SuperAPI AI Directory
Partner with us中文AI newsFavorites
Skillsmlb-statcast-player-comparison

mlb-statcast-player-comparison

Data & analytics

View repository

Updated 10-15 20:29

About this skill

MLB Statcast player comparison

Access and scope

Set CRAWLORA_API_KEY to your Crawlora key. Prefer the listed Crawlora MCP tools when connected; otherwise use the bundled scripts/crawlora.sh, which sends x-api-key to https://api.crawlora.net/api/v1. Read reference/endpoints.md for exact methods and parameters. Check application code as well as HTTP status; payloads are inside data. Stop on authentication errors, back off on 429, and retry a transient upstream failure once. Bound calls/pages by the requested scope and credit budget. Retain source IDs, URLs, source dates, and observation/crawl times separately.

Build a role-specific performance comparison for selected players or a declared cohort. Establish season/date window, batter versus pitcher/fielding role, competition scope, qualification, and metrics before retrieving leaderboards.

Discover boards, identities, and qualification

  • Use mlb_discovery for board-specific selectors/fields and MLB search/team/player routes for identity. Keep numeric player/team IDs distinct from a leaderboard's local row/entity keys. A matching display name does not resolve the player.
  • Each board has its own season coverage, role values, team parameter, thresholds, split/grouping, and sorting fields. Do not copy team_id into a board expecting team, or reuse batter fields in a pitcher table. Request only the relevant expected/percentile/arsenal/batted-ball/movement/rolling views.
  • Qualification depends on the board: plate appearances, balls in play, pitches, or other events differ. minimum=q means the source's qualification rule, not the same sample size for all boards. Preserve the selected thresholds, event count/denominator, role, year, and returned league-average context.
  • Some boards return league-wide tables with local sorting/pagination rather than an arbitrary player_id filter. Match selected players in actual returned rows; do not invent unsupported input fields or assume a top-N page includes every eligible player. Keep total, offset, limit, and missing-player reasons.
scripts/crawlora.sh /mlb/discovery
scripts/crawlora.sh /mlb/statcast-expected type=batter year=2025 filter_type=bip minimum=q limit=10
scripts/crawlora.sh /mlb/statcast-percentile type=batter year=2025 limit=10
# Resolve player identity before joining rows from different boards.

Compare metrics on their actual bases

Preserve raw metric labels, units, direction, and calculation scope. Expected statistics are provider models, not realised outcomes or promised future results. A percentile is a relative ranking in its source cohort, not a success probability; its year/type denominator can differ from another board. Do not average percentiles or convert them into a universal talent score without an explicit justified rubric.

Pitch usage, whiff rates, run values, velocity/spin/movement, and batted-ball metrics have different units and favorable directions. Compare equivalent pitch families, role, event windows, and board definitions; do not merge a pitch-specific row into a whole-player mean without its actual weights. Rolling windows are plate-appearance samples rather than equal calendar time; do not claim a calendar trend from a mislabeled window. Year-to-year views need matched definitions and coverage, with rule/measurement changes retained as limits.

Return identity/scope and metric tables, board/qualification evidence, event counts, league context, supplied-versus-calculated values, and unknowns. Disclose weights for a requested ranking and keep small samples visible. These observations do not establish injury, future performance, transfer value, or a complete scouting assessment; no betting or account action is part of a comparison request.