stathead-mcp

Remove custom projections/rankings overrides previously saved via import_excel, so analysis tools revert to StatHead's own model. Optionally clear just one kind. See also: get_metadata for season coverage, valid enums, and analytic caveats.

Export StatHead projections, rankings, or rookie rankings to a styled .xlsx workbook on the local disk — the same boards the website lets you download. Edit the highlighted columns in Excel/Sheets (Proj PPG for projections; My Rank for rankings/rookies), then feed the file back with import_excel so YOUR numbers drive every later analysis. Returns the absolute path of the written file. See also: get_metadata for season coverage, valid enums, and analytic caveats.

Get StatHead ADP: StatHead's blend of the draft markets (FantasyPros expert consensus, Sleeper draft ADP, FantasyFootballCalculator), each weighted by freshness and confidence, for players priced by at least two of them. Third-party rankings and ADP are inputs only: no single source's number is returned. Includes a disagreement spread (max minus min across sources) and the number of sources. Use for draft value analysis. See also: get_metadata for season coverage, valid enums, and analytic caveats.

Join preseason StatHead ADP (StatHead's blend of the archived Sleeper and FantasyFootballCalculator draft boards; players on both) to actual season-end fantasy production in one call — find draft-day values and busts without joining ADP and stats yourself. For a season, returns each drafted skill player's StatHead ADP alongside their actual PPR points, positional ADP rank, positional PPR finish (among drafted players), and value = adp_pos_rank − finish_pos_rank (positive = beat draft slot). Skill positions (QB/RB/WR/TE). Coverage 2020–present. No single source's ADP is shown. Note: value is a single-season residual — a relative bust/value ranking, not a predictive probability; average multiple seasons before relying on it. See also: get_metadata for season coverage, valid enums, and analytic caveats.

Get PFR advanced stats: passing (pressure rate, drop rate, bad throws), rushing (yards before/after contact, broken tackles), receiving (drops, YAC). Use for efficiency and process metrics beyond box score. See also: get_metadata for season coverage, valid enums, and analytic caveats.

Head-coach scheme + usage tendencies and a coach×player reunion lookup — so you don't have to hand-reconstruct what a coach does or how he used a player. Per coach (career + last-3-season averages): neutral-script pass rate, pass rate, plays/game (pace), shotgun rate, RZ TD rate, PPG, target concentration (HHI), and WR1 / RB / TE target share (the WR1's name per season). Pass a coach by name, or a team+season to resolve that team's coach. Add player_name to see that player's target share / PPG / games UNDER this coach (the reunion signal, e.g. A.J. Brown under Mike Vrabel), or player_name alone for the player's full coaching history. Coverage: 2016–present (head coach = nflverse credited game coach, a play-caller proxy). Note: this is descriptive data; it is not (yet) wired into the share-model projection. See also: get_metadata for season coverage, valid enums, and analytic caveats.

Get ESPN college QBR ratings for quarterback prospects. Includes total QBR, points added, EPA, and breakdown by pass/run/sack. Coverage: 2004–2020 only (historical; the upstream source is not updated for recent seasons — use get_college_stats for current college production). Use for evaluating college QB efficiency, comparing draft prospect QBs across classes. See also: get_metadata for season coverage, valid enums, and analytic caveats.

Get college football statistics for individual players OR a whole class/cohort. Counting stats (passing, rushing, receiving, tackles, sacks, INTs, etc.) by season. Source: CollegeFootballData (CFBD); coverage 2005–present. For one player, pass player_name. For a cohort (e.g. "all 2023 RBs"), omit player_name and pass season (optionally position/school) and sort_by to rank. Use for evaluating college production, dominator rating, market share analysis. See also: get_metadata for season coverage, valid enums, and analytic caveats.

Get NFL Combine athletic testing results. Includes 40-yard dash, bench press, vertical jump, broad jump, 3-cone drill, shuttle, height, weight, draft position. Use for athletic profile analysis, draft capital evaluation. See also: get_metadata for season coverage, valid enums, and analytic caveats.

Get NFL player contract details from OverTheCap. Includes total value, APY, guaranteed money, cap percentage, inflation-adjusted values. Use for salary analysis, team cap situations, player investment vs production. See also: get_metadata for season coverage, valid enums, and analytic caveats.

Get NFL team depth charts showing starter/backup designations. Shows position rank (1=starter, 2=backup, 3=third string). Use for projecting snap shares, identifying handcuffs, roster battles. See also: get_metadata for season coverage, valid enums, and analytic caveats.

One devy (college) player's card: his StatHead devy numbers (composite value and rank, value-model price, NFL career projection, dynasty value / pick equivalent, estimated age, breakout age), his college season stat lines (the last five seasons plus the current season to date) and the current season's game log (week, opponent, result and his passing / rushing / receiving line). Covers every player on the devy board (get_devy_rankings, ~6,100). College stats from CollegeFootballData.com; the season to date and game log run through the last completed week (refreshed weekly in season). If the name matches several players, the matches are listed: narrow with school, position or draft_year. See also: get_metadata for season coverage, valid enums, and analytic caveats.

