How LoserQ Counts Your Games

Every verdict rests on a defined sample, fixed rules, and a versioned method. This page spells out all three.

Which games count

LoserQ analyzes Ranked Solo/Duo on Summoner's Rift only — no Flex, no Arena, no customs. We request your most recent Solo/Duo match history from the Riot API and analyze every game that passes the checks below. Your result always states how many eligible games it is based on.

What gets excluded

A game never enters the sample if any of the following applies:

  • Not Ranked Solo/Duo
  • Invalid match timestamp
  • Remake (under 5 minutes)
  • Wrong number of players
  • Incomplete player data
  • Malformed teams
  • Player not found in match data

Every excluded game is counted, and the exclusion breakdown is shown in the evidence section of your result.

Carry and Feeder are team-relative

A player is classified relative to their own team in that one match — never against a global bar. A Carry stood far above their teammates in that game; a Feeder fell far below and pushed the game toward losing. The same stat line can be a carry performance in one lobby and unremarkable in another.

Neither label is decided by a single stat. Deaths, KDA, damage, vision, kill participation, and objective play are compared against teammates, weighted by role. Every badge on a result lists the specific reasons behind it.

The balance score

Each eligible game gets one of three outcomes: it counts toward loser queue when your teammates fed clearly more than the enemy team, toward winner queue when the enemy fed clearly more, and as balanced otherwise. The balance score is the average of those outcomes across your sample — 0% means your matchmaking luck was even.

Uncertainty

Alongside the score we show a 95% descriptive interval: the range the average balance outcome plausibly falls in, given how much your recent games vary. Small samples swing hard, so the interval is wide when few games are eligible.

Sample maturity turns that into a plain label: limited (under 10 eligible games), developing (10–29), and full (30 or more). Treat a limited sample as an anecdote, not a verdict.

Method versioning

Every result is stamped with the version of the method that produced it. The current version is Method v1. The version changes only when the eligibility rules, the balance outcome definition, or the interval math change meaning — so two results with the same version are always computed the same way.

Data freshness

Results show when the underlying match data was last fetched from Riot. A cached result is a snapshot, not a live feed — use the update button on the result page to pull your newest games.

What LoserQ cannot conclude

LoserQ describes your recent games — nothing more. It cannot observe Riot's matchmaking system, and no result here proves that matchmaking is rigged for or against you. The interval reflects variation in the analyzed matches only. LoserQ is not endorsed by Riot Games and shows no official rating or MMR.