KICK Rating

What KICK Rating is

KICK Rating is a single per-game score between 0 and 100 derived from the box-score stats of an individual player's match. It answers a blunt question: how well did this player play, in this game, compared to a typical professional performance?

It is the headline figure used on every player profile, every match recap, and the leaderboard on this site. The headline number you see next to a player's name is their rolling 40-game KICK Rating — the average of their KICK Rating across their last 40 games where they played 50% or more of the match. We chose 40 games rather than a calendar window so a player on a hot run doesn't drop overnight in the off-season, but also can't rest on ancient form.

How to read a KICK Rating

KICK uses one formula and a 0–100 scale, but the distribution you're reading depends on whether the number is a single-game score or the rolling 40-game average (the underlying number on every player profile, retained since v1.8 as the Brownlow-aligned reference under the KICK-P headline — see "Position and role still matter" below). Averaging 40 games compresses the distribution toward the middle, so the bands are different. Percentiles below are derived from the actual rated population, not vibes.

Single-game KICK — what one game was worth

• 85+ — once-a-season performance. Top ~1% of all eligible games.
• 75–84 — best-on-ground calibre. Top ~3%.
• 65–74 — clearly elite game. Top ~8%.
• 50–64 — good to very good. Top ~20%.
• 35–49 — median senior performance.
• Under 35 — quiet, role-restricted, or truncated game.

Rolling 40-game KICK — the underlying overall score

Across ~5,660 leaderboard-eligible players (30+ games in the rolling window). This is the Brownlow-aligned overall score, retained on every profile as the secondary “reference” badge under the position-fair KICK-P headline (since v1.8).

• 70+ — elite of elite. Top 0.3% of active players. Bontempelli / Gawn / Xerri tier.
• 60–69 — All-Australian squad contender. Top 1.5%.
• 50–59 — established star. Top ~5.5%.
• 40–49 — consistent best-22. Top ~20%.
• 30–39 — solid senior contributor.
• Under 30 — fringe, depth, or injury-interrupted.

The overall rolling-40 number doesn't break 85 because a rolling 40-game average smooths out the once-a-season games that ceiling single-game scores at 100. Bontempelli's leading rolling-40 KICK of 81 means he has averaged an elite performance across two seasons straight by the overall formula — not that any individual game hit 81. His position-fair KICK-M scaled value reads 100 (elite midfielder ceiling), which is the headline on his profile; the 81 is the Brownlow-aligned secondary. Same player, same games, two scales.

Position and role still matter — a shutdown defender and a midfield ball-winner with the same KICK Rating are doing different jobs. That's what Positional KICK (KICK-M / KICK-D / KICK-F / KICK-R) is for. Since v1.8 (2026-05-09) every player profile leads with their KICK-P score as the headline number, with the overall KICK retained as a secondary “Brownlow-aligned reference” badge. Harris Andrews reads KICK-D 86 (elite defender) on his profile instead of overall 41 (average); Tony Lockett reads KICK-F 87 career, with peak 90 in his 1990–1992 window (top-5 forward) instead of overall 43. Same player, same games, different lens — the lens that ranks within position is the right default for “is this player good at their job?”.

Coaches' eye — the AFLCA lens

KICK Rating measures on-field actions in the box score. The Coaches' eye is a parallel lens, not a re-rated KICK and not a blended composite. Every match the two senior coaches award AFLCA Champion-Player votes (5-4-3-2-1 each, top six players get votes). Aggregated per game and scaled 0–100 within each position, the Coaches' eye is the answer to which players do coaches actually respect.

The two lenses correlate strongly on midfielders (Bontempelli, Daicos, Petracca, Cripps, Neale top both). They diverge most on defenders and key forwards, where structural contributions (shutdown defending, decoy running, intercept positioning, marking presence) earn coach votes but don't always make the stat sheet. The structural ceiling that closed v1.2 / v1.3 / v1.4 means we can't move headline KICK to fix this; the honest move is to surface the other signal alongside, transparently.

Worked example: James Sicily's overall KICK rolling-40 reads 53 (“solid”) on the box score. His Coaches' eye reads 100 — he is the most-respected active defender in the AFLCA record over 2003–2025. Reading both numbers together is sharper than reading either alone. The Coaches' eye card on every voted-era player profile shows both, plus the delta.

Eligibility: 80+ career games and 40+ games in the voted era (2003–2025). Data window: 2003 (first season AFLCA published per-game votes publicly) to 2025; 2026 votes backfill at season's end. Anchors: p50→50, p80→70, p95→85, p99→95, max→100, computed within each position pool. Position labels: inherit the same career-aggregate classifier + V1.8 manual overrides used by KICK-P.