StatHead devy (college player) rankings for dynasty leagues. The headline is StatHead's COMPOSITE value and rank: the devy market's price blended with StatHead's NFL career projection in rank space, priced on a smooth 0-9999 value curve. composite_weight = the career projection's share, set by the career model's own held-out accuracy at that position and distance from the draft (0.75 x Spearman, halved where it does not beat last-season production: about 0.11-0.17 for QBs, 0.17-0.25 for TEs, 0.26-0.35 for RBs and WRs), except where a backtest on the 2010-2022 classes found a better weight on held-out classes both within position and across the board (WR three seasons from the draft: 0.5); never above 0.5, so the market always leads. In that backtest the composite out-ranked a market-style price alone on NFL outcomes 0-2 seasons before the draft. Third-party values and ranks are inputs only: no field is a third-party number or rank. market_value / market_rank = the StatHead devy value model's price for the player's profile (what the market pays for a player like him: P(listed) x value if listed, from estimated age and draft age, breakout age, share of the offense, usage, counting stats, program, competition level and recruiting; held-out rank correlation with the market ~0.79 superflex / ~0.68 1QB). market_listed = on the market's devy list. hit_pct = StatHead's NFL career model IN THE CHOSEN FORMAT: the chance (percent) of at least one fantasy-starter season in his first four NFL seasons (6+ games above replacement PPR PPG for a 12-team league; single QB: QB13/RB30/WR42/TE13; superflex/2QB: QB25). About 3-5% of college players at this level hit, so even top prospects read in the 20s-60s. career_rank / career_pct = over the whole board by hit chance x what a hit is worth at his position. career_vs_market = market rank minus career rank (positive: the career model likes him more than the market model). draft_day = his chance of round 1 and of rounds 1-3; p_day1 / p_day2 / p_day3 / p_undrafted = chance he goes in round 1, rounds 2-3, rounds 4-7, or undrafted (finer splits by round did not validate). For the nearest class, a player StatHead's big board ranks gets the board's chances (projected pick -> actual draft day curve from the last draft; on the 2026 draft the board alone beat any blend with college stats). Otherwise from the college profile (round 1 recalibrated by distance from the draft), blended with an early mock draft where one projects him, or failing that with the pick his devy market price implies (a log-log line fitted on the big-board class, widened). QBs with no pick source have no draft outlook. format sf = superflex / 2QB, 1qb = single QB: every value and rank switches. dynasty_value / pick_equiv = the composite class rank priced as a rookie-draft slot on a smooth curve fitted to future-pick values, so he reads against NFL players and picks (get_dynasty_values). During the college season the profiles run through the last completed week (a calibrated full-season estimate). Ages are ESTIMATED from the high-school class, or without a recruiting record from his first college season in any division (JUCO and lower-division years count). In 1QB the market input for a listed player is his superflex price through a smooth per-position line, and the composite order WITHIN a position is the same in both formats (set by the superflex blend; posRank is identical): the format only moves positions against each other. Covers QB/RB/WR/TE in the next three draft classes, as deep as the data goes: every current college skill player the models score (about 6,100), not just the market's list. In season, only players on a current FBS/FCS roster or with stats this season at an FBS/FCS school: players off their team (left, out of eligibility, gone below FCS) are not ranked. Deep in the board the values are small and flat (one decimal below 10) and the rank carries the information; college_seasons = seasons with stats, this one included (0 = recruit only), i.e. how much evidence the ranking rests on. Reach deep players with player_name, school, position/draft_year filters, or page with offset. See also: get_metadata for season coverage, valid enums, and analytic caveats.

Get historical NFL draft picks with career outcomes. Includes round, pick, team, college, career approximate value (car_av), Pro Bowls, All-Pro selections, Hall of Fame status. Note: career columns (car_av, pro_bowls, etc.) are cumulative and are blank/low for recent draftees who are still active — they populate as careers progress. Use for draft analysis, career success by pick, team drafting history. See also: get_metadata for season coverage, valid enums, and analytic caveats.

Get historical draft prospect profiles: school, draft slot, physical measurements, and scouting report text (strengths/weaknesses). ESPN's own grades and prospect ranks are not shown (third-party rankings are never shown raw). Coverage: draft classes 1967–2021 only (this ESPN/JackLich10 source is not updated for 2022+). For recent/2022+ classes use get_rookie_class or get_college_stats. Use for prospect evaluation, draft class comparisons, historical draft analysis. See also: get_metadata for season coverage, valid enums, and analytic caveats.

