors (100 = neutral). Batter park multiplier: run_factor / 100. SP park multiplier: 100 / run_factor (inverse — a hitter-friendly park hurts pitchers). Updated annually.

A correction, 19 September 2026. Until this date the scores averaged the matchup and park multipliers instead of multiplying them as described above — so each counted for half — and relievers got no park factor. The code now does what this page says. The typical score moved by about two points; games at Coors Field and extreme pitcher matchups moved most.

Availability. Only players on the active 26-man roster receive a score — players on the IL or not rostered are excluded automatically. The ETL checks active rosters nightly via the MLB Stats API. Always verify injury and lineup status on game day; roster status can change after scores are computed.

Position buckets. Players are grouped into eight buckets: C, 1B, 2B, SS, 3B, OF (batters) and SP, RP (pitchers). Scores are percentile-ranked within each bucket — an 80 at catcher means top-20% of scorable catchers, not top-20% of all players.

Data. Player stats, schedule, and active rosters from the MLB Stats API (statsapi.mlb.com, official, no auth required). Park factors from FanGraphs (2025 regressed values, updated annually). Scores are pre-computed nightly by the ETL and cached in Supabase; this page never calls either API directly. No machine-learning model is used — every weight and assumption is published here.