Your end-of-game dialog might show you gained fifteen points while your leaderboard rank barely budged. That gap confuses almost every new player in Beyond All Reason. The two numbers track different things, and they are supposed to drift apart sometimes. Understanding what each one measures will save you from chasing the wrong target.
Game rating is what the match lobby uses to put you against evenly matched opponents and teammates. Before every game starts, the server assigns each player an OpenSkill value. Think of it as the system's best guess about your current playing strength. It updates after every single match — win or lose, 1v1 or 8v8 — by comparing the actual outcome against what the skill values predicted. The bigger the surprise, the bigger the adjustment. Upset a favored opponent and you gain a lot. Lose to a player the system already ranked below you, and you drop more.
Leaderboard rating, on the other hand, is a seasonal accumulation built on top of a subset of ranked play. The leaderboard filters for competitive integrity. Queue dodging, imbalanced team sizes, and certain casual modes do not always count. The leaderboard also resets with each season so new players have a realistic shot at climbing. Your game rating might sit at 1850 after two hundred mixed matches, but your leaderboard number could be 1620 because half those games were unranked lobby fills or 3v3s where the matchmaking was stretched thin.
Why does the gap matter in practice. The game rating is the number that actually decides who you face next. If you are stuck in a loop of stomps or getting stomped, your game rating is still finding its center. The leaderboard rating is the badge that shows how well you performed in clean, competitive conditions over a season. Treat it like a scoreboard, not a report card.
OpenSkill is not Elo, even though people use the terms interchangeably. Elo was designed for chess — a closed, zero-sum, two-player game. BAR throws teams, imbalanced sides, disconnects, and uneven information into the mix. OpenSkill handles that uncertainty by maintaining both a mean (your estimated skill) and an uncertainty (how confident the system is in that estimate). New accounts start with high uncertainty, so the first twenty games swing your rating wildly. That is by design. The system is gathering data, not punishing you for a shaky start.
A few things that regularly confuse players:
If you want to climb your leaderboard rank specifically, focus on ranked duels and small-team queues where the matchmaking is tightest. Those results carry the most weight. If you want to improve your actual game, treat the raw game rating as noise and look at your replays instead. The number that matters is the one that keeps you growing, not the one that changes after a lucky win against a disconnected opponent.
Chasing numbers leads to bad habits. You see a low rating, you turtle up and play scared, and that passive style teaches you the wrong things. The rating is not a measure of your worth as a player. It is a matching tool. Play more games, play them attentively, and the rating will settle where it belongs. The players who climb fastest are the ones who treat post-game screens as feedback loops, not verdicts.
If rating confusion is part of your frustration, playing with a regular group makes a real difference. Creed of Champions runs team games where people actually talk through their matches afterward. Nobody yells when your individual rating dips after a rough lobby. The crew looks at what happened, shares replays, and helps each other read the stats instead of obsessing over a single number. It is the kind of environment where rating anxiety disappears because you have real people to talk to who care more about improvement than leaderboard placement.
[Crd] Before discovering Creed, I was thinking the only thing that separates BAR from the perfect RTS is a friendly and safe social environment for new players to learn and feel included.