Get StatHead's dynasty trade values and rankings: StatHead's blend of two dynasty markets (the geometric mean of FantasyCalc's value and KeepTradeCut's value calibrated to the same scale), shown in tens. Third-party values and ranks are inputs only; no field is a market's raw number or rank (position ranks are on StatHead value). Includes 1QB and SuperFlex values, position ranks, age, and stable ids (gsis_id, sleeper_id) for joining without name-string matching. Board depth is the market source's top ~500; a player absent from the board is a market judgment (valued below the top 500), not missing data. No TE-premium variant exists (the underlying market composites don't publish one) — for a TEP league, re-score get_projections (scoring tep0.5/tep1.0) and map the TE uplift through the value-vs-points curve. Use for dynasty trade evaluation, roster building, value comparisons. See also: get_metadata for season coverage, valid enums, and analytic caveats.

Week-level fantasy play log: returns per-player ordered scoring/touch events already reduced to fantasy shape (gsis-attributed), so consumers don't need the crosswalk or a client-side play reducer. One small call per week. Offensive events: {kind: pass|rush|rec|incomplete, yards, td, is_reception, is_target, two_point}. Committed-turnover events (attributed to the offensive player, not the defense): {kind: int_thrown (passer) | fumble_lost (ball carrier), turnover: 1}. Return events: {kind: kr|pr, yards, td} (kickoff/punt returns, attributed to the returner). Kicker events: {kind: fg|xp, distance, result}. Team-defense events (keyed by team = the defense): {kind: def, sack, interception, fumble_recovered, def_td, safety} plus a points_allowed summary per team-game. Per-defender IDP events (opt-in via idp=true; each credited defender's gsis): {kind: tackle (tackle_type solo|assist) | tfl | sack (sack=1 or 0.5 for half) | qb_hit | int | pd | ff | fr | def_td}. Every event carries game_id + play_id (traceable back to the get_play_by_play source row; note one play can fan out to multiple events, so play_id is not unique here) and game_date + time_of_day for chronological ordering. Filter to a roster with player_ids. Use output_format=jsonl for extraction. See also: get_metadata for season coverage, valid enums, and analytic caveats.

Get StatHead's redraft rankings: each player ordered by the average of two StatHead ranks — StatHead ADP (StatHead's blend of the draft markets, which include the FantasyPros expert consensus, Sleeper and FantasyFootballCalculator) and StatHead's own season projection (PPR points above a 12-team replacement level: QB13/RB30/WR42/TE13; within a position, plain projected points). Pass position for a positional board. Third-party rankings are inputs only; no expert or market rank is shown. Columns: rank (StatHead), pos_rank, adp_rank and proj_rank (the two StatHead ranks behind it), proj_pts, adp. Use for draft strategy and value picks (proj_rank far ahead of adp_rank = the projection likes him more than the market). See also: get_metadata for season coverage, valid enums, and analytic caveats.

Get FTN play-level charting data (2022+). Includes play-action, RPO, screen pass, motion, blitzers, pass rushers, drops, contested catches, QB pocket movement. Use for scheme analysis, play-calling tendencies, process-over-results evaluation. See also: get_metadata for season coverage, valid enums, and analytic caveats.

Get NFL game results and schedules. Includes the canonical nflverse game_id (e.g. 2024_04_BUF_BAL, handles neutral-site/relocated games correctly), scheduled kickoff (gameday = date, weekday, gametime = kickoff in US Eastern HH:MM), home/away/week, scores, spreads, totals, weather, surface. Use for team records, point totals, home/away splits, deriving day/time windows, bye weeks (a team absent from a week's games is on bye), or to enumerate game_ids to feed get_play_by_play. See also: get_metadata for season coverage, valid enums, and analytic caveats.

StatHead's high-school devy board: QB / RB / WR / TE / ATH recruits in classes not yet in college, ranked ACROSS positions by NFL fantasy outlook. hit_pct = chance (percent) of at least one fantasy-starter season in his first four NFL seasons (6+ games above replacement PPR PPG for a 12-team league in the chosen format); the rank weighs it by what a hit is worth at his position; p_drafted = chance he is drafted at a skill position. Calibrated on every high-school skill recruit in the 2007-2016 classes. The model's input is the recruiting composite rating, which is never shown; held out by class, nothing beat it (size, sub-position, committed program), so within a position the order is the recruiting services' and this tool has no position rank or position filter. Across a whole class it orders NFL outcomes as well as the raw rating, not better; it adds the cross-position, per-format scale (superflex lifts QBs). class_rank = rank within his class, across positions. See also: get_metadata for season coverage, valid enums, and analytic caveats.

Get weekly NFL injury reports (nflverse, 2009+). Per player-week: gsis_id (joinable to get_play_by_play / get_player_crosswalk), primary + secondary injury, practice participation (primary/secondary injury + practice_status), and final game report_status (Out/Doubtful/Questionable). Use for injury impact analysis, weekly availability, and attaching injury status to a roster. See also: get_metadata for season coverage, valid enums, and analytic caveats.

