How to read a tier list and win rate without fooling yourself
A tier list is a map, not an oracle. A win rate number summarizes thousands of games played by people who aren’t you, in matchups that aren’t yours. Once you understand what each percentage measures and what distorts it, you can use the tier list to find candidates and then let the math confirm whether the build and matchup hold up for your case.
What win rate, pick rate and ban rate really measure
The win rate is the percentage of games won with a champion out of all games played with that champion in the sample. It sounds simple, but it’s an average: it blends people who master the champion with people touching it for the first time, and all the good and bad matchups into a single figure. The benchmark is 50%: by construction, the pool of all champions tends to hover around that value, so a “neutral-looking” 52% can be good or mediocre depending on the champion and role.
The pick rate measures how often a champion is chosen (what % of games include it). It speaks to popularity and to how comfortable or flexible the champion is, not to how strong it is: a fun, easy champion can have a sky-high pick rate and a mediocre win rate, while a strong niche champion can have a low pick rate.
The ban rate measures how often a champion is banned. It usually signals frustration or perceived power more than real power: champions that are miserable to face pile up bans even when their win rate is ordinary. A high ban rate also “hides” data, because a banned champion generates no games.
The classic trap is reading these three figures as if they measured the same thing. They don’t: win rate points to outcome, pick rate to popularity, ban rate to perception. A serious tier list combines them (sometimes with a “presence” note = pick + ban) instead of sorting on win rate alone.
Illustrative example for reading the tiers: S = almost always picked, A = solid, B = situational. The real tiers shift every patch — open the tool to see the current one.
The biases that distort win rate
Sample size. With few games, win rate is noise. A champion with 40 games showing 58% may be a statistical mirage; one with 50,000 games at 51.5% is a solid signal. Before trusting a percentage, look at how many games back it: uncertainty shrinks slowly, roughly with the square root of the number of games, so quadrupling the sample only halves the margin of error.
Elo. A champion can be dominant in pro hands and a disaster in normal games, or the reverse. Demanding mechanics (hard combos, cooldown management, team dependency) gain win rate as elo rises; champions that “play themselves” tend to do better in lower ranks. That’s why a serious tier list lets you filter by rank, and a win rate with no elo context says little. If climbing the ladder is exactly your goal, see how to climb.
One-trick inflation. Hard or unpopular champions tend to be in the hands of dedicated specialists. That self-selection inflates their win rate: it’s not that the champion is strong for anyone, it’s that almost no one who doesn’t master it picks it. A high win rate with a very low pick rate is the typical signature of this bias.
New-champion dip. When a champion or rework releases, its win rate starts low: everyone is learning it at once, with no settled builds or matchups. That doesn’t mean it’s weak; it means the sample hasn’t matured. After a week or two, the number settles. The same happens, on a smaller scale, after every patch that touches it.
None of these biases makes statistics useless: it makes them something to read with judgment. Statistics are honest about what they measure (real results of real games); the mistake is asking them to answer questions they can’t.
| Games | Margin (±) | Real range |
|---|---|---|
| 10 | ±31% | 19% – 81% |
| 30 | ±17.9% | 32.1% – 67.9% |
| 100 | ±9.8% | 40.2% – 59.8% |
| 500 | ±4.4% | 45.6% – 54.4% |
| 3,000 | ±1.8% | 48.2% – 51.8% |
95% confidence interval (±1.96·√(0.25/n)). That’s why 53% over 30 games means almost nothing, and over 3000 it does.
Why a high win rate isn’t “the best for you”
A tier list’s win rate is a global average; your game is a particular case. Three things the aggregate number doesn’t know about you usually decide more than the ranking.
Mastery. A “tier A” champion you’ve played 300 times will almost always win you more games than a “tier S” one you’ve never touched. The learning curve is real: the list’s win rate assumes an average player of that champion, not you in your first 20 games. Unless the meta is badly skewed, mastery usually outweighs a couple of tier points.
Matchup. The average hides per-matchup variance. A champion at 52% overall can sit at 58% against one lane and 44% against another. If your natural rival or the counter of the moment keeps landing across from you, the global number lies to you about your reality. That’s what specific matchup tables and the matchups and counters guide are for.
Team composition. A champion’s value depends on what surrounds it: engage that capitalizes on your area damage, a frontline to protect you, a second threat to split attention. A champion can be strong “in a vacuum” and mediocre in a comp that doesn’t complement it, or vice versa.
The takeaway isn’t to ignore the tier list, but to read it as a shortlist of candidates rather than a buy order. It’s excellent for discovering what’s strong right now; it’s poor at telling you what suits you, in your matchup, with your team and your practice level.
Statistics + theorycrafting: the workflow
There are two honest ways to know what’s strong, and they complement each other. Statistics tell you what the community does: aggregates of real games, in the style of the win-rate sites. They capture the meta as it’s actually played, fads and biases included. Theorycrafting tells you what the math says: it runs the game’s real formulas over stats, damage, mitigation and scaling. It doesn’t depend on how many people play something, only on whether the numbers add up. That’s what this site’s calculators compute.
Statistics answer “what’s winning right now?”. Theorycrafting answers “why, and does it hold for my case?”. One measures popularity and outcome; the other, cause and mechanics. Using them together saves you from the two classic mistakes: following the list blindly, or theorizing a perfect build that no real games ever validate.
The concrete flow: (1) Open the tier list to spot strong candidates in your role, filtering by your elo and checking the sample size before trusting each percentage. (2) Cross-reference it with the meta to understand what kind of champions the patch rewards, and with duos if you play as a lane pair. (3) Pick two or three candidates compatible with what you already master.
(4) Verify the mechanics with the calculators: use versus to see whether your combo kills in the real matchup (with the patch’s resistances and penetration), build to compare items by damage, survivability and gold value, and the mitigation and scaling tools to understand which phase of the game you win. That way the tier list gives you the “what to try” and the calculations give you the “why it works” — with no invented numbers.
FAQ
How many games does a win rate need to be reliable?
There’s no magic threshold, but uncertainty falls with the square root of the number of games, so small samples are treacherous. With a few hundred games the margin of error is still several points; with tens of thousands, the percentage is stable. Rule of thumb: distrust any “spectacular” win rate with a very low pick rate, and always check the sample size the tier list reports before drawing conclusions.
Should I always pick the “tier S” champion from the list?
Not as a rule. A “tier S” you don’t master usually yields less than a comfort pick where you have mastery, because the list assumes an average player of that champion, not your first week on it. Tier also ignores your matchup and your composition. Use the tier list to find strong candidates compatible with your pool, not as a closed order; then confirm the matchup and build with the calculators.
What does theorycrafting add if I already have the statistics?
Statistics tell you *what* wins in the community; theorycrafting tells you *why* and whether it applies to your exact case. Statistics don’t cleanly isolate your specific matchup or evaluate a new build no one plays yet; the math does, because it runs the patch’s real formulas. Combine them: the list proposes candidates, and tools like versus and build verify whether the combo kills and whether items pay off in gold and survivability.