What is a good chess rating?

The short answer, and why it is a percentile

There is no rating that is good in isolation. A number only means something once you know how many people are below it, which is why every answer on this page is a position in a population rather than a threshold to clear.

The table below is measured, not estimated. It counts every Lichess player with an established blitz rating — 657,474 of them — and reports where each band of 200 points falls. Nothing here is modelled: Lichess publishes the distribution, and this is it, grouped.

The result that surprises people: 1400 sits at the 42nd percentile and 1600 at the 61st. A 1500 blitz player is almost exactly average, and 1500 is the number most beginners are told to aim at as though it were a distant achievement. It is the middle of the room.

Read down the right-hand column and the shape of the problem appears. The four bands from 1000 to 1800 hold about two thirds of everybody. Almost everyone playing chess online is in a band with tens of thousands of near-identical neighbours.

Why the middle is so crowded

A rating system is a feedback loop. Win and you are given harder opponents; lose and you are given easier ones. Left running long enough that loop drives everybody toward the rating at which they win about half their games, and since most players are of broadly similar strength, most ratings pile up in the same place.

The consequence is that equal rating gains are not equal achievements. Climbing from 1400 to 1600 means overtaking roughly 18% of all rated players. Climbing the same 200 points from 2000 to 2200 overtakes about 7%. The second climb is worth far more and feels far slower, and both of those facts have the same cause.

It also explains the plateau nearly everyone describes and few expect. You do not stop improving when your rating stops moving. You reach the part of the curve where improving enough to pass another thousand people takes a great deal more than it did to pass the previous thousand.

What we measured

How many people you are actually being compared with

A percentile is only as meaningful as the pool it is drawn from, and these pools are wildly different sizes. That matters more than it sounds: being in the top 5% of a pool of 39,000 and the top 5% of a pool of 657,000 are not the same claim, even when the two ratings look similar.

Blitz is where almost everyone is. Lichess classical has a small fraction of the blitz pool, which is why classical ratings there behave oddly at the edges — there are simply not enough players in some bands for the number to settle.

The FIDE pools are smaller than any of them, and they are not a subset of the same people. A FIDE rating requires playing rated over-the-board games in an organised tournament, which is a filter far narrower than making an account. That is the single most important thing to hold on to before comparing an online rating with an official one.

Good for whom

If the honest question is whether you are good at chess, the honest answer is that the population you are measured against is almost entirely people who chose to play rated chess online. Against everyone who knows how the pieces move, any established rating puts you a long way up.

If the question is whether you are good relative to people who take it seriously, the bands above 2000 are where that conversation starts, and about a tenth of the rated online population is there.

And if the question is whether you are improving, the rating is a poor instrument on any timescale shorter than a few hundred games. It moves for reasons that have nothing to do with you — who you were paired with, what time of day it was, how long since you last played. The percentile is the stable part. Watch that instead.