Start with the plain-English version below. The complete master equation, full pseudocode, every constant and the trained weight tables follow further down. Values shown are the parameters currently deployed.
1 · The bar. Before grading anyone the engine predicts the match: team skill, who's attacking, bots, console share, premades, new players, this map's quirks — and where the matchmaker put you. Battle rating runs BR 1–5, and what counts is whether you were placed above the BR you brought: an up-tier is something done to you, so it lowers the bar. That prediction E is the bar — win as the heavy favorite and the result barely moves you; upset strong odds and it pays.
Your equipment does not rate you. Bringing better gear never raises your rating and bringing worse gear never lowers it. The only BR question the engine asks is the up-tier one above.
2 · The grade. Win and it's 70% how you played, 30% that you won. Lose and the stat line counts 60%, the result 40% — so a pretty scoreboard can soften a loss but rarely erase it. And the stat line is judged against what players of your rating usually score in a lobby like this: topping a weak lobby is what the engine expects of a strong player, so it pays little; doing it in a strong lobby pays in full. "How you played" is 17 stats, each measured against your own lobby and weighted by your job that match — and dying while DEFENDING costs nothing by itself (defenders don't burn tickets; the penalty measured as noise). Valves: losing records get their stat credit damped, blowout winners aren't punished for coasting, and a newcomer's stat line counts less for their first 8 games (from 75% of its usual weight).
3 · The move. Step size scales with uncertainty squared — new players swing, veterans barely move. Your uncertainty also shrinks with matches played, so a long record can't keep swinging just because your lobbies were one-sided. If you lost and the math still says "up", the gain is cut to 70% — losing is never a good way to climb. First 10 games clamped to ±75; skill estimate never goes below 200.
4 · Deserting. Leave a match more than a minute after you first spawn and it counts as a loss your stats can't soften, even if your team goes on to win. Leave within the first minute, or before spawning, and the match isn't rated. That is the whole rule: the quit costs you once, and nothing follows you into later matches. It applies to matches from 5 October 2026 on; in older matches a quit counted as a loss only after a third of the match. If every real player on one team has left, the match isn't rated for anyone — nobody takes a loss, and nobody gets a win against an empty team.
5 · The ladder. Ranked by μ−2σ, experience-shrunk — a confidence statement: "you're at least this good." Tiers are daily percentiles of the same number, so tier and rank can't disagree.
6 · Two ladders. Matches with at least one PC player rate on the Main ladder; matches without PC players (all-console lobbies) rate on the Console ladder only. The relationship is one-way: a console player's games against PC players still inform their Console rating, but console-only games never touch the Main ladder — so a Main-ladder rank is always earned against the PC player pool. Console players carry both ratings; tiers are cut separately within each ladder.
Everything above, stated exactly: one master equation, the full pseudocode, the per-stat transforms and the complete 9×17 weight matrix.
Everything the engine does to your rating in a single match is this one update (each piece defined below):
