"Win rate" can mean three different numbers — a champion's overall rate, its rate against one opponent, and your own record on it — and they routinely disagree, resting on very different sample sizes. Should I Dodge uses only the first, since it's the one large enough to stay stable lobby to lobby; trust your own matchup or personal numbers instead, whenever you have them.

Three different numbers, one name

"Win rate" gets used loosely to mean at least three different things, and they can disagree with each other by a lot:

  • The champion's overall win rate — wins divided by games, across every matchup and every player who picked it.
  • The champion's matchup win rate — how it performs specifically against one opposing champion in the same lane or role.
  • Your own win rate on that champion, across whatever games you personally have on it.

A champion can have a mediocre overall win rate, a strong matchup win rate into whatever it's laned against that game, and be a champion you personally have a losing record on. All three numbers are true at once. Knowing which one to trust depends on what you're trying to predict.

Which one to trust

Sample size is usually what decides it. Overall win rate is built from the largest sample, so it's the most statistically stable number, but it's also the least specific to your actual game — it says nothing about the lane you're about to play. Matchup win rate is more specific but drawn from a much smaller sample, since it only counts games where that exact pairing occurred, so it swings more on noise and is more sensitive to which patch and tier it was measured on. Your personal win rate is the most specific of all, but for most players the sample is tiny — 20 games on a champion isn't enough to separate real skill from streaks — so it's usually the least reliable of the three, right up until you have genuinely a lot of games on the pick, at which point it starts to mean more than either aggregate number.

As a rough rule of thumb, a few hundred games is usually enough for a win rate to start meaning more than noise, and a few thousand is comfortably stable. Aggregate champion win rates are built from far more games than that. Matchup win rates for common pairings often clear that bar too, but for rare pairings they don't. Personal win rates almost never do, for anyone who isn't a full-time specialist on one champion — which is exactly why they're the number most worth being sceptical of, including your own.

A hypothetical to make it concrete

Say a champion's overall win rate sits at a fairly ordinary 50%. Against one specific lane opponent, its matchup win rate might sit at 56% — a real, meaningful edge in that lane, because something about the kit genuinely answers that particular opponent. Your own win rate on the champion, over your own games, might be 40%, because you've only played it a handful of times and haven't found the timing windows yet. None of these numbers is wrong. They're answering three different questions: how does this champion do in general, how does it do in this specific lane, and how do you personally do on it. A tool that only reports the first number isn't lying to you — it's just answering the first question, not the other two.

Why this app stops at the first number

Matchup-specific and personal win rates require much finer-grained data — matchup win rates need enough games of one exact champion pairing to be statistically meaningful, and personal win rates require tracking an individual player's history rather than aggregate stats. Should I Dodge is built around the coarser, more available number precisely because it's the one with enough sample size behind it to be stable lobby to lobby. That's a real limitation, not a hidden one: a matchup or personal number, if you know it, is more specific evidence than the app's estimate, and should generally win when the two disagree.

Matchups aren't perfectly symmetric

It's tempting to assume a matchup that favours one champion by some amount disfavours the other by exactly that amount, but the two numbers aren't perfect mirrors. Scaling differences, item spikes, and playstyle asymmetries mean a matchup can be genuinely close for both sides in practice even when one champion is "supposed to" win it on paper, or lopsided even when both champions are individually strong. Treat a matchup win rate as a description of what tends to happen, not a guarantee of what will.

What this app uses

Should I Dodge uses the first number only: aggregate, overall champion win rate, the same figure discussed in what a win rate actually means. It doesn't know the specific matchup in any lane, and it doesn't know your personal history on any champion — it has no way to. What it does with that aggregate number once it has it, and how far a single champion's own figure can actually move the final estimate, is a separate question — see why a 43% champion isn't scored at 43%. Read its estimate as "how this champion performs on average, across everyone who plays it," not as a prediction about your specific lane or your specific hands. If you know your matchup or personal numbers tell a different story than the aggregate one, trust the more specific number for the decision that actually matters — the app's estimate is a starting point, not a correction to override what you already know. See how the estimate is calculated for the full picture of what does and doesn't go into it.