NFL statistics for AI assistants

Choose the right statistical view

nfl_player_stats returns season-level passing, rushing, or receiving production. nfl_compare compares two players or teams in one call. nfl_team_stats covers offense, defense, and situational efficiency; nfl_search_plays finds and aggregates plays.

Use nfl_game_log for week-by-week player results rather than treating season totals as weekly stats. For rosters, schedules, injury reports, and snaps, use the corresponding focused tools. Advanced coverage splits and pressure analysis have their own coverage windows.

• Player stats
• Player and team comparisons
• Play-by-play search
• Weekly game logs

Specify the sample before comparing

Include the season, regular season or playoffs, player/team, stat category, and any week filters. For non-quarterbacks, request receiving or rushing rather than relying on the passing default.

EPA means expected points added: how a play changes the offense's expected scoring value. Success rate measures the share of successful plays under the tool's definition. Ask the assistant to explain denominators, sample size, and filters before drawing conclusions.

Resolve names instead of guessing

Full player names and supported short forms are accepted. If a name is ambiguous, choose a returned candidate and pass the team or GSIS identity supported by that tool. Two people with the same name should never be silently merged.

No rows can mean an unavailable season, unmatched name, or genuinely empty filter. Read the hint. A missing row is not automatically a zero.

Inspect datasets for custom analysis

Use nfl_catalog to discover loaded tables, season coverage, and refresh metadata. Pro includes nfl_schema for live column inspection and nfl_query for bounded, read-only SQL when no structured tool covers the question. Do not invent column names or assume every dataset spans the same seasons.

• Dataset coverage
• Dataset catalog