Can anyone tell where a player will be in 50 games?

What we measured

Every day we read the whole ranked ladder of each region. For a sample of players we took one day, everything known by then (their LP, games and win rate this season, how their LP had moved, and the games of theirs we had collected), and the first later reading at which they had played 50 more games. The question is the LP between the two.

• LP is counted on one continuous ladder: 100 LP a division and 400 a tier, so a demotion from Gold IV to Silver I counts as the same 100 LP as any other division.
• The forecast was trained on outcomes known by 13 October 2026 and judged only on forecasts made after that, the way a real forecast would be used.
• 889,608 players (a tenth of each ladder) and 62,856,468 daily readings of them, 30 August (NA) or 1 September (EUW, KR) to 13 November 2026.
• We do not show any player’s forecast, here or on profiles: a personal rank prediction is too close to an MMR calculator, which Riot’s developer policy does not allow.

LP moves a lot in 50 games, and mostly by luck

Across the three regions the average player ended 50 games 38 LP higher: the ladder drifts upward through a season. The spread around that is 167 LP, so about one player in three ended further than that from the average, up or down.

Most of that spread is luck. Even a player whose true win rate were known exactly would see about 138 LP of spread from how 50 games happen to fall. That leaves roughly a third of the variation for anything about the player to explain.

Most guesses barely beat “they stay where they are”

Extending a player’s recent trend is the worst guess, worse than assuming nothing changes: 194 LP of average miss against 133 LP. A run of wins or losses over 50 games does not carry into the next 50.

Our best forecast misses by 119 LP on average. Luck sets a floor of about 110 LP that no forecast can go below, and the forecast closes 60% of the gap between “they stay where they are” and that floor.

Who climbs: ten groups of players by forecast

All 252,788 forecasts sorted into ten equal groups. Each group ended close to what it was forecast, and the groups really do differ: 65% of the highest group climbed at least a division and 5% dropped one, against 14% and 37% of the lowest.

Even inside one group the spread is about 148 LP. A forecast moves the odds; it cannot say where anyone will land.

Early in the season LP is predictable; later it mostly is not

For players with fewer than 50 games this season, the forecast explains 43% of what happens over their next 50 (their spread is 234 LP): they are still moving toward the level their play has earned, and their first games show which way. For players past 150 games it explains 10% (spread 155 LP): their LP already sits close to their level, and what moves it is mostly luck.

So being “hardstuck” is mostly the luck above. A settled player whose LP has not moved in 50 games is usually where their games say they belong, not held there. How long a player had stayed near the same LP added almost nothing to the forecast once their win rate was known.

What does separate the climbers

Of everything in a player’s history, the forecast leaned most on how many games they had played this season and their season win rate; on whether they gained more LP for a win than they lost for a loss; and on their share of their team’s damage and gold. Account level, how many roles they play and how long they had stayed in one division added almost nothing on top.

What should a player do with this?

Judge a climb over hundreds of games, not 50: over 50 games luck alone is worth about 138 LP either way. The two numbers the client shows after every game, LP won and LP lost, say more about where you are heading than your last streak does.

What this cannot prove

• A tenth of each ladder, read once a day, over one part of one season (30 August to 13 November 2026). A new season or split resets the picture.
• Our games are the ones we collected, not every game each player played; kills, deaths and farm were not in the data.
• The signs above predict a climb; nothing here shows they cause it. Dealing more of your team’s damage may be what a stronger player does, not what makes a player stronger.
• Everything here is about groups of players. No forecast for any single player is published.