Not Fantasy. Not SuperCoach.

If the weights table below looks Fantasy-shaped, that's partly because there are only so many publicly-available stats to weight — but the goals, outputs and validation targets are different. Fantasy and SuperCoach exist to make a weekly game fun: their weights are tuned for ownership dynamics, price movement and breakevens, and they accumulate unbounded per game. KICK is tuned to correlate with the game's own recognition signals — we back-test against Brownlow top-10 overlap, All-Australian overlap, and AFLCA Champion Player votes. v1.1 hits 4.36/10 on Brownlow agreement; not Fantasy's job to care about that.

The practical consequence: a player with high KICK and low Fantasy (or vice versa) is a signal, not a rounding error. KICK rewards job-specific work the Fantasy weights don't credit (contested possessions, clearances, goal assists, rebound-50s, one-percenters); Fantasy rewards raw disposal volume. We're measuring different things on purpose.

The formula

A per-game KICK Game Score is the weighted sum of a player's on-field actions, divided by a flat normaliser so that a typical professional performance sits around 50 and a once-a-season game ceilings around 100.

Per-stat weights (v1.1)

Normalisation

The weighted sum is divided by 1.9. That's a scaling choice, not a natural constant — we picked 1.9 because it maps a typical AFL men's game (roughly 95 raw-score points on these weights) to KICK 50. Any other divisor would still be a scaling choice. The 0–100 scale only means something in context, which is why the "How to read" section up top anchors specific values to actual percentiles of the rated population (per-game 80+ = top ~1%, rolling-40 headline caps around 80 for current elites).

Eligibility filter & time-on-ground weighting

A player's KICK Rating only counts games where they were on the ground for 50% or more of game time. This prevents sub-ordinary injury-substitution cameos from dragging a rating up or down on noise.

From v1.1 we also weight by time-on-ground, because sustaining elite output for a full game is harder than peaking in a cameo:

• 90%+ TOG — ×1.06 (played the game out at elite level)
• 80–89% TOG — ×1.03
• 65–79% TOG — ×1.00 (baseline)
• 50–64% TOG — ×0.95 (qualifying, but modestly rebated)

Rolling 40 vs career

Every profile shows both. Rolling 40 is sustained recent performance — a player's last ~1.5 seasons of games, smoothed. It's not "hot-form last week"; a 10-game window would be, but 10 games is too noisy for a single-number rating (Champion Data's official Player Rating uses the same 40-game window for the same reason). Career is legacy. The leaderboard uses rolling 40 and a 30-game minimum — we're not interested in small-sample top-tens driven by three good games.

What's not in v1

Honest limits, because transparency matters more than flattering the rating:

• No efficiency data. We don't yet have disposal efficiency, kick-to-handball ratios, or effective kicks. These are proprietary Champion Data. If they land in a public feed we add them.
• No positional adjustment in headline KICK. A key-defender's job is to lower the opposition score; their overall KICK will always trail midfielders for that reason. This is why we ship Positional KICK (KICK-M, KICK-D, KICK-F, KICK-R) — a separate per-position calibrated overlay, live on every profile and leaderboard since 2026-04-21. The headline KICK number is kept pure so cross-position career comparisons stay honest; Positional KICK is the right lens when comparing within a position.
• No era adjustment in headline KICK. A 1985 rover's KICK Rating lives on 1985 weights; a 2025 rover's lives on 2025 weights. v1.6 looked at era rebaseline offline (2026-04-25) and held — both naive and percentile-matched approaches produced indefensible top-20 lists. Era-fair viewing lives on the decade leaderboards and positional all-time leaderboards.
• Hit-out-to-advantage acquired but unused. The v1.4 AFL.com.au scrape pulled hit-outs-to-advantage for 2012+ matches; weighting it crashed AFLCA alignment in the sweep, so headline KICK keeps the v1.1 raw-hit-out weight. Ruckmen remain underweighted on overall KICK; KICK-R (positional) is the right view for ruck rankings.
• Position classification is career-aggregate, not per-year. Players who shifted roles across their career get one label based on dominant career stats. The position-fair leaderboards and team-page splits use that single label. Where we know the classifier is wrong for current usage, we maintain a small manual override registry: Lachie Ash (GWS half-back) reroutes Midfielder → Defender; Mark Blicavs (Geelong) reroutes Ruck → Utility because his role is genuinely hybrid (early ruck, late-career third-tall back). Blicavs appears under the "Honourable mentions" footnote on team pages rather than in any single position bucket. Overrides are routing-only — the player keeps their originally-computed KICK score, mathematically consistent with how the formula scored them; only the navigation label changes.
• AFLW is not yet covered. v1.1 is men's-only. AFLW ships with v2 once a separate AFLW normaliser is calibrated against the shorter-game baseline.