Weekly strength of matchup by offensive fantasy position (QB/RB/WR/TE) for 2026: what each defense has ALLOWED per game to a position this season, ranked 1-32 (1 = most allowed = softest matchup for the offense), laid over the schedule so every team's upcoming opponent reads as a matchup, in your scoring. Every table also carries OVER EXPECTED (oe): points or a metric allowed per game minus what the offenses faced produce in their other games this season (leave-one-out), i.e. the schedule-adjusted ledger — a defense that only looks soft because it met Detroit's tight ends has a high raw rank and an oe near zero. Views. WEEK (default; week defaults to the current one; position and team optional): one row per team and position with the opponent, the opponent's points allowed per game to that position (oppFPA), its rank, % vs league average, its over-expected (oppOE) and the StatHead model factor the weekly projections actually apply for that opponent (prior season blended in, shrunk toward 1) with its own rank, plus a lean (soft/tough = top/bottom quarter). DEFENSES (view=defenses): the defense-vs-position table — points allowed, rank, over expected, last 3 games, last season, model factor, next game; without position all four positions side by side. ROS (view=ros): strength of REST OF SCHEDULE per team — the mean of what the remaining opponents allow to the position (rosFPA, rank 1 = softest rest of schedule), its over-expected, the mean model factor, fantasy-playoff weeks 15-17 on their own and the bye; without position all four positions side by side. METRICS (view=metrics, position required): points allowed split into component metrics where the position has the volume — QB passPts/passAtt/passYds/passTD/int and rushPts/carries/rushYds/rushTD; RB rushPts/carries/rushYds/rushTD/ypc and recPts/targets/rec/recYds/recTD; WR and TE recPts/targets/rec/recYds/recTD/ypr — each per game with rank, over expected and oe rank; pass metric for one metric's full table. Pass metric in the week, schedule or defenses view to read that metric instead of points (e.g. position TE metric rec: which defenses give up the most TE receptions, and who faces them this week). SCHEDULE (team without week, or player_name): a team's or a player's week-by-week matchup strip from the current week on (from_week=1 shows the played weeks too; weeks_ahead caps it) — player_name "Brock Bowers" gives LV's TE matchups: the defense he faces, its TE points allowed, rank, over expected and the model factor. The raw ledger and the model factor disagree on purpose: four weeks of points allowed is loud while the model keeps about 40% of a deviation, so answer "who gives up the most to tight ends" with the rank, "is that real or schedule" with oe, and "how much to move his projection" with the factor. Points allowed are the opponents' actual fantasy points in the chosen scoring (receptions carried, so half, standard and TE-premium are exact; over-expected converts the same way), computed by StatHead from nflverse weekly stats; season to date counts every final game, including the finished games of a week in progress. Rates (ypc, ypr) are totals over totals. Every response quotes the 2016-2025 backtest reliability of what it shows (docs/matchups-validation.md): after 4 weeks only 10-25% of a defense's deviation reaches the players facing it (RB rushing the most, QB rushing and TDs none), a top-8 matchup is worth about +5-15% of a player's own average and a bottom-8 about -3-6%, and the model blend out-ranks the raw ledger until midseason — so lean on the factor for projections and on the raw rank only for the question of who concedes the most. Pairs with get_weekly_projections (the points) and get_schedule_strength (season-long model factors for every position incl. K, DST and IDP). See also: get_metadata for season coverage, valid enums, and analytic caveats.

Describe this server's capabilities: available tools, data sources, per-source season coverage, and valid enum values (positions, scoring formats, league sizes, ADP sources). Call this FIRST to scope a question correctly — it tells you what seasons/sources/formats are supported so you don't have to probe by trial and error. Takes no arguments.

How StatHead's models work and what drives them: a methodology overview (scored-player VOR hit/bust model, season projection pipeline, dynasty value, share models), the hit/bust thresholds and share-model cross-validation fit, and the TOP FEATURE IMPORTANCE per position (each feature's category, weight, and the direction of its relationship to the projection). Use to understand or explain the model. For one player's feature breakdown use get_player_features. topic=devy: the devy (college) models instead (validation, calibration, feature importance, backtest, data checks). Coverage: 2026. See also: get_metadata for season coverage, valid enums, and analytic caveats.

Get NFL Next Gen Stats — advanced tracking data powered by AWS. Passing: time to throw, air yards, aggressiveness, completion probability (CPOE). Receiving: separation, cushion, YAC above expectation, target share. Rushing: efficiency, rush yards over expected (RYOE), time to LOS, stacked box rate. Use for elite efficiency analysis beyond box score stats. See also: get_metadata for season coverage, valid enums, and analytic caveats.

