How to tell if someone is smurfing in League of Legends
Someone is smurfing when a new account is performing like an old one. Both halves matter: a fresh account by itself is just a fresh account, and a great KDA by itself is just a good game.
That is the whole test, and it is why most of the advice you find on this is useless. "They had high CS" describes every strong player at their correct rank. The question is whether the account has any business being that good.
What counts as a new account
Ranked needs level 30, twenty champions owned, and ten normal Summoner's Rift games (the current requirements are laid out here). Levelling comes at roughly one level per game or two, so the useful mental line sits well above 30.
- Under level 60 the account has played on the order of a hundred games total. In a Platinum lobby that is a genuinely new account, whatever it is doing there.
- Level 150 and up means hundreds of games. Elite numbers on an account like that are much more likely to be a good week than a hidden main.
- Between the two, level tells you almost nothing on its own.
You can see summoner level on any profile lookup. It is the fastest check available and almost nobody does it.
The win rate test, and the sample it needs
A win rate is only evidence once there are enough games behind it. Below about 30 ranked games, ignore it completely. A 70% win rate over 10 games is what ten games look like.
Over 30 or more:
- 65% and above means the account has not settled at its rank yet. Matchmaking pushes everyone toward 50%, and it only fails to do that when the player is better than the bracket they are in. At 30 games, 65% is 19 or 20 wins, which happens by chance roughly one time in twenty. Evidence, not proof.
- 55% and below over a real sample is the signature of an account that has found its rank. It argues against smurfing, even if the last three games looked terrifying.
Put those two checks together and most "is this a smurf" questions answer themselves. Level 34 at 71% over 40 games is a smurf. Level 312 at 53% is a player having a good run.
The performance test, per role
Only now do the stats matter, and they have to be compared against the right thing. Four benchmarks are worth reading, and if any of them are new to you, here is what each one actually measures:
| Benchmark | What it catches | The trap |
|---|---|---|
| CS per minute | Farming that is out of place for the rank | Meaningless for supports, and for one game where they were behind |
| Gold per minute | Converting a lead into more lead | Inflated by a stomp that was not their doing |
| KDA | Not dying, which is mostly experience | Inflated by playing safe in a won game |
| Kill participation | Being where the game is happening | Naturally high in short games |
Two rules keep this honest. Compare each game against the baseline for the role it was actually played in, because a support's 270 gold per minute and a mid laner's 400 are not the same achievement. And compare against the rank of the lobby, not against an absolute number, because an absolute bar flags most high-elo players and misses the low-elo smurf who wins without farming.
Somewhere around 25% above the role baseline, on two or more of the four, is where "good game" stops explaining it. Under three recent games, do not form an opinion at all.
What does not work
Username patterns. "Word plus digits" gets repeated everywhere as a smurf tell and it is a bad one. It is unfalsifiable, and it accuses real people on the basis of their name. We deliberately do not use it and neither should you.
One good game. Everyone has them. The player with a 14/2 score last game went 3/9 the game before, and you only saw one of those.
Map awareness and ward placement. Real signals in principle, and completely unavailable to you in a loading screen. Advice that requires you to have already played the game against them is not advice.
Champion mastery of zero. Suggestive when it comes with everything else, worthless alone. Plenty of people first-time a champion in ranked.
What Riot does about it
Riot has not published the details, so treat what follows as community reporting rather than fact. The consistent account is that accounts winning far too much for their bracket get pulled into matches against each other within ten to fifteen games, that hardware identifiers collected by Vanguard connect a "new" account to a main on the same machine, and that smurfing on its own rarely leads to a ban because the system prefers to simply match you upward.
The practical version: if the enemy mid laner is genuinely a Diamond player on a fresh account, the system is probably already fixing it. That does not help you in the next 25 minutes.
What to do in the game you are already in
Assume the answer is yes and play as if it is. It costs you very little if you are wrong.
- Give up the trade you are not certain of. Smurfs win by punishing small mistakes repeatedly, so stop offering them.
- Play to the second wave, not the first. Their early lead compounds. Take the safe farm and let your jungler decide whether the lane is worth contesting.
- Tell your jungler once, with a reason. "Enemy mid is level 41 on a 70% win rate" is information. "Mid diff incoming" is not.
- Track their summoner spells. Experience shows up as flash timing more than anything else, and it is the one thing you can actually exploit.
Where we fit
Doing the checks above by hand means opening five tabs while the loading bar runs. LaneDiff does them for all ten players once the game starts: the account level, the ranked win rate and sample, and the per-role performance comparison, in the sixty seconds you actually have.
Two honest caveats. It is a signal and not a verdict, so it is worded that way in the app. And the smurf flag is off by default, because it depends on correctly working out who is playing which role, and a wrong role guess against a support baseline used to manufacture false flags. You can turn it on in your settings once you know what it is telling you.