Can an engine calculate how fast Human GM lose ratings after 55 ?
Posted: Tue Jul 20, 2021 2:11 pm
by Chessqueen
Has anybody calculated precisely with a software Engine how many Elo Human GM lose after the age of 55 ? Here is the Link or Website ==> 16 Years old Praggnanandhaa will demolish the 58 old GM
Re: Can an engine calculate how fast Human GM lose ratings after 55 ?
Posted: Tue Jul 20, 2021 7:08 pm
by Uri Blass
Not all GM's lose rating after age 55
I do not see that the following GM lost rating after age 55(he was 55 years old in 2014)
You just flippantly ignored the fact that this GM lost about 140 points between 45 and 55.
He gained some back since he was 55, but he's still far from the rating he had almost 20 years ago.
Re: Can an engine calculate how fast Human GM lose ratings after 55 ?
Posted: Tue Jul 20, 2021 10:33 pm
by mvanthoor
lkaufman wrote: ↑Tue Jul 20, 2021 8:38 pm
At age 63 my fide rating was about the same as it averaged over the previous forty years. Only after that I started to decline about 20 elo per year.
I've noticed that "lower" rated GM's (the ones that are between 2600 and 2700 their entire life) seem to decline less rapidly, compared to the >= 2700 ones, and less severely. I wonder if someone like Jeff Sonas officially researched that at one time or another.
Re: Can an engine calculate how fast Human GM lose ratings after 55 ?
Posted: Wed Jul 21, 2021 1:47 am
by DrCliche
It's certainly no mystery that not-quite-world-class athletes (and chess players, etc.) have the potential to maintain a relatively consistent level of performance as they age, whereas the best of the best do not.
Obviously absolutely no disrespect meant to Larry Kaufman (or anyone else), but as he "wasn't that good" to begin with, even as he aged he was still capable of exerting himself sufficiently to maintain his 99.999th percentile (or whatever) rating.
As a point of comparison, probably less than one in a 100,000 people can run a sub-5-minute mile, yet elite 60 year olds can do it! But they're still more than a minute slower than the absolute best, and probably slower than the fastest high schooler in your city.
Anyway, you don't need an engine to figure out how fast you might expect people to lose rating as they age. Just plot the highest rating ever achieved at each age. That's about as good as anyone can be, and everyone else is likely to deteriorate somewhat more quickly, depending on their effort and interest.
Re: Can an engine calculate how fast Human GM lose ratings after 55 ?
Posted: Thu Jul 22, 2021 6:32 pm
by MikeB
In summary, the good news is that you only get old once. ;>)
Re: Can an engine calculate how fast Human GM lose ratings after 55 ?
Posted: Thu Jul 22, 2021 11:58 pm
by Branko Radovanovic
I wonder if rating loss has something to do with one's playing style. If I remember correctly, Karpov once said something to the effect that Kasparov's style, based on tactics and calculation, could not be maintained at a high level for long, and was the main reason behind his early retirement.
Also: did Karpov's preference for positional play effectively prolong his career?