Smurfs and Unranked Players in Dota 2 Fair Customs

Updated 2026-07-13

How do you spot a smurf in your Dota 2 group?

A smurf shows up in your group as a rank medal that doesn't match how the player actually performs — a Herald or Guardian medal paired with someone consistently winning lanes against players two or three tiers above that band. The signature to look for is recent win rate on a low-game-count account: a player sitting on a 70%+ win rate over 15-20 games at a low medal is very likely still calibrating or intentionally smurfing, since new and low-MMR accounts move through placement quickly once results are lopsided. A genuine low-rank player's win rate tends to hover much closer to 50% over the same sample, because that's what the matchmaking system is designed to converge toward at any stable rank.

Smurfing is common enough in Dota 2 that it's worth assuming at least one account in a 10-stack of casual friends is either a smurf or someone who hasn't played ranked in a long time and is sitting on a stale medal. Neither case is malicious most of the time — people smurf to play with lower-ranked friends, or their medal simply hasn't caught up to months of inactivity — but both distort a straightforward MMR-average split the same way: the number on the profile stops describing the player actually walking into the lobby. Valve, for its part, declared smurfing a bannable offense in 2023 and banned 90,000 smurf accounts in a single wave, with penalties reaching linked main accounts — but that enforcement targets ranked matchmaking, and none of it fixes the stale medal sitting in your custom lobby.

How should you weight a known smurf when balancing teams?

Once you know a player is smurfing, the fix is to stop trusting their medal and manually weight them at the skill level you actually know them to be. If a Divine-level player is running games on a fresh Herald account, entering them into your split at their real rank — not the account's — prevents the balancer from stacking three genuine low-rank players against what is effectively a fifth core carry. Dota 2 Groups lets you add and adjust players manually for this reason: when the public data is misleading, a human correction based on known information beats trusting a number that's actively wrong.

How do you handle unranked or private-profile players in a fair split?

Unranked or private-profile players carry the opposite problem — there's no medal to distrust, because there's no public data at all. The most fair approach is to spread unknown players across both teams rather than clustering all of them on one side, so the uncertainty each unknown represents is distributed evenly instead of concentrated. Dota 2 Groups handles this by letting you add any player manually by typing a name even without a Steam profile, and the algorithm still balances using whatever information is available for the rest of the group — an unknown player isn't excluded, just balanced around.

For best results, ask players to enable public match data in the Dota 2 client — Settings, Options, Advanced Options, Expose Public Match Data — which lets OpenDota read their rank and history going forward. That single setting change turns a fully unknown player into a fully verified one for every future lobby, which is worth doing once for a group that plays together regularly instead of re-guessing every time someone new joins.

Why do smurfs and unranked players break simple MMR splits?

Both smurfs and unranked players break a simple MMR-average split for the same underlying reason: the split trusts whatever number it's given, and both cases feed it a number that's wrong in a different direction. A smurf's medal understates their real strength, an unranked player has no number at all, and a balancer that only looks at rank tier has no way to catch either problem on its own. That's why recent win rate, not just medal, is a second signal worth checking manually whenever a Dota 2 customs lobby includes players you suspect aren't accurately represented by their rank — the medal tells you the band, the win rate tells you whether that band still fits, and a healthy balance score depends on getting both inputs right before the split ever runs.

Frequently asked questions