A friend is 21–9 on a new champion. A pick sits at 58% after 40 games. A team swept its last series 3–0. Each number feels like it means something, and each can be mostly luck. This guide gives you one rule of thumb for telling the two apart, then applies it to the statistics you see on StatChecked.
Every game is a weighted coin
Think of each game as a coin toss, weighted by how good a player or champion really is. A player who truly wins 55% of the time still loses 45% of games, and nobody can say in advance which ones. Over a handful of games, weighted coins produce streaks that look like skill.
Take a perfectly average player, someone whose true win rate is exactly 50%, and have them play 30 games. The chance they win 21 or more, a 70% record, is about 2%. That sounds rare until you remember how many players there are: out of 200 average players, about four will post 70% or better over 30 games by luck alone. The chance of finishing at 60% or better (18 wins or more) is about 18%, close to one in five.
The rule of thumb
Divide 100 by the square root of the number of games. The answer is roughly how many percentage points the true win rate could plausibly sit on either side of the record you see, about 19 times in 20.
- 20 games: about ±22 points
- 25 games: ±20
- 100 games: ±10
- 400 games: ±5
- 1,000 games: about ±3
- 2,500 games: ±2
Some worked examples:
- A 14–6 record over 20 games is 70%, but the plausible range runs from about 48% to 92%. The player could be ordinary or exceptional.
- A 60–40 record over 100 games is 60%, plausibly anywhere from 50% to 70%. That is promising, but not proof.
- A champion with 550 wins from 1,000 games is at 55%, plausibly 52% to 58%. Now we can say with some confidence that it is above average.
The rule is exact at 50% and slightly generous for records near 30% or 70%. For a 70% record over 30 games, the precise margin is about ±16 points rather than ±18, a difference that won’t change your conclusion.
Two consequences are worth remembering. Halving the margin takes four times the games. And fine precision is expensive: pinning a win rate down to within one point takes about 10,000 games.
Comparing two win rates
When you compare two numbers, each carries its own margin, so the gap between them is less certain than either number alone. With similar sample sizes, the margin on the difference is about 1.4 times the margin on each. Two champions at 53% and 51%, with 1,000 games each, have margins of about ±3 points apiece. The two-point gap between them is well inside the noise.
Champion statistics
Our tier list shows champions with at least 30 games in a role. At that minimum the margin is about ±18 points. That is why the tier score adds 150 imaginary games at 50% before it uses a win rate, as the tier list guide explains. With several thousand games the margin shrinks to a point or two. By then the differences that remain are mostly about who plays a champion and against whom, not luck.
Matchup tables need even more care. A matchup on a champion page appears after five games, and five games carry a margin of about ±45 points. Read a 4–1 matchup as a hint to keep an eye on, not a counter to learn.
It also helps to remember how many numbers you are looking at. Scan a table of 150 rows and a few will look remarkable by chance. The more numbers on a page, the more extreme the most extreme one will be, even when nothing unusual is going on.
Ranked records
Ranked win rates hide a twist: matchmaking tries to give you even games. Once a player reaches the rank that matches their skill, they face opponents of similar strength, and their win rate drifts towards 50%. A season win rate therefore says more about how fast someone was climbing or falling than about how good they are. A Diamond player at 50% is not worse than a Gold player at 58%. The Gold player is probably on the way up.
The same rule applies to a record on one champion. An 8–2 record on a profile is 80%, with a margin of about ±32 points.
Streaks deserve the same scepticism. A player with a true 50% win rate who plays 200 games has about a 97% chance of losing five in a row at some point, and slightly better than even odds of a seven-game losing streak. Even a 55% player has close to a 90% chance of a five-game skid over 200 games. Streaks are what randomness looks like, not proof of tilt or a rigged queue.
Short pro series
Professional matches are short samples by design. Give one team a 60% chance of winning each game and it wins a best-of-one 60% of the time, a best-of-three 65% and a best-of-five 68%. Even over five games, the weaker team wins about one series in three. Between two evenly matched teams, a 3–0 sweep one way or the other happens in a quarter of best-of-five series. A sweep alone doesn’t prove a gulf in class.
Head-to-head records are thinner still. Two teams that met twice in a season might have played five or six games, which carries a margin of more than ±40 points. That is why our Worlds model rates teams game by game over a whole season and simulates each series, rather than reading form from a few results.
A short checklist
- Look at the games column before the win rate.
- Apply the rule: 100 divided by the square root of the number of games.
- Prefer the bigger sample, as long as it comes from the same patch or meta.
- Expect extremes to fade. A player who started 21–9 will usually finish their next 30 games closer to average.
- Trust differences that persist across patches and samples, not ones that appear once.