Get play-by-play data with EPA, WPA, win probability, air yards, YAC, and stable player IDs (passer/rusher/receiver_player_id = nflverse gsis_id, joinable to get_player_crosswalk). Default view is lean; pass `fields` to project any of these slim columns — including kicker/defense/special-teams columns for K & DST fantasy scoring: kicker_player_id, kicker_player_name, kick_distance, extra_point_result; sack, interception, fumble, fumble_lost, fumble_recovery_1_team, safety; td_player_id, td_team, return_touchdown (separates defensive/ST TDs from offensive); two_point_attempt, two_point_conv_result; and kick/punt return columns: return_yards, return_team, kickoff_returner_player_id, kickoff_returner_player_name, punt_returner_player_id, punt_returner_player_name, fair_catch (note: yards_gained is 0 on return plays — use return_yards); committed-turnover + recovery attribution: interception (+ passer_player_id = who threw it), fumble_lost (+ fumbled_1_player_id/_name/fumbled_1_team = who lost it, fumble_recovery_1_player_id/_name = who recovered, forced_fumble_player_1/2_player_id = who forced it); per-defender IDP attribution (project via fields): solo_tackle_1/2_player_id, assist_tackle_1..4_player_id, tackle_for_loss_1/2_player_id, sack_player_id, half_sack_1/2_player_id, qb_hit_1/2_player_id, interception_player_id, pass_defense_1/2_player_id (all gsis, crosswalkable; defensive/ST TD scorer is td_player_id with td_team=defteam); and real (wall-clock) time: game_date, start_time (kickoff), time_of_day (UTC time the play was run — vs the game-clock `time` field; note nflverse populates it for most recent-season plays but it can be sparse/older-season gaps). Use for situational analysis (red zone, 3rd down, 2-minute drill), play type breakdowns, EPA-based efficiency, scoring K/DST, or pulling a complete game with game_id. WARNING: Large dataset — always filter by game_id, player, team, or situation. To pull an ENTIRE WEEK set limit ~3000 (a week is ~2400 plays) with week=N; to pull a WHOLE SEASON set a high limit (e.g. limit=50000, season ~43k plays) — for these large pulls always use output_format=jsonl or csv (optionally + fields) to keep the payload compact. You can also page with offset (the header reports the next offset and the total). When projecting with `fields`, include game_id/play_id/posteam/defteam yourself if you need them for joins. See also: get_metadata for season coverage, valid enums, and analytic caveats.

Player ID crosswalk: maps the stable nflverse gsis_id (the *_player_id in get_play_by_play) to a canonical full name and cross-source IDs (pfr_id, sleeper_id, espn_id, pff_id, yahoo_id, sportradar_id, rotowire_id, fantasy_data_id, esb_id), plus ready-to-use headshot URLs: espn_headshot (ESPN CDN, when espn_id is known) and headshot (best available — falls back to the roster headshot_url for fresh/in-season rookies ESPN hasn't indexed yet). espn_id and sleeper_id are backfilled from the season rosters when the static crosswalk lags. Use to resolve the abbreviated PBP names (e.g. J.Allen) to a real player, to join PBP against name-keyed endpoints, and to surface player photos in apps. Filter by player_id (gsis), player_name, position, or active season. See also: get_metadata for season coverage, valid enums, and analytic caveats.

Feature evaluation for a SCORED player, broken out PER MODEL that scores them: Hit/Bust (VOR), Season Projection (PPG), Rookie Career/Prospect (rookies), Usage Share (RB/WR/TE), and a market-derived Dynasty Value note. Each model shows its prediction plus the top feature 'drivers' — where the player lands vs the positional cohort FOR THAT MODEL (percentile band) and which way the feature pushes that model's prediction. Proprietary/paid-sourced features are shown only as a qualitative magnitude band + direction (never a raw value or source). Use to explain WHY a player is scored the way they are. Coverage: 2026 scored skill players (QB/RB/WR/TE). See also: get_metadata for season coverage, valid enums, and analytic caveats.

Get a comprehensive analytics profile for a player. Combines weekly stats, play-by-play, NGS tracking data, PFR advanced stats, FTN charting, and snap counts into a single metrics object. QB metrics: comp%, Y/A, AY/A, passer rating, EPA/dropback, CPOE, pressure rate, time to throw, play-action rate, scramble rate, designed rush rate. Skill (RB/WR/TE) metrics: YPC, catch rate, YPRR, target share, WOPR, separation, YAC above expected, RYOE, snap%, opportunity rate, drop rate. Use for in-depth player evaluation, cross-position comparisons, efficiency analysis. Returns one row per player matching the query. See also: get_metadata for season coverage, valid enums, and analytic caveats.