## CONSTANTS & PER-PLAYER STATE Q = ln(10)/400 = 0.005756 # classic Elo logit scale μ = mean of players with 20+ rated matches − 25, once 200 such players exist # skill estimate at first sight (1000 while the field is small) σ = 377.19 # uncertainty; floor 25.81, shrinks every game games = rated matches so far # W/L/D rows only record = career wins/games # never includes the current match [AI] = ONE shared rating for every bot ## STAGE 1 — BEFORE THE MATCH: predict P(team A wins) = E # team skill: honest pre-match roster, softmax by μ with β=0.0007: your strongest players weigh a little more than the plain mean. # map-veto: drop a player in the battle under 60 s, the same line as a quit counting # (a bot: under 2% of the match): loaded in, disliked the # map/mode and left — never really on the team: w_i = exp(0.0007 × (μ_i − max μ on team)) teamμ = Σ w_i μ_i / Σ w_i ; teamσ² = Σ w_i² σ_i² / (Σ w_i)² g = 1 / √(1 + 3 (Q √(teamσ²_A + teamσ²_B) / π)²) slope = SLOPE[mode] # inv 3.45 ass 3.09 dst 2.57 cnf 3.30 dom 3.56 other 1.45 f_map = shrunk(map skill-agreement, k=20, centered on mode) slope = slope × (1 + 0.25 × (f_map − 1)) ctx = −0.0578 × (premade-frac diff) + 0.2482 × (biggest-party-share diff) + attacker offset: inv +0.047 ass +0.782 dst +0.540 + −0.4 × (bot-count diff) + −0.60 × (console-frac diff) + 0.5 × shrunk side-bias(map, k=20) # symmetric maps only + −1.0 × (tier̄_A − tier̄_B) # tier = lobby BR − BR you brought, clamped to ±1 + −0.45 × (new_A − new_B) # new = Σ humans exp(−games / 10) # new = a team's newcomer load: 1.0 for a first rated match, 0.37 at 10 # games, 0.14 at 20, ~0 for a regular. A fresh rating says "average" but # first matches are won ~44% of the time; lowering the starting rating # only shifted the whole scale, so the correction is made HERE, in the # expectation. Same roster as team skill (map-veto tail dropped); bots # and incognito rows carry no load. Prediction only — no rating is # moved for being new or for playing beside someone who is. # # tier = (lobby BR) − (your own BR), clamped to one tier either way and # to the real BR scale (BR runs 1..5, so no "BR 0" and no "BR 6"): # tier = matchBR − clip(yourBR, max(1, matchBR−1), min(5, matchBR+1)) # POSITIVE = up-tiered = placed above what you brought = bar drops. # It lands in {−1, 0, +1}: a placement flag, NOT a gear score. Your # own BR is read from the loadout you SPAWNED with, so a gun picked up # off any corpse cannot move it, and the enemy's gear never enters at # all — equipment does not rate you. # Needs ≥3 BR-known humans per team, else the term drops out. # # Weight 1.0: a team up-tiered by one whole bracket more than its opponents # has its expectation lowered by about what a 170-Elo gap does at slope 1. # The older lineup-BR term (team mean BR difference) is RETIRED: it let # your own equipment move your rating, which this engine must not do. E = logistic( slope × Q × g × (teamμ_A − teamμ_B) + ctx ) ## STAGE 2 — AFTER THE MATCH: grade every human player x[stat] = raw / minutes_played # 17 counting stats per minute # z-pools: captures/points/obj_kills_att pool ATTACKERS only, obj_kills_def DEFENDERS; # deaths pool the whole lobby. Transform per stat (table below), then: z[stat] = (transform(x) − lobby mean) / lobby stddev # missing → 0; winsor caps ±2 z = Σ W[mission:side][stat] × z[stat] # full 9×17 matrix below perf = logistic(z / 1.8) if record < 50% (after 10 games): perf ×= max(0, 1 − 0.5(0.5−record)/0.5) S = 1 if won else 0 a = 0.7 if won else 0.6 # a lost match grades 60/40, not 70/30 (alpha_loss) if WON a blowout: a ×= (1 − 0.7 × margin) # coasting a stomp isn't punished if games < 8: a ×= 0.75 + 0.25 × games/8 # newcomers lean on the RESULT target = (1−a)×S + a×perf gap = your μ − mean μ of the humans in the match expected = usual perf at this gap # table of earlier matches, 100-point steps ## STAGE 3 — MOVE THE RATING (Glicko step, exact) g_o = 1 / √(1 + 3 (Q √(opponent teamσ²) / π)²) e = your team's E par = (1−a)×e + a×expected # result judged against the win chance, stats against expected σ = min(σ, max(25.81, 600 / √games)) # uncertainty capped by matches played d² = 1 / (Q² g_o² max(e(1−e), 1e-6)) denom = 1/σ² + 1/d² Δ = (Q / denom) × g_o × (target − par) # scales with σ² when new if LOST and Δ > 0: Δ ×= 0.7 # loss-gain cap — a gain in a loss keeps 70% if games < 10: Δ = clamp(Δ, ±75) # placement cap — no first-game jackpots μ = max(200, μ + Δ) # hard