6

I've heard often about Elo Inflation among the top chess players, meaning that a player rated 2650 Elo nowadays is weaker compared to a player rated 2650 Elo player 10 years ago. I'm curious if there's an opposite phenomenon on the lower end of the Elo bell curve:

Is a 1500 Elo rated player nowadays stronger compared to a 1500 Elo rated player 10 years ago?

I have this 'feeling' based on my personal experience but no data to back it up.

Update: I'm referring to FIDE Elo ratings.

4

These questions of rating inflation/deflation over time are very tricky to answer objectively. It is very easy to find mid-range players who will say something like "My rating today is about 200 points lower than it was 10 years ago but I'm just as strong if not stronger [spot the wishful thinking] therefore there must have been rating deflation at my level during this period". Full disclosure: I'm one of those mid range players :-)

To get more objective views (I think definitive answers are another, more complicated, matter) you have to look at how many players were above a given level today and 10 years ago and perhaps also the percentage of players who were above a given level today and 10 years ago.

That data is available via the FIDE rating website where it is possible to download rating data going back to 2001.

Here are the figures from the January 2019 and January 2009 lists:

January 2019
Players rated above 2650 = 109 (0.0335%)
Players rated above 2100 = 40940 (12.6%)
Players rated above 1500 = 224873 (69.1%)
Total rated players = 325306

January 2009
Players rated above 2650 = 75 (0.0756%)
Players rated above 2100 = 40044 (40.35%)
Players rated above 1500 = 98669 (99.4%)
Total rated players = 99232

Interpreting these figures is difficult, particularly the percentage ones.

If you just look at absolute values then there are a lot more players graded over 2650 today than 10 years ago and the same for players rated over 1500. This suggests rating inflation at both levels. I've also included values for players over 2100 because, intriguingly, these are almost unchanged.

If you look at percentage values then a much smaller percentage of players are rated over 2650 today than 10 years ago and the same for players rated over 1500. This would suggest a great deal of rating deflation at both 1500 and 2650 levels. The deflation is highest, however, at the 2100 level (out of the three levels I have looked at).

However, I think the percentage values are problematic. First if you look at the 1500 level 10 years ago there are only 563 players with ratings <= 1500. I suspect that this is because the only FIDE rated tournaments for players below 1500 were tournaments like continental and world junior tournaments going down to levels as low as 1001.

By the way, the first player I can find to have had a rating just above 1500 was the Polish player Sebastian Kobus. His first rating was 1500 in January 2006. For 1000 it was Indian player Hakeem Nikhil with a rating of 1003 in September 2012. The first sub 1300 player was French player Michel Khechab with a rating of 1260 in September 2009.

There were, of course, a lot of players playing tournament chess whose strength was below 1500 but they were all playing nationally graded chess because there simply were no FIDE rated tournaments for players that weak outside the big FIDE championship tournaments. Those figures show that today over 30% of players have a FIDE rating of 1500 or below compared to just 0.6% 10 years ago.

My tentative conclusion is that there has been some rating inflation at the level of 2650 and above because there has been a 45% increase in the number of players rated over 2650 which I think cannot be accounted for by increasing number of chess players.

I think there has been a lot of deflation in higher mid-ranges (typified by the 2100 figures) because there has been some increase in chess playing numbers which is not reflected in higher numbers at these levels.

At the lower mid-range (1500) I think the data is unreliable 10 years ago due to the very small number of tournaments open to players of these levels and so it isn't possible to draw any conclusions about this level.

Regarding the data for players at the 2100 and 2650 levels I think this is reliable because I think FIDE rated tournaments were capturing most players at these levels.

  • 1
    Thanks Brian for such a detailed answer! However IMHO the lack of data is related to some kind of a cutoff in fide rating list - I can hardly believe that there were only a handful of 1500ish players. – ThinkTank May 2 at 16:04
  • @ThinkTank 10 years ago there were lots of players of around 1500 strength however there were almost no FIDE rated tournaments for these players as I pointed out above. The rating cutoff in the FIDE lists was 1000 as it is now. Those players were playing national graded chess rather than FIDE rated, again as I pointed out in my answer. – Brian Towers May 2 at 17:13
  • That's the fact I'm not completely sure about because I recall that any player could have played in an open tournament. There were no sections but after few rounds the players were 'stratified' based on their tournament points. – ThinkTank May 3 at 7:42
  • @BrianTowers That is wrong. The cut-off has only been in 1000 FIDE points for quite a short time. Indeed, when I first got a rating in mid 2010, my first results were removed because they would have assigned me a rating below 1400. FIDE-rated tournaments rarely have any low-end threshold – David Sep 12 at 7:18
  • @David As I point out in my answer the first player to have a sub-1300 rating was Michel Khechab with a rating of 1260 in September 2009 (ratings.fide.com/id.phtml?event=639443). What you claim sounds very unlikely. Can you give me your FIDE ID so that I can verify your claim? – Brian Towers Sep 12 at 10:45
3

I think most of the inflation-deflation debate is an excuse by formerly good players who have worsened or been surpassed by younger ones over time. I don't see any reason to believe that current 2650 ELO players are weaker than their predecessors.

On the lower part of the spectrum, though, you may have a point, as lower and lower ratings get allowed into the FIDE lists, but, if anything, this should create inflation, as 1400-ish rated players now get rating for defeating 1100s, who a few years ago would've been unrated.

In any case, I don't really know if a definite answer is possible

2

Elo inflation occurs when the player base increases which allows the top players to take more points from the bottom players. The top ratings in a rating system are highly dependent on the size of the player pool. If there were only 10 rated chess players, even magnus wouldn't be much higher than around 1800 with a default rating of 1200.

Ratings are used to compare people in the same pool, and comparing ratings from two different time periods are strictly different pools. In terms of pure ranking, a 2650 10 years ago would be higher up than a 2650 now, but it does not say anything about what would happen if they played each other. It could be the case that everybody got significantly better at chess in 10 years, so the 2650 player now would beat the 2650 10 years ago. It could also be the case that people didn't improve, so the 2650 player before would beat the 2650 player now.

For in the middle ratings, there's barely any rating inflation to begin with. 1500 is still higher than the default 1200 so it would be inflated over time. However it's nowhere near as exaggerated at the end of the ratings pools. Deflation only occurs on ratings below the default rating.

  • If you're assuming a 1200 default rating, you certainly aren't talking about the FIDE rating system, which bases initial ratings on the player's opponents and not any particular default rating. FIDE didn't even go below 2000 for a long time. In any case, I disagree that "more players" is the cause of inflation. Magnus's rating would likely by well above 1800 even with a 10 player pool, too, unless you think someone rated around 1000 now would have a rating of 0. – D M Apr 25 at 9:32
  • A 2650 rated player from 20 years ago would be smashed by a current one. Just check their games – David Apr 25 at 13:25
0

When two established players play a rated game there is a transfer of rating points from the loser to the winner. Elo deflation occurs when young players gain strength and rating points from other players and then lose interest and stop organised play. They gained rating points from the pool of established players and then departed. Eventually they drift off the ratings lists and the rating points go away.

protected by Phonon Sep 12 at 11:36

Thank you for your interest in this question. Because it has attracted low-quality or spam answers that had to be removed, posting an answer now requires 10 reputation on this site (the association bonus does not count).

Would you like to answer one of these unanswered questions instead?

Not the answer you're looking for? Browse other questions tagged or ask your own question.