Get aggregated season totals for NFL players — offence, defence (IDP), kicking and punting. Returns rushing, passing, receiving, fantasy stats and 2-point conversions, and via `fields` every other counting column nflverse carries: defensive (def_tackles_solo, def_tackle_assists, def_tackles_for_loss, def_sacks, def_qb_hits, def_pass_defended, def_interceptions, def_fumbles_forced, def_tds, def_safeties), punting (pt_att, pt_yards, pt_net_yards, pt_inside_20, pt_long, pt_touchback) and kicking (fg_made/fg_att by distance band, pat_made/pat_att). Filter by position: QB/RB/WR/TE/K/P/FB, the nflverse defensive codes (DE, DT, LB, CB, SAF), or the IDP buckets DL (DE+DT+NT), LB and DB (CB+SAF+FS+SS). Sort by any stat column. Punting columns need a season pulled from nflverse's unified table — 2025 onward locally, any season when fetched live. Use this for player performance, fantasy points, rankings, comparisons. See also: get_metadata for season coverage, valid enums, and analytic caveats.

Get week-by-week stats — game logs — for one player or for a whole position in a season. Pass player_name for one player's log, or omit it and pass position (and optionally week_start/week_end or team) for every player at that position, which is how you get a weekly board. Covers offence, defence (IDP) and specialists: with `fields` you can project any raw nflverse weekly column, including def_tackles_solo, def_tackle_assists, def_sacks, def_qb_hits, def_pass_defended, def_interceptions, def_tackles_for_loss, def_tds, punting (pt_att, pt_yards, pt_net_yards, pt_inside_20) and returns (punt_returns, punt_return_yards, kickoff_returns, kickoff_return_yards). Positions accept QB/RB/WR/TE/FB/K/P, the nflverse defensive codes (DE, DT, LB, CB, SAF) and the IDP buckets DL/LB/DB. A whole position-week is a few hundred rows, so narrow with week_start/week_end or team, and raise limit deliberately. Use for game logs, weekly trends, consistency and boom/bust analysis, and week-by-week IDP scoring. See also: get_metadata for season coverage, valid enums, and analytic caveats.

StatHead's first-party season fantasy projections: in-house projected fantasy points-per-game for the upcoming season — veterans via a prior-year-actual / 2-year-average / age-curve blend, rookies via the rookie career model. This is the season-level companion to get_prospect_outcomes (rookies) and get_dynasty_values (long-horizon value), and the answer to "what does StatHead project for a veteran." Coverage: 2026, QB/RB/WR/TE plus kickers (position K), team defenses (position DST) and individual defensive players (positions DL, LB, DB — or IDP for all three), each projected separately with components. Punt/kick return components ride on every row that has a return role, whatever the position. IMPORTANT for ranking: ppg is projected points divided by projected games played, so it is a rate CONDITIONAL ON PLAYING, not a season expectation. A backup projected for one game divides a one-game line by one and lands beside the best starters in the league. Every row therefore carries `games` (the denominator) and `projPts` (the season total). To rank players against each other — a draft board, a lineup, a roster sort — use projPts, or pass min_games; use ppg only when reading one player's per-game rate. IN-SEASON rank on rosPts instead: projPts is a FULL-season total that still includes games already played, while rosPts covers only what is left, with rosPPG, rosGames and gamesRemaining alongside it. Responses flag any returned rows with games <= 4. RE-SCORING UNDER CUSTOM SCORING: ppg is a scalar priced under standard PPR. Every row also carries the projected season stat line it rolls up from — pass_att, pass_cmp, pass_yd, pass_td, pass_int, rush_att, rush_yd, rush_td, tgt, rec, rec_yd, rec_td — selectable via `fields`, and zeroed (not omitted) for categories a position doesn't accrue. Re-derive the total from those under your own scoring catalog; re-deriving standard PPR from them reproduces projPts to within a mean 0.07 pts/season. Kickers and team defenses DO carry components: K has fga/fgm split 0-29, 30-39, 40-49, 50+ plus xpa/xpm; DST has pts_allow (season and per game), sack, def_int, fum_rec, def_td, st_td and safety, plus pa_points_pg (the points-allowed bracket integrated over the per-game distribution) and pa_game_sd (the spread it was integrated over, ~9.4). In-season both K and DST blend toward what the current season has shown, on curves fitted against rest-of-season outcomes (n/(n+9.5) for kicker volume, n/(n+10.5) for DST points allowed / sacks / INTs — about 30% weight by week 4, 45% by week 8); rows carry inSeasonWeeks and inSeasonWeight so you can see how much is current form. Do not front-run it with your own hot-hand adjustment: measured over 2017-2025, current-season form ALONE predicted the rest of the season WORSE than the preseason prior alone, for both positions. So leagues pricing field goals by distance or re-tiering points allowed can re-score both. Re-tiering points allowed: integrate your tiers over a per-game normal around pts_allow_pg with pa_game_sd, do NOT score the season mean against them — the bracket is convex, so point-scoring the mean overstates the good defenses in a consistent direction. Note DST's projection does NOT beat a flat league mean on RMSE (1.384 vs 1.375 over 2023-25) and carries only modest ordering value (r~0.25) — use its components to re-score, not its ordering to rank. IDP is the opposite case and the strongest projection here: DL/LB/DB rows carry tackles/solo/assist/tfl/sack/qb_hit/pd/def_int/ff/fum_rec/def_td/safety, and the model beats a flat positional mean by 24% on RMSE (1.367 vs 1.801 over 2023-25, r=0.72) because defensive per-game rates persist far better than offensive ones (QB hits r=0.82, sacks 0.74, solo tackles 0.73). Weekly IDP bonuses: weeks_2plus_sack and weeks_3plus_pd are the expected game counts, with sack_game_sd/pd_game_sd/tackles_game_sd for integrating other thresholds — never score a season mean against a weekly threshold. Return components: pr, pr_yd, kr, kr_yd, ret_yd, ret_td (plus pr_pg/kr_pg/ret_yd_pg and the pr_role/kr_role depth-chart flags), season totals scaled to each row's own projected games. Not modelled anywhere: fumbles, first downs, two-point conversions and yardage milestones for skill players. Preset boards carry rescaled ppg without a stat line. See also: get_metadata for season coverage, valid enums, and analytic caveats.

