A champion's win rate hides more than the number suggests: who actually plays it (skill self-selection inflates hard champions), how recent the patch is, which tier it's pulled from, and how much a low pick rate weakens the sample. That's why Should I Dodge clamps every win rate to 30-70% — anything more extreme is almost always small-sample noise.
What the number is
A champion's win rate is wins divided by games played, over some window of ranked matches pulled from a third-party League of Legends statistics provider. It's a real, measured number — not a guess — but it's a summary of a huge number of very different games compressed into one percentage, and a lot gets lost in that compression.
Sample composition
The players who pick a given champion aren't a random sample of the player base. A mechanically demanding champion is disproportionately picked by players who've put in the practice to play it well, which inflates its measured win rate relative to how it would perform in the hands of an average player. A simple, forgiving champion doesn't get that same boost. The win rate reflects who plays the champion, not just the champion.
Patch recency
Balance changes shift win rates within days, and the effect of a big buff or nerf on a niche pick with a small player base takes time to fully show up in the data — early adopters skew the sample before the wider population has caught up. A win rate pulled right after a patch is a noticeably noisier read of "how strong is this champion right now" than that same number would be three weeks later, once the sample has had time to settle.
Tier skew
The same champion can have a meaningfully different win rate depending on which skill tier the data is filtered to — a pick that rewards mechanical precision tends to look stronger at higher tiers, where players can actually execute it, and weaker lower down. A win rate pulled from the wrong tier is answering a slightly different question than the one you're asking. See how tier affects the spread of these numbers for more on this.
Pick rate changes what a win rate can tell you
A 54% win rate built from a champion picked in 2% of games, and a 54% win rate built from a champion picked in 20% of games, are not equally trustworthy signals. The low pick-rate number rests on a smaller, more self-selected sample — often a narrow group of specialists — and swings harder on small-sample noise. The high pick-rate number is averaged over a much broader, more representative slice of the player base, so it takes a stronger real effect to move it. Two identical-looking percentages can carry very different amounts of evidence.
One number can span multiple roles
Some champions are played in more than one role, and their published win rate is often a single blended figure across all of them, even though performance can differ a lot role to role. A champion that's strong in one role and mediocre in another shows up as one average number, which understates how good it is in its best role and overstates how good it is in its worst. If you know a champion is being played somewhere unusual for it, that's a reason to treat the aggregate number as less directly applicable, not more.
Queue type matters too, in a smaller way — win rates pulled from solo queue and from flexible team queues reflect somewhat different player behaviour and team coordination, even for the same champion on the same patch. Most aggregate figures don't separate the two, which is one more reason to read a single percentage as a useful approximation rather than a precise measurement.
Why the number is still worth looking at
None of this makes an aggregate win rate useless — it makes it a somewhat rougher instrument than the single decimal point on the page suggests. Across a large enough sample, sample composition and patch effects tend to average out for most champions most of the time, and a win rate that's persistently well above or well below 50% over a stable patch is still telling you something real. The caveats above matter most exactly when a reading is close to the edges of plausibility — a very high or very low number, a very low pick rate, a reading taken right after a big patch — which is when it's worth asking why, rather than taking the percentage at face value. This app's own per-tier calibration data gives one concrete way to make that judgment — see how to tell a real win-rate gap from ordinary noise.
Why the app clamps every win rate
Should I Dodge clamps every champion's win rate to a range of 30% to 70% before using it in the win estimate. A raw win rate outside that band is almost always a small-sample artefact — a rarely-picked champion with a handful of games can post a 65% or 35% win rate that says more about sample size than about the champion — and letting an outlier like that swing the estimate unchecked would make the whole calculation less honest, not more precise. See exactly what that clamp bounds and why, or how the rest of the calculation works for the fuller picture.