Ban Rate Is Not a Power Ranking

Ban rate repeats identically across every role row a multi-role champion shows with a third-party League of Legends statistics provider, because a ban removes a champion before anyone has a role — unlike win rate and pick rate, which genuinely vary by lane. It measures how badly players want a champion gone, not strength: Nasus is banned in 37% of games despite an unremarkable 51.8% top-lane win rate.

A number that doesn't change when the row does

Look at a multi-role champion across a third-party statistics provider's per-role tables and one column behaves differently from the other two. Win rate and pick rate genuinely move from row to row, the way you'd expect a champion's performance to differ by lane. Ban rate does not — it repeats, to the tenth of a percentage point, in every role that champion is shown in:

Compare that against the same rows' win rates. Nasus posts 51.8% top, 48.0% jungle, and 53.0% mid — a real, five-point spread depending on where he's played. His ban rate barely twitches across the same three rows. That's not a coincidence specific to Nasus; it's the pattern every multi-role champion on the board shows.

Why: a ban happens before anyone has a role

The reason is structural, not statistical. You cannot ban a champion "in mid" — a ban removes that champion from the entire lobby, for both teams, before a single person has picked anything, let alone been assigned a lane. Win rate and pick rate are naturally per-role numbers, because a game where Nasus was played top is a genuinely different data point from a game where he was played jungle. A ban carries no role information to split by at all — there is exactly one ban count and one games-in-tier count behind it, full stop. A per-role ban rate isn't a smaller or noisier version of a real quantity; it's not a measurable quantity in the first place. Every source that prints one — including the third-party statistics provider this app's own data comes from — is taking that single champion-level number and repeating it once per role row, because the table format calls for a value in that column and the real number doesn't vary by row to put there.

The small wobble in the Nasus figures (37.1 / 37.0 / 37.1) isn't a genuine per-role effect either — it's sampling drift between separate data fetches for each role's table, each pulling the same underlying champion-level ban count at a very slightly different moment or rounding. Worth saying plainly: that wobble is noise, not signal, and treating a 0.1-point difference as if Nasus is "banned slightly less in the jungle" would be reading precision into a number that was never that precise to begin with.

What ban rate does measure

None of this makes ban rate meaningless — it answers a real question, just a different one than win rate does. A ban is a pre-emptive decision made by someone who hasn't seen the matchup yet, based entirely on how unpleasant they expect facing that champion to be. That tracks frustration, popularity, and reputation far more directly than it tracks measured strength. A champion can be banned constantly because playing against it is miserable — long stuns, hard-to-punish scaling, a kit that feels unfair to lose to — regardless of whether its actual win rate backs that reputation up.

Nasus is the cleanest illustration on the board. At 37% banned, he carries one of the highest ban rates of any champion shown, and yet his top-lane win rate — 51.8% — is a modest, unremarkable edge. Run that edge through the tier's own noise floor and it's about 0.33 standard deviations at Gold (see the noise floor behind every tier's win-rate data for how that conversion works and what counts as a meaningful gap): comfortably inside the range this app's own calibration treats as ordinary variation, not a standout performer. The highest ban rate on the board sitting next to a win-rate edge that small is exactly the gap between "widely disliked to play against" and "measurably strong."

What it looks like to take this seriously

Should I Dodge's own "most common bans per role" band is worth pointing at, because it's an example of building a feature around the fact above instead of around the illusion the per-role columns create. Its ranking logic, ChampionBanScorer, is deliberately simpler than the equivalent pick-rate ranking sitting right next to it: it sorts candidates by ban rate alone, with no blending against win rate or pick rate at all. The reasoning is stated directly in the code: "ban rate is already a direct, revealed-preference measure of how badly the playerbase wants a champion gone. Re-weighting it by win rate or pick rate would double-count the very popularity that ban rate is already expressing, not add independent information." That's the same distinction this article is making, applied to a design decision rather than an observation — ban rate isn't an input to fold into a strength score alongside win rate, because it isn't measuring strength in the first place.

The band still groups its top-banned champions by role, which is worth being precise about: that's the same real, per-role split win rate and pick rate get, filtering the ban-rate ranking down to champions eligible for that role — it is not a claim that Xin Zhao is banned more often "as a jungler" than in some other role he doesn't play. Ban rate itself never becomes a role-specific number just because the surrounding list is organized by role. The distinction matters precisely because it's easy to miss: a well-designed per-role "most banned" list and a naive per-role ban-rate column can look identical on the page while one of them is measuring something real and the other is repeating a single number under a misleading heading.

The honest scope of the claim

This isn't an argument that ban rate is worthless, and it isn't a claim that a high ban rate causes a mediocre win rate, or the reverse — nothing here establishes a causal story in either direction, only that the two numbers routinely disagree and measure different things. High ban rate paired with an unremarkable win rate is a common, ordinary pattern, not a paradox that needs explaining away: a champion can be genuinely frustrating to lose to at a rate its raw performance doesn't fully justify, and player behaviour around bans reflects that frustration directly, without needing the underlying strength to match it.

The narrower, defensible version of this whole argument is worth keeping separate from the broader one: ban rate answers "how much does this playerbase want to avoid facing this champion," win rate answers "how often does this champion actually win," and those are different questions with different answers. A per-role ban rate, specifically, answers no question at all — it's the same champion-level number printed three times, and the tenths of a point of difference between those copies are an artefact of when the page happened to be fetched, not a real signal about role. Read ban rate for what it is — a measure of how a playerbase feels about a matchup — and it's genuinely useful. Read it as a ranking of strength, and it will mislead you about exactly the champions, like Nasus, where the gap between reputation and reality is largest. The practical habit worth taking from this: when a champion's ban rate looks unusually high, check its win rate before drawing any conclusion about how strong that champion actually is. The two numbers answering different questions is exactly why they're worth reading side by side, not a reason to throw either one out.