Does your chess rating decline with age?
Yes, and the size of it is remarkably steady: about 11 points a year after fifty, 4 in your thirties, and a gain of 32 a year if you are under eighteen. Measured on the same players, across ten consecutive years, on the whole FIDE population.
Why the usual table says the opposite
The chess titles page on this site carries a table of median rating by age band, and read carelessly it says something absurd: that the median forty-year-old outrates the median twenty-year-old, and that chess players improve until middle age.
That table compares different people at one moment. A twenty-year-old with a FIDE rating has usually just arrived; a forty-five-year-old with one has usually been playing rated chess for two decades and kept going, which is a filter that removed nearly everyone who was not committed. It is a picture of who is still here, not of what happens to a person.
The question needs the same player measured twice. FIDE gives every player a permanent ID, so anyone appearing on two consecutive January lists can be matched to themselves and the difference taken — 143,735 players in the most recent pair alone.
The decline is real, gradual and astonishingly consistent
Across ten usable year-pairs the pattern does not wobble. Players under eighteen gain a median 32 points a year. The eighteen-to-twenty-nine band gains about 6. Thirty-somethings lose about 4, forty-somethings about 7, and players fifty and over about 11.
Look at the last column before anything else. Every band moved in the same direction in every one of the ten years — the under-eighteens gained in 10 out of 10, and the three older bands lost in 10 out of 10. Not one year broke the pattern in either direction.
That consistency is what makes this worth trusting more than the size of any individual number. A finding that reverses in three years out of ten is a finding about those years; a gradient that holds across a decade, on hundreds of thousands of players, is describing something structural.
| Age | Matched | Typical year | Range across years | Years gained |
|---|---|---|---|---|
| under 18 | 230,085 | +32.3 | 22 to 49 | 10/10 |
| 18–29 | 209,968 | +5.8 | 4 to 9 | 10/10 |
| 30–39 | 123,471 | -3.8 | -5 to -2 | 0/10 |
| 40–49 | 141,504 | -6.6 | -9 to -6 | 0/10 |
| 50 and over | 363,032 | -11 | -13 to -9 | 0/10 |
Within-person across the whole rated population: every player on two consecutive FIDE January lists, matched by ID and grouped by age. 10 usable year-pairs of 13 — two frozen by the pandemic and one distorted by a rule change are excluded. The most recent pair matched 143,735 players.
The crossover is around thirty
Somewhere between the twenties and the thirties the sign flips. The eighteen-to-twenty-nine band is still gaining a few points a year and the thirty-to-thirty-nine band is already losing a few, which puts the turn close to thirty.
That agrees with the companion article on peak age, which followed titled players individually and found a median peak at 33. Two different measurements, on two different samples, using two different methods, landing within a few years of each other — which is the most reassuring thing either of them has going for it.
It is worth putting the decline in perspective. Eleven points a year at fifty means a hundred and ten points over a decade — real, and far less than the collapse most players imagine. Someone who was 1800 at fifty is typically around 1690 at sixty, which is the same player having a slightly worse time, not a different one.
The three years that were thrown out
Thirteen year-pairs exist and ten were used. Two of them, 2020 to 2021 and 2021 to 2022, are the pandemic: 45.4% and 35.4% of matched players had an identical rating a year later, against about 1.2% in a normal year. They were not holding steady, they were not playing at all, and every age band comes out at exactly zero.
The third is 2024 to 2025, where the median moved 155 points in a single year with a gradient running from 87 points in the oldest band to 323 in the youngest. Playing games does not produce a gradient like that. A rule change does, and FIDE made one.
Both disqualifications are made by rule rather than by name — an unchanged share above 20%, or a median move above 50 points — so a future list carrying the same signature is excluded without anyone having to remember. The rating-inflation article throws out the same three years for the same reasons.
What this does and does not say
It measures rated results, not ability. A large part of getting worse on paper after fifty is entering fewer events, playing weaker fields and preparing less, and none of that is the same as seeing less at the board. The number is real; the interpretation is not forced.
It also cannot see anyone who stopped. A player has to be on both lists to be matched, so the people who quit between two Januaries — disproportionately those having a bad time — are invisible. If anything that makes the declines here understatements.
The figure combined across the ten pairs is a weighted mean of medians, weighted by how many players each pair contributed. That is not itself a median and the table says so. The per-year range beside it is the honest picture of how much it moves.
Common questions
At what age does a chess rating start to decline?
Around thirty. Players aged 18 to 29 still gain a median of about 6 points a year, while players in their thirties lose about 4. That agrees closely with the separate finding that titled players peak at a median age of 33.
How many rating points do you lose per year as you age?
About 4 points a year in your thirties, 7 in your forties and 11 from fifty onward — measured on the same players from one January list to the next, across ten years. Over a decade that is roughly 110 points at fifty-plus, which is considerably less than most players fear.
Do young chess players really improve that fast?
Yes. Players under eighteen gain a median of 32 rating points a year, and they gained in all ten of the years measured. The eighteen-to-twenty-nine band still gains, but at about a sixth of that rate.
Data updated 2026-08-02
Chess.com publishes no rating distribution. Those figures are derived from the Lichess distribution mapped through the conversion anchors, and are rounded accordingly.