StatHead's own prospect model: draft grade, tier, projected draft slot, and CALIBRATED boom/bust probabilities. IMPORTANT — boomProb/bustProb/outperfPctile are RELATIVE to the player's draft slot, not absolute fantasy outcomes. boomProb = P(the player OUTPERFORMS the model's draft-slot-informed PPG expectation for a profile like this); bustProb = P(underperforms it); outperfPctile = percentile of that outperformance vs slot. So a Day-3 pick with a high boomProb is likely to beat Day-3 expectations — NOT likely to be an absolute fantasy star. For absolute expected value use grade / tier / projPick. Use these instead of multiplying your own marginal factors into a "P(Hit)" — that is over-confident because factors are correlated (see get_metadata analytic caveats). Boom/bust probabilities cover the 2026 class; grades cover 2026 and 2027. Probabilities are model estimates, not certainties — treat as calibrated guidance.

Get ESPN QBR (Total Quarterback Rating) — a comprehensive QB efficiency metric from ESPN. Season-level or week-by-week data from 2006 onward. Includes QBR total, points added, EPA breakdown (pass, run, expected sack, penalty), and raw QBR. Use for QB evaluation, comparing efficiency across eras, weekly performance tracking. See also: get_metadata for season coverage, valid enums, and analytic caveats.

One-call rookie-class join: NFL draft picks for a draft year + combine testing + that player's rookie-season (same year) production — so you don't have to join draft, combine, and stats by hand. Skill positions (QB/RB/WR/TE) by default. Each row carries pfr_id for joining to other tools. Great for prospect/dynasty evaluation and hit-rate analysis. Anti-pattern: do not multiply marginal factors (athleticism × age × draft slot) into a "P(Hit)" — the factors are correlated and the product is over-confident; use as ordinal ranking or compare similar-player (joint) comps instead. See also: get_metadata for season coverage, valid enums, and analytic caveats.

Rookie snap-share RAMP tracker — weekly offensive snap % for a season's rookie class, with a ramp signal (last-3-week avg − first-3-week avg) that surfaces mid-season role expansion a static depth chart misses (the classic late-bloomer pattern, e.g. a rookie RB climbing from 30% to 70% snaps). Identifies rookies from the roster (years_exp 0 / rookie or entry year = season). For an in-progress season this is a live leading indicator; for a completed season it's the role-trajectory record. Sort by ramp (biggest risers), latest, peak, or avg. See also: get_metadata for season coverage, valid enums, and analytic caveats.

Get NFL team rosters with player details: position, status, height, weight, college, birth date, years of experience, draft info, headshot URL. Use for player biographical info, roster composition, experience levels. See also: get_metadata for season coverage, valid enums, and analytic caveats.

Strength of schedule per team, per position, for every game of 2026 — how easy or hard each team's slate is for QB/RB/WR/TE/K/DST. A factor above 1 means an easier schedule (that position's opponents concede more than league average); below 1 means harder. Season factors are the mean over a team's games; per-game factors name the opponent and venue. CRITICAL, read before using: these factors are ALREADY APPLIED to StatHead's K and DST projections (their season lines are built alongside these numbers), so re-applying them there double-counts. They are deliberately NOT applied to QB/RB/WR/TE — the weekly matchup multipliers for skill players are normalized to mean 1, so they redistribute points between weeks without moving a season total, which the projection pool owns. If you want the season-level schedule effect for skill players, apply it yourself from here. DL/LB/DB follow the skill-position rule — published, normalized out of the weekly IDP strip, never applied to the season line — and are deliberately close to 1 because offense-concedes-to-IDP barely repeats year over year (r = +0.25 DL, +0.24 LB, +0.08 DB). Spread across 2026: ~8% DST, ~6% QB, ~5.5% WR, ~4.5% TE, ~3.3% K, ~3.1% RB, ~2.5% DL, ~1.3% LB, ~0.4% DB between the easiest and hardest schedule. Derived from the same defense-vs-position table the weekly projections use, so published and applied numbers cannot drift. These are model multipliers, not points allowed: for what each defense has actually conceded per game to QB/RB/WR/TE this season, ranked 1-32 in your scoring and laid over upcoming weeks, use get_matchups. See also: get_metadata for season coverage, valid enums, and analytic caveats.

