How BrawlPick Works

The three kinds of information on this site

• A. Observed statistics — a real win rate measured from actual battles BrawlPick collected, always shown with its sample size and a last-updated date (see the map statistics page). This is a description of what happened in those specific battles, not a prediction.
• B. The picker's recommendation— the output of BrawlPick's scoring algorithm for your specific draft: your map, mode, bans, team, the enemy picks and (optionally) the brawlers you own. It uses observed statistics as one input when enough exist, but it is a model's judgment call, not a measured fact.
• C. General strategic explanation — background knowledge about roles, positioning and how a game mode plays, used to explain reasoning and to fill gaps where no reliable observed data exists yet. This is not a statistic and is never presented as one.

What the picker actually considers

For every brawler still available, the picker combines several factors into a single score. The exact weight given to each factor shifts through the draft — early on, when little is known yet, flexibility and general map/mode fit matter more; once picks are visible, countering the enemy and covering your team's roles matter more.

• Map and game mode fit — how well a brawler suits the specific map's layout and the mode's objective.
• Observed win rate — when enough real battles exist for that brawler on that map or in that mode (see the reliability section below); otherwise this factor contributes nothing rather than a guess.
• Enemy counters — brawlers that tend to beat the specific brawlers the enemy has already picked.
• Team synergy and role balance — whether a pick complements your team's existing picks and covers a role your team is missing.
• Flex safety — early in a draft, when the enemy team is still mostly unknown, the picker leans toward brawlers that stay strong across more scenarios.

How bans and picks affect eligibility

Banned brawlers, your own team's picks and the enemy's picks are all removed from the pool of brawlers the picker will suggest — this is a hard eligibility filter, not a scoring penalty. A brawler either can be picked or it can't.

Ownership and power level

If you add the brawlers you own (and, optionally, import them automatically from your player tag), the picker can limit its suggestions to brawlers you can actually field. Brawl Stars itself requires Power 9 or higher for a brawler to be Ranked-eligible, so BrawlPick uses that same floor when checking ownership.

Power level does not affect a brawler's score. A Power 9 and a Power 11 version of the same brawler you own are ranked identically on strategic fit — power level is only used to decide whether a brawler counts as available to you at all, never to make one ranked higher than another.

Where the observed statistics come from

BrawlPick reads recent battle logs from the official Brawl Stars API, using player tags that BrawlPick users have added to their own profile (plus, optionally, a small number of seed accounts). Each battle is identified by its timestamp and all six participants' tags, so a battle seen through more than one tracked player is only ever counted once — including across separate collection runs, not just within a single run.

This is a convenience sample: the players who happen to have added a tag to BrawlPick, not a random or representative sample of the entire Brawl Stars player base. Trophy range, region and play style of BrawlPick's own users can all skew it. Treat the numbers as real observations from that sample, not as a claim about every player everywhere.

Reliability thresholds — and why the picker and this website use different ones

Internally, the picker will use a brawler/map or brawler/mode row once it has a small minimum number of observed battles — enough to be better than nothing when blended with map fit, counters and synergy. That bar is appropriate for a recommendation that also weighs several other factors, but it is not high enough to publish as a standalone win-rate claim.

The map statistics page therefore applies a stricter, independent bar before showing a number in public: at least 30 observed battles for a specific brawler on a specific map, and a map is only listed at all once at least 5 brawlers individually clear that bar on it. Below that, BrawlPick shows nothing for that map rather than a number that looks precise but isn't.

As of the last update (September 19, 2026), 53 of 54 maps that had at least one brawler with enough battles to consider went on to meet that full bar.

How often the data refreshes

The Ranked map rotation shown across BrawlPick updates automatically every day from real competitive Ranked battles. The win-rate statistics on the map statistics page and the guides are different: they come from a deeper collection run that is triggered manually, not on a fixed daily schedule. The "last updated" date shown next to any statistic is the actual date of that collection run — check it before treating a number as current, especially right after a balance update.

Balance updates

Observed statistics only reflect battles played under whatever brawler balance was live at the time they were collected. BrawlPick does not currently tag rows with a specific game version, so a statistic could in principle mix battles from just before and just after a balance update if one landed during the collection window. In practice each collection run captures a short, recent window (typically a few days), which limits how much this can happen — but it is not eliminated. If a statistic looks off right after a known balance change, the last-updated date is the first thing to check.

When there isn't enough data

For maps or modes without enough observed battles, the picker falls back to general brawler traits and role/matchup knowledge (category C above) instead of a statistic it doesn't have. That fallback is reasoned, but it is not a measurement — the picker's explanation for a pick tells you which kind of evidence it actually used.