Chess, age and Roger Federer
In the provocatively titled “Can Anand Be The Federer Of Chess?“, Chess.com’s Mike Klein brings up the question of whether chess players can really keep performing at the top level as they age. Proponents of this view typically note that chess is not subject to intense physical stresses, that it depends a lot more on past experience than other competitive sports do, and typically end curtly with “Korchnoi.”
All plausible arguments. On the other hand, opponents like to point out that scientists typically ‘peak’ before they turn 30, judging by the age when the key work for Nobel prizes was conducted (though this age may be increasing). This is supported by a famous Einstein quip:
“A person who has not made his great contribution to science before the age of thirty will never do so.”
All well and good, but it’s perhaps not fair to compare chess to either physical sports or science, as our passion only partly overlaps with either comparison. It’s probably better to look at how chess rankings and ages have changed over time. And here, there’s undeniably been a trend towards the world’s elite becoming younger. Hardly surprising, really, given the exponential increase in availability of chess materials and resources. The more easily dispensable chess knowledge is younger players, the less older players can rely on their experience to their advantage. And that’s the key point: If younger players can more easily (and quickly) pick up the knowledge that took older players years to acquire, then the days of chess veterans dominating the charts are numbered.
A quick aside: Of the world’s top 15 as of the time of writing, only two are over 40: Former World Champions Vladmir Kramnik and Vishy Anand. Is it a surprise that Kramnik is known as a meticulous opening preparer and innovator, while Anand was one of the very first professional chess players to use computers for chess training? Of the remaining 13 players, three are in their 30s, nine are in their 20s, and the future World Champion Wei Yi is still old enough to play in junior events…
There’s a reason I highlighted some special characteristics about Kramnik and Anand. Statistics that look at the average rating of different age cohorts often make a fundamental error that econometricians like to call ‘selection bias’. The players who drop out of professional chess as they get older are not random, but instead are exactly the types who are ‘feeling their age’ with regard to their chess performance. So when we look at the average of older players who are still competing, we’ll actually be calculating the average of the better players, so we’ll always think age is having less of an effect than it actually is.
This seems like a pretty basic thing to keep in mind, but it’s shocking how often this mistake is made in research. Anyway, some economists have recently shown that once we take care of this bias, the results show that chess players’ performance indeed decreases steadily as we get older. Moreover, this decline happens much earlier than we used to think. They found that the average peak age for chess performance is a remarkably young 21.6 years, with a median around 24 years. One of the most remarkable findings from their study is that the average chess player’s level at 40 will be similar to when he or she was 15 (obviously, this assumes the player was already playing tournaments at 15). I find this very hard to believe from an intuitive standpoint, and I haven’t yet had time to go through their paper in close detail. But from a quick read, it seems like the results are solid.
As an ageing player myself, I should find this depressing (especially to think that in eight years time, I’ll get beaten by my 15 year old self). But taking a step back, the age we live in is the most exciting of all time to be a chess lover, as I explained in a recent interview. The exponential increase in chess knowledge, playing opportunities and tournament coverage (all thanks to computers) has brought the joy of the game to hundreds of millions. And if that means I have to get beaten up by a couple of kids every now and then, do I really have a right to complain?
Then again, back in my day…
Another small study pubished on my blog is http://schaken-brabo.blogspot.be/2017/02/ervaring.html (again only available in Dutch).
There you see that above 800 games players don’t improve anymore.
Still many players continue playing even beyond that number.
So many players won’t give up the game just because they passed their peak rating.
More interesting here is the fact that the majority of the players haven’t reached their peak rating yet as they haven’t played yet the 800 games.
The self-selection story is too simple. In my article http://schaken-brabo.blogspot.be/2016/08/vakantie-deel-2.html (unfortunately only available in Dutch) I point out with another small study some interesting facts again based on official data.
1) Only a minority of the players keeps playing after the age of 20.
2) However we see almost no people dropping out once they passed that filter. You can clearly see that when comparing e.g. age 30-39 of july 2009 with age 40-49 of july 2016. Both are 628 !
3) We only see a clear decline of number of players when we look at the 2 oldest groups. This can be easily explained by mortality.
So I am rather sceptical about a self-selection story linked to players willing to keep learning and peak later. The only real big selection happens around the age of 20 which has to do with big changes in your life. As a teenager you have plently of free-time and chess can be a nice pass-time. Work, marriage, kids,… often leaves you very little free-time so you ask yourself how much fun chess means for yourself. If it is not much then you just quit. Fun has no direct link with playing strength/ learning/ peak.
Very interesting study. Those numbers seem consistent with a self-selection story: the types of players who keep learning and therefore peak later are the ones who are more likely to keep playing longer.
I made a study of all active + 2500 players above 50. They peaked averagely at age 39 see http://chess-brabo.blogspot.com/2016/07/peakrating-part-2.html
That contradicts completely the study of those economists. I used official data of fide so I wonder what kind of data the economists used as I can’t access their paper without paying.
Thought you might like this diagram, https://blog.ebemunk.com/a-visual-look-at-2-million-chess-games/ Might give you ideas for future illustrations.