Get Sleeper draft data. Pass draft_id for that draft's full pick board (round, pick, team, player, position). Or pass username (+ season) to list that user's drafts and their ids. Use for draft recaps, keeper/rookie draft analysis. Source: Sleeper public API. See also: get_metadata for season coverage, valid enums, and analytic caveats.

Open a Sleeper league: format/settings, standings (W-L, points for/against, each manager's owner_id), and every team's roster with starters (by lineup slot) and bench, resolved to player names. Identify the league by league_id, OR by username + name (e.g. "the Football Warriors league") — no id needed. Each manager's owner_id is emitted so you can walk to get_sleeper_user_leagues / get_sleeper_user_snooper for that manager. Source: Sleeper public API. See also: get_metadata for season coverage, valid enums, and analytic caveats.

List the managers in a Sleeper league — display name, team name, owner_id, roster_id, and record — without pulling full rosters (a cheap hop). Identify the league by league_id, OR username + name. Use each manager's owner_id (or display_name + this league_id) to walk to get_sleeper_user_leagues / get_sleeper_user_snooper. Source: Sleeper public API. See also: get_metadata for season coverage, valid enums, and analytic caveats.

Get head-to-head matchups and scores for a Sleeper league in a given week. Pairs teams by matchup, with each team's points and (optionally) starters. Source: Sleeper public API. See also: get_metadata for season coverage, valid enums, and analytic caveats.

Kept for compatibility: returns StatHead's own projections, not Sleeper's (third-party projections are inputs only, never shown raw). Pass week for StatHead's weekly projection (same as get_weekly_projections), or omit week for StatHead's season projection (same as get_projections). See also: get_metadata for season coverage, valid enums, and analytic caveats.

Get a Sleeper league's transactions — trades, waiver claims, and free-agent adds/drops — with players and teams resolved to names. Pass a week, or omit it to sweep the season (weeks 1-18). Filter by type. Identify the league by league_id OR username + name. Use for trade history, waiver activity, and roster churn. Source: Sleeper public API. See also: get_metadata for season coverage, valid enums, and analytic caveats.

Get trending player adds or drops from Sleeper fantasy platform. Shows which players are being most added/dropped across all Sleeper leagues. Use for waiver wire analysis and league-wide sentiment. See also: get_metadata for season coverage, valid enums, and analytic caveats.

Multi-season Sleeper career history for a user: every league they fielded across recent seasons with record, points, regular-season finish, and championship/runner-up result, plus a career summary (total record, titles, leagues). Identify the user by username/user_id, OR display_name + league_id. Source: Sleeper public API. See also: get_metadata for season coverage, valid enums, and analytic caveats.

Look up a Sleeper user's fantasy leagues. Identify the user by username/user_id, OR by display_name + league_id (to scout a manager you just saw in get_sleeper_league). Returns each league's id, name, size, status, and format (Dynasty/Keeper/Redraft, Superflex/2QB/1QB, Best Ball, IDP), plus the resolved user_id. This is the entry point of the Sleeper graph: take a league_id from here into get_sleeper_league / get_sleeper_matchups / get_sleeper_waiver_wire. Source: Sleeper public API. See also: get_metadata for season coverage, valid enums, and analytic caveats.

Scout a Sleeper user across all their leagues for a season: their league list plus cross-league player exposure — which players they roster in the most leagues (and start most), with the league names. Identify the user by username/user_id, OR by display_name + league_id (to scout a manager you just saw in get_sleeper_league). Use to see who a manager is heavily invested in. Source: Sleeper public API. See also: get_metadata for season coverage, valid enums, and analytic caveats.

Find the best available (un-rostered) free agents in a Sleeper league. Cross-references the league's rostered players against all NFL players, then joins Sleeper trending-add counts and StatHead's projections: projected PPG, projected games, and rest-of-season points (ros_pts, the in-season ranking column). Sort by trending (default), ros, or ppg — ppg ranks players projected for at least min_games first, because a backup's ppg is a per-game rate conditional on playing (sorting on it alone returned thirty backup quarterbacks). Use for waiver-wire and streaming decisions. Source: Sleeper public API + StatHead projections. See also: get_metadata for season coverage, valid enums, and analytic caveats.