floor σ = max(25.81, √(1/denom)) ## STAGE 4 — DESERTION # matches from 2026-10-05 on: left early, in the battle ≤ 60 s → not rated (clock starts at your first spawn) left early, in the battle > 60 s → a LOSS with target = 0 (stats can't soften it) # older matches: the line was 35% of the match instead of 60 s # a quit costs once; it never changes how a later match is graded every human on one team left (bots holding it) → match NOT RATED for anyone, any date ## STAGE 5 — WHAT THE LADDER SHOWS Elo = μ ± σ Rating = 825 + (μ − 2σ − 825) × (1 − e^(−games/30)) # experience shrink: 49% of your edge at 20 games, 81% at 50, 96% at 100 Tier = live percentile of Rating (rated ≥ 10, recut daily): Diamond top 0.5% · Platinum 3% · Gold 12% · Silver 40% · Bronze rest
Each stat is reshaped before it's z-scored, so long-tailed counts don't swamp the grade. Grouped by the transform applied:
| transform | applied to | what it does |
|---|---|---|
| sqrt | kills, building_usage, obj_kills_att, long_range_kills, kills_as_inf | compresses big numbers — a 40-kill game isn't 4× a 10-kill game |
| log1p | deaths, obj_kills_def, fortification_built | strong compression for counts with long tails; log(1+x) so 0 stays 0 |
| rank | squad_deaths, vehicle_kills, rally_usage, kills_as_veh | replaced by lobby rank — only the ordering matters, not the raw gap |
| winsor ±2 | assists, headshots, points_captured | kept linear but clipped at ±2σ so one outlier can't dominate |
| linear | captures_credited, battle_score | used as-is (after per-minute), then z-scored like everything else |
Your job that match picks the column: inv=invasion, ass=assault, dst=destruction, cnf=confrontation, dom=domination; att/def=side, sym=symmetric. Green = raises your grade (deeper = stronger); red = lowers it. A dash (—) means the stat isn’t scored for that side — capture stats only count for attackers, defend-kills only for defenders — so its trained weight there never touches the grade. Death penalties on the three DEFENDER columns are zeroed. Defenders have had a ticket pool since 29 September 2026, but a re-fit on the matches after that date still found the penalties made predictions worse, so assault is graded the same way before and after the update.
| stat | inv:att | inv:def | ass:att | ass:def | dst:att | dst:def | cnf:sym | dom:sym | other |
|---|---|---|---|---|---|---|---|---|---|
| kills | 1.37 | 1.42 | 1.46 | 0.90 | 0.15 | 0.99 | 1.83 | 1.06 | 0.88 |
| deaths | −0.48 | 0 | −0.92 | 0 | −0.60 | 0 | −0.27 | −1.49 | 0 |
| assists | 1.04 | 1.81 | 1.29 | 0.53 | 0.31 | 1.44 | 0.14 | 0.23 | 0.29 |
| captures_credited | 0.63 | — | 0.40 | — | 0.66 | — | 1.56 | 0.99 | 1.74 |
| squad_deaths | −1.10 | 0 | −0.82 | 0 | −0.24 | 0 | −1.47 | −0.21 | −1.82 |
| vehicle_kills | 0.46 | 1.07 | 0.83 | 0.36 | 0.49 | 0.52 | 0.91 | 0 | 0.71 |
| building_usage | 0.46 | 0.33 | 0.44 | 0.17 | 0.88 | 0.59 | 0.30 | 0.60 | 1.44 |
| battle_score | 0.34 | 0.61 | 0.06 | 0.25 | 0.88 | 0.86 | 0.61 | 1.49 | 1.04 |
| obj_kills_att | 0.54 | — | 0.39 | — | 0.92 | — | 0.54 | 1.07 | 0.21 |
| obj_kills_def | — | 0 | — | 0.62 | — | 0.46 | 0.88 | 0.62 | 0.59 |
| headshots | 0.07 | 0.93 | 0.59 | 0.49 | 0.25 | 0.72 | 1.13 | 1.14 | 0.37 |
| points_captured | 0.49 | — | 1.13 | — | 1.08 | — | 0.19 | 1.82 | 0.49 |
| fortification_built | 0.18 | 0 | 0.45 | 0.54 | 0 | 0.89 | 1.24 | 0.13 | 0.38 |
| long_range_kills | 0.53 | 0.03 | 0.65 | 0.89 | 0.76 | 0.70 | 0.35 | 1.23 | 0.15 |
| rally_usage | 1.32 | 0.18 | 0.83 | 0.07 | 1.81 | 0.33 | 0.04 | 0.47 | 1.05 |
| kills_as_inf | 1.45 | 0 | 0 | 0.54 | 0.20 | 0.84 | 0.54 | 1.18 | 0.78 |
| kills_as_veh | 0.35 | 0.41 | 0.21 | 0.40 | 0.27 | 0.30 | 1.14 | 1.29 | 1.06 |
Not scored at all (weight 0 in every column): kd, human_kills, killstreak. They mostly repeat what the stats above already measure, and leaving them out made predictions more accurate.
Everyone is re-rated from scratch, oldest match first, whenever the formula changes — the ladder is always one consistent formula end to end.