Why this is open

Rating players is a market with walled gardens. Champion Data runs the AFL's official numbers. Fantasy stats use their own weightings. Neither publishes a formula.

KICK Rating is reproducible from public stats. Everything on this page is the full spec: weights, normaliser, eligibility, window. Anyone with an AFL Tables match scrape and 50 lines of Python gets the same numbers. That's the point.

We'd rather you argue with the formula than trust it blindly.

Version history

v1.7 — 2026-04-25 — Investigated and held. Game-level WOWY (With Or Without You) — for every player, average team margin in games they played vs games they missed within their career. Computed from lineup + match-result data already on disk; no new scrape required. Two compute attempts (100/20-game floors with per-season aggregation, then 200/12 floors with career-total cohorts) plus a 9-combination parameter sweep all failed the face-validity gate. Best result was 175/12 floors with 3 of 10 face-validity targets in top 20; gate required 6. The metric measures team-context confounds (when player happened to be present during good team era) rather than individual impact at AFL public data granularity. Structural ceiling: 23 games/season + 22-player rosters + co-absence clusters + career-team-trajectory correlation (Andrews' career improved AS Brisbane improved — "without" cohort clusters in pre-improvement years). NBA WOWY works because of minute-precise on/off-court tracking + 82-game seasons + clean replacement causality — AFL public data has none of those. Held permanently. Companion shipping: Career-best games block on every player profile (top 10 highest-KICK individual matches, expandable to 20). Full hold writeup on the validation page.

v1.6 — 2026-04-25 — Investigated and held. Pre-2000 era rebaseline was attempted offline against two approaches (naive era divisor + percentile-matched mapping). Both produced indefensible top-20 lists — Jack Clarke at #1, Peter McKenna at #3 above Bontempelli, with 12 modern stars dropped from the top 20. The deeper finding: overall KICK is structurally mid-biased by design (correct for Brownlow alignment), so rebaseline can lift pre-2000 midfielders but cannot lift specialists like Lockett. The all-time positional leaderboards were instead switched to rank by Peak 40 (matching the renamed "Peak-form leaderboard" semantics on the overall page), which puts Lockett top-5 all-time forwards on KICK-F. Pre-2000 caveat copy on the all-time leaderboard was also rewritten to route readers to the positional + decade views where era fairness already exists.

v1.5 — 2026-04-21 to 2026-04-24 — Metric expansion (headline KICK formula unchanged from v1.1). Positional KICK overlay: KICK-M / KICK-D / KICK-F / KICK-R, per-position calibrated variants that strip the midfielder bias out of rankings. Live on every profile, every match recap, and the leaderboard. Explainer: /kick-rating/positional/. Peak 40: best rolling-40-game window across a player's career. Ranks peak form rather than career average. All-time leaderboards + team all-time lists now sort by Peak 40 so Dustin Martin (peak 81) correctly ranks above Matt Crouch (peak 73) even though their career averages tie.

v1.2 / v1.3 / v1.4 — all held — v1.2 weight-tuning sweep (48 configs), v1.3 mechanism sweep (32 configs), and v1.4 advanced-data sweep (972 configs) all failed to produce a configuration that cleared the binding pass criteria. v1.4 acquired the missing data — intercept marks, score involvements, hit-outs-to-advantage and four other fields scraped from AFL.com.au for 2012–2025 matches — but the same trade-off held: every weight that closes position gaps drags Brownlow / AFLCA / AA alignment below the floor. Three holds confirms the ceiling is structural to the single-number formulation, not a data poverty problem. Full decisions on the validation page.

v1.1 — 2026-04-19 — Clangers scaled by clanger-to-disposal rate (baseline 10%, slope 5, cap 2×, floor −10). Time-on-ground multiplier added (0.95× @ 50–64% TOG; 1.00× @ 65–79%; 1.03× @ 80–89%; 1.06× @ 90%+). Every historical game re-scored under the new formula.

v1.0 — 2026-04-19 — Initial release. AFL men's only. 13,279 rated players, 678,319 per-game scores, across 1897–2026. Flat clanger penalty (−1.0 per clanger, capped at −10). No time-on-ground multiplier. AFLW deferred to v2.

How v1.1 performed vs v1.0

We back-tested both formulas against our own validation harness. Honest result: the changes were a mixed bag.

