FIDE Rating inflation/deflation

Over the years there have been many discussions here about whether there has been rating inflation. Many people point to the sheer number of GMs, the high ratings of Carlsen and the rise in the number of players rated over 2700 to argue that there must be rating inflation.

But statisticians have not found that inflation when they study the quality of play. In fact, Kenneth Regan argued about 15 years ago that there has been a slight ratings deflation over time.

Chessbase has just published a new study which analyses over 300,000 games and has determined not only that there is deflation, but has listed the likely cause.

Young scholastic players are entering the FIDE rating pool at younger ages and start with lower ratings than ever before. They tend to be underrated and steal points from more established players.

Here is a link to the latest fascinating article:

https://en.chessbase.com/post/the-elo-ratings-inflation-or-deflation

I am almost certain there is deflation.

My chess “career” has been to jump in, play 3-4 tournaments, then quit for 5-6 years and I have been doing this since the late 70’s.

I have always maintained a rating in the 1600’s.

I know my tactics at least are far far better now, because back then I had a couple of the Reinfeld 1001 tactics books, which I never finished because they were too hard and I didn’t stick around long enough.

Now I regard them all as pretty easy puzzles. So I am pretty sure I could wipe the me from the 70’s and 80’s off the board today.

According to the article, players rated 2600 today are making fewer and smaller errors than 2600 rated players 20 years ago.

Because of the way rating is scaled, if that is true for 2600 then it will be true across the rating spectrum.

It makes sense.... magnus has been the only player to be able to stay consistently above the 2800... before him it was only Kasparov who was able to stay that long above the 2800s, who as brilliant as he is, due to a lot of factors such as age an such had trouble playing caruana in 960 which requires no opening knowledge... caruana who has been #2 for a while and just fell below 2800

Well..... numbers are myth

I personally think the deflation is worth 100 points, minimum.

That is a GUESS. It’s based on subjective stuff.

I have been making the same point for a while now. What the article doesn't state, although it can be inferred from the graphs, is that the deflation in FIDE ratings since 2001 (when the floor was lowered to 1800) is a full one hundred elo points, or a bit more, based on move quality. That makes sense to me. My FIDE rating has dropped about 200 elo in that span, and while I can believe that half of it is due to old age (I'm 73), the other half is apparently due to this deflation. It all makes sense now. 

Holy cats my wild guess is backed up by the god of statistics in chess.

It makes me feel a little less bad about “barely maintaining” my own rating. It means I have been improving, which is certainly how it has felt subjectively.

“I know I am better since the last time I played in tournaments, why doesn’t my rating reflect it” has been my battle mantra for decades!

Or so I will tell myself.

While this may be true for the last 15 years, the fact remains that in the 1960s and 70s you could count on one hand, and sometimes no hands, the number of players 2700+. On the latest rating chart from FIDE, there are 38.

     That is true, but I'm pretty sure that a comparison of move quality would show that Wesley So (for example) has a smaller error-rate than Bobby Fischer (1970 to 1972), although their FIDE ratings are about the same. Computers and internet have drastically raised the level of chess. If Bobby Fischer was born in 1995 and had the same interest and access to chess as he did in the 1950s, he might be 2850 or even higher now, but if the 1972 Fischer just magically appeared now he would be at most 2700, even with some opening study (but without practice vs. comps or top modern players). 

There are also a metric ton more chess players.

I think playing skill at any “rating level” (we didn’t always have ratings of course, but if we did…) has been going up continuously since Morphy.

Players are getting better because they are learning more from those that have come before, books and coaching are more readily accessible, etc.

When I grew up in Texas in the 60’s and 70’s, chess was a desert for kids. Basically, no one in my school even played. No internet, no
clubs for kids. Very hard for me to get good.

Compare that to now. “Not having an opponent until you are 16 years old” is inconceivable today.

People are getting better, earlier.

Part of the reason I started coaching chess in elementary schools a few years ago is to live vicariously through my students…. you have no idea what I would have done to be able to enter a chess tournament for kids when I was in like 4th or 5th grade. Omg.

Now I enter them into tournaments with HUNDREDS of kids. It’s unreal. Where did they all come from lol.

Meh. ELO is not meant to compare strength between groups, but rather relative strength within a group. Pointing to elo as a metric of comparison of absolute strength of players is committing a cardinal sin of statistics.

Let’s do a thought experiment, shall we? Imagine everyone in the current rating pool’s “true” strength was 50% worse. The win percentages would be identical and thus the elo calculation would be identical, right?

You just absolutely cannot use elo to compare groups. You just can’t.

Yes exactly, that scenario is what is being discussed. In any rating bracket, we (Larry Kaufman and I, feels cool to assert that!) are saying that the playing strength has gotten stronger for the reasons outlined, plus others.

If everyone had like chimp brains you would still statistically have 2800 rated chimps, they would just play awful compared to us humans.

Then why use elo at all since it’s obviously flawed? Everyone knows that players today are objectively better. We all know that Magnus would surely beat Morphy. What we absolutely have no way of knowing is if Magnus is inherently a better player, or if it’s just a function of his tools and training. Additionally, many of Morphy’s games that have been recorded are from simuls and informal games while those are not in Magnus’ database. Morphy had a handful of books, played chess for a much shorter period, and mostly relied upon his own brain. Saying this or that person now is better than this or that person then yields trivial results.

We use elo to match up players TODAY, like in swiss tournaments.

Your question of whether Magnus is “inherently” better isn’t one I concern myself with because we have zero ways of finding this out and no way to really objectively compare this.

We can compare games they have played. We can’t compare their innate “talent”.

Also Nikki is not true that “everyone knows” Carlsen would best Morphy, In fact I think you were on another thread where that very point was being argued.

I argue that, yes, but that is because there are people on the other side who don’t believe it and think Morphy time traveled to today would dominate everyone including Carlsen, and I think even Fischer believe this.

I think of it as a hero worship thing or a “They were just better in my generation” thing. idk.

I like to point out that Fischer could learn from Morphy games. Morphy could never learn from Fischer games. Etc etc etc.

I think what I’m getting at is that at base, the question of inflation/deflation is in the minds of most a question of whether players are better or worse. Many people wrongly assert that player x is better than player y because of an elo. Then someone says that it doesn’t count because of inflation. Both are wrong. Elo ratings and differences amongst groups come down to statistical calculations based upon the composition and distributions of the groups.

In truth, whether or not elo is inflating or deflating is rather uninteresting.