Methodology

The site's database statistics are accumulated from recent battles returned when player tags are searched. The sample may be biased toward regions, skill levels, or communities that search more often, and the number of unique stored tags is not the same as the number of unique users.

The official battle log response covers up to the latest 25 battles. When the same tag is searched again, battles already stored are skipped through UNIQUE constraints on battle time and battle fingerprint. The fingerprint uses time, mode, map, and sorted participant tags.

• For normal two-team battles, each battle fingerprint is selected once, then participants from both teams are expanded to calculate results for the winning and opposing teams.
• For Duo and Trio Showdown, even when there are multiple teams, the API does not provide each opponent's individual result, so only the placement from the searched player's perspective is used.
• Friendly battles are excluded from recommendations and accumulated win-rate statistics.

Win rate and sample count are shown when a map-brawler combination has at least 5 battles. The recommendation score is an internal score that discounts win rate more when the sample is small; 5 battles do not guarantee representativeness. Values shown on screen may be cached for up to 60 seconds.