• Brownlow alignment (top-10 overlap, 2000–2025 seasons): v1.0 = 4.06/10, v1.1 = 4.36/10 — a +7.4% improvement. v1.1 is marginally closer to the umpires' read of the best players each season.
• Position coherence (career-KICK gap between midfielders and other positions): all three gaps widened under v1.1 — mid-vs-defender 22.35 → 22.64, mid-vs-ruck 18.81 → 20.21, mid-vs-forward 30.38 → 31.92. The clanger-rate scaling rewards high-disposal players (midfielders) more than low-disposal positions, which we didn't anticipate in the design.
• Era normality: effectively unchanged under v1.0 or v1.1. v1.6 looked at a dedicated era rebaseline offline and held — both naive and percentile-matched approaches produced indefensible top-20 lists, and the deeper finding was that overall KICK is mid-biased by design (correct for Brownlow alignment), so era rebaseline can't lift specialist forwards/defenders even after correction. Pre-2000 readers are routed to positional + decade leaderboards instead.

Net assessment: v1.1 is a narrow improvement on Brownlow alignment, a narrow regression on position coherence. We're shipping it as labelled because the Brownlow win is real and the position regression is modest. The position-gap problem was instead solved by shipping Positional KICK as a separate per-position overlay (2026-04-21) rather than by rescoring headline KICK — v1.2, v1.3 and v1.4 sweeps all confirmed that closing position gaps inside the headline metric tanks Brownlow/AFLCA alignment (see below).

What we tried — v1.2, v1.3, v1.4, v1.7 all held; two structural ceilings

v1.2 ran a 48-configuration weight-tuning sweep. v1.3 moved past weights to three new mechanisms — role-aware multipliers, scoring-involvement composite, and era normalisation — in a 32-configuration sweep. v1.4 acquired seven advanced fields from AFL.com.au's match centre (intercept marks, score involvements, hit-outs-to-advantage, intercepts, metres gained, pressure acts, f50 ground-ball gets) covering 2,879 matches 2012–2025, then ran a 972-configuration sweep across the new fields crossed with the v1.3 mechanism toggles. Zero configurations across all three sweeps cleared their binding pass criteria.

The pattern across all three: every mechanism that closes position gaps tanks Brownlow and AFLCA alignment. You can't lift defenders / rucks / forwards relative to midfielders without midfielders dropping out of top-N rankings, which are dominated by mids in both the Brownlow umpire vote and the AFLCA coach vote. v1.4 specifically: intercept_marks at weight 5.0 with role-aware + era_on closed the defender gap from 22.17 to 11.92 (a 46% reduction), but every top-50 config dropped Brownlow below the 3.811 floor and AA peaked at 45.85% (target 50%). AFLCA bifurcated — configs weighting hit-outs-to-advantage crashed AFLCA to 65–74%; configs without it held v1.1's 82.61%. The 87% target was unreachable.

v1.7 attempted a different angle: a parallel WOWY metric measuring team-margin shift with-vs-without each player. Two compute attempts (different floors) plus a 9-combination parameter sweep all failed face validity — best result was 3 of 10 target stars in top 20 (gate required 6). The metric was measuring team-context confounds (player happened to be present during good team era) rather than individual impact at AFL public-data granularity. Game-level WOWY hits a structural noise ceiling in AFL specifically: 23-game seasons + 22-player rosters + co-absence clusters + career-and-team trajectories that often correlate. Different ceiling from v1.2–v1.4, same outcome.

All four held. v1.1 stays live. The honest read across the four attempts: the binding criteria require KICK to simultaneously rank positions proportional to their on-field impact and agree with awards that strongly over-index midfielders. Those are incompatible when midfielders genuinely win most Brownlows (they do) AND when the position-gap target says defenders deserve scores closer to a ratio inconsistent with those awards (it does). The ceiling on the single-number formula is structural to the target, not to box-score poverty — and v1.7 confirmed the team-impact angle has its own structural ceiling at AFL public-data granularity.

The fixes shipped instead: Positional KICK (KICK-M / KICK-D / KICK-F / KICK-R) as a parallel rating where each position is calibrated against its own peers; Peak 40 as the all-time leaderboard headline so short-peak careers don't tie with long-peak-plus-decline careers; and the Career-best games block on every player profile from the v1.7 work.

Full hold decisions and sweep results: validation page.

On the name

Yes, the rating is called KICK and one of the 17 stats it weights is also called kicks. The double meaning is deliberate, not oblivious. Footy is named after the act of kicking the ball; so is this rating. When the context is ambiguous: KICK in uppercase is the rating, kicks in lowercase is the stat.

Got an issue with the formula, found a stat we're double-counting, or want to propose a v2 weighting? Get in touch.