A skill-based rating for OC Battle — built to reward beating strong pilots, not just racking up wins against easy opponents. Here's exactly how it works.
The Battle leaderboard has always ranked pilots by total wins. That rewards playing a lot, but it doesn't tell you anything about the strength of who you beat — grinding wins against brand-new accounts counts exactly the same as beating the best pilot on the platform.
Battle Rating fixes that. It's a per-pilot skill estimate that goes up more for beating strong opponents and barely moves for beating weak ones — and it's specifically hardened against the obvious exploit: spinning up a second account to farm easy wins, or to snipe rating off an established pilot.
Battle Rating uses Glicko-2, the same family of rating system used by chess and other competitive ladders — a refinement of the classic Elo system.
Every pilot has a rating that starts at 1500. Before each race, the system knows every participant's current rating and how confident it is in that number. During the race, for every opponent you're in a room with, it calculates the probability you should have beaten them based on the ratings going in — then compares that to what actually happened.
Beat someone you were expected to beat easily? Small gain. Beat someone rated well above you? Bigger gain. Lose to someone far below you? Bigger loss. This happens individually for every opponent in the room, then combines into one rating update after the race — so a 12-pilot free-for-all is scored on who you specifically beat and lost to, not some average difficulty of "the room."
New pilots start provisional — the system doesn't have enough data to be confident in their rating yet. Provisional status clears after your first 15 rated races, once the system's confidence in your number has converged.
This matters because of what's next.
A brand-new account is, by definition, an unknown quantity — the system doesn't yet know if it's a genuine beginner or an established pilot's second account. Two extra safeguards sit on top of the standard Glicko-2 math to make sure that uncertainty can't be weaponized:
How much rating an established pilot can lose to provisional opponents in a single race is capped — both per opponent and in total for that race. Getting beaten by one new account barely dents your rating. Getting beaten by several new accounts in the same lobby is capped too, not multiplied.
Only your first few rated meetings against the exact same opponent within a rolling window count toward rating. Facing the same pilot over and over — say, two accounts queuing into each other on purpose — stops moving the number well before it could be farmed.
Both safeguards only ever apply to the provisional/repeat-opponent portion of a race. Losing fair and square to strong, established pilots always costs full rating — the guardrails target manipulation, not legitimate upsets.
To be explicit: this is a Battle Rating, scoped entirely to Battle Royale races. It is not an overall OCGP skill score, and it does not use anything from Grand Prix or Solo results. Grand Prix is a different format — endurance, no elimination, no HP — and a shared number wouldn't mean anything comparable between the two. Grand Prix will get its own, separate rating system in the future.