Why Your Win Rate Feels Stuck

How much variance actually exists in League

Each ranked game has a binary outcome: win or loss. If your true skill level corresponds to a 52% win rate — meaning you win slightly more than you lose in the long run — you still lose 48% of your games. Those losses don't happen in a neat alternating pattern. They cluster. Randomness produces streaks.

With a true 52% win rate, going on a five-game losing streak has roughly a 3% chance of happening in any given five-game window. That sounds small, but if you play 200 games across a season, you have dozens of five-game windows — meaning multi-game losing streaks happen to even improving players multiple times a season, not because something has gone wrong, but because that's what variance looks like.

Why 20 games tells you almost nothing

After 20 games, a player with a true 52% win rate could plausibly show anything between 35% and 70% — the confidence interval is enormous. The scoreboard of 20 games is not a reliable reading of skill. It's a sample that's too small to drown out noise.

The actual confidence intervals for a player with a true 52% win rate:

• 20 games: could show anywhere from 30% to 74%. Essentially meaningless.
• 50 games: 38%–66%. Still very wide.
• 100 games: 42%–62%. Starting to be directionally useful.
• 200 games: 45%–59%. Reasonably reliable.
• 500 games: 48%–56%. Now you can trust the number.

You need 100+ games before you can say with any confidence whether you're a winning or losing player. You need 300–500 before the precise number means much. Below 100 games, you're reading tea leaves.

This is why looking at your win rate after a bad week and concluding you've stopped improving is almost always wrong. One bad week — 15 to 20 games — is not enough data to conclude anything meaningful about your skill trajectory.

52% is genuinely very good

Players in ranked often feel like a 52% or 53% win rate represents a plateau — something short of actually being good. This is a miscalibration. High-elo players, people who consistently rank in the top 1% of the server, rarely sustain win rates above 58–60% over a full season. The matchmaking system is designed to push everyone toward 50% by matching players against opponents of similar skill.

A 53% win rate, sustained over 150+ games, means you are consistently winning more than you lose despite the matchmaker working against you. That is not a plateau — that is genuine positive LP pressure that will result in climbing, just not as fast as it feels like it should.

Why losing streaks feel worse than winning streaks feel good

Loss aversion is a well-documented cognitive bias. Psychologically, losses feel approximately twice as impactful as equivalent gains. In practical terms: a five-game winning streak feels satisfying, but a five-game losing streak feels catastrophic — even if both are the same length and both are driven by the same variance.

This bias causes players to draw conclusions from losing streaks that they would never draw from winning streaks. After five losses in a row, it's common to feel like everything has broken down, teammates are worse, the game is unfair. After five wins in a row, players rarely think “something unusual is happening, I should investigate.” The reality is that both streaks are usually just variance.

What to track instead of session win rate

Instead of checking your win rate after every session, track a four-week rolling trend. Are your stats improving week over week — CS per minute up, deaths trending down, vision score gradually improving? If yes, your win rate will reflect that over time. It may lag by weeks due to variance, but the underlying improvement is real and it will compound into LP eventually.

The weekly win rate graph on LOLgraphs is useful for exactly this: it shows you whether your win rate over each of the last four weeks is trending upward, flat, or declining — which is a much more stable signal than any single session or short streak.

When to actually be concerned

If your weekly win rate has been below 48% for three or four consecutive weeks, that is a signal worth paying attention to — not because it proves something is wrong, but because the sample is now large enough that it might. At that point, look at your stats rather than your record: are your deaths per game trending up? Is your CS per minute dropping? Is there a specific stat that has been getting worse for a month? Those are the questions worth asking. A losing streak alone doesn't answer them.