Chess.com rating inflation

I've heard that there's been a big chess boom recently, does this effect the chess.com rating system? Has elo on chess.com become even more inflated than it was about a year ago?

The average rating got lower, this means that percentile for any given rating is now higher than it was before. I don't know if this has any effect on rating itself.

Hoping someone with more knowledge that me will answer because I'm also curious about this

Actually, I believe the players are getting stronger. Someone at a rating range (say 1600) a few years back, if they were to play exactly at the same level today, might not achieve the same rating as they had.

@TheDarkKinghtShow why do you think players are getting stronger?

I don't know if this is true but both online and OTB ratings are calculated by you winning against other players. More players to play = higher ratings.

Lots of chess courses and YouTube videos, Chess.com game analysis report tool, engines for analysis and much more. Combination of these things are definitely responsible for helping lots of players improve. The players are definitely more knowledgeable than before.

I was interested in this and looked into the math and ran some simulations... players joining has no effect on the skill-to-point ratio. In other words a 1500 one year ago will be a 1500 today if the only effect is players joining.

There are many other effects, but first of all, the rating system is designed to limit them as much as possible, and it does a really good job at this. Secondly, the small inflationary and deflationary effects seem to balance each other.

When in doubt, you can look at rating graphs of people who have been active for many years... however it's better to look at players like @ziryab than players like @hikaru because players at the very top don't necessarily have accurate ratings. As more strong GMs join, it would make sense for the top 50 players to gain rating, but pretty much everyone else stay the same.

There is no limit to the number of people who can have the same rating, so ratings are not changed by an already large population doubling in size (for example).

If many players gain knowledge, then those players will also gain rating. It only becomes an issue if there is not a continuum of skill or not a continuum of laziness

Sure, there are more learning tools these days, but there are still a large number of people who don't study anything, and also a large number who study a little, and then a large number who study a little more than that, etc.

A few basic ideas...

It's useful to think of it as a skill-to-point ratio.
It's also useful to think of it in terms of currency. If you travel to different countries with a lump of gold, you'll be able to exchange it for a different number of bills, but the value stays the same. In this analogy the gold is your skill, and the bills is your rating.

If someone joins the economy (chess website) and is given the correct exchange rate for their gold (given a proper rating for their skill) then they have no effect (no inflation or deflation). When two active players play, the number of points one wins will equal the number of points the other loses, so nothing changes. However when a new player joins, they can win or lose 100s of points while their opponent may only win or lose 5 points. This is a large reason why new players have no effect... because points are being created or destroyed to fit them in.

---

Examples of inflationary / deflationary effects

If a cheater joins, and wins 100 games, and then is banned, then the points they took are removed from the system... over time this changes the point to skill ratio. However all the time new players are joining, losing a few games, then quitting forever because they decide chess isn't for them. These people add points to the system.

Improvement also changes the point to skill ratio. Sure RD creates points for you if you're inactive, but people who are very active slowly gain in skill. They gain skill, but there are no points being created for them so they take them from others... however to balance this there are also active players who are getting worse over time (due to age for example) and so they're injecting points into the system.

If you draw any conclusions from my rating graph that answers the OP's question, please let me know. I don't have an answer beyond what has already been noted: averages are going down with the boom.
I am finding more new users who prefer software testing to matching wits.

I don't know enough players like you to draw any conclusions from chess.com graphs. I only know what I simulated.

Averages can go up, or down, or stay the same all without having an effect on the skill-to-point ratio. New players joining does not affect ratings.

It is true that when many below average players join, the average goes down, and that has definitely been the case over the last many years.

Here's a graph I made years ago from regularly visiting this link: https://www.chess.com/leaderboard/live

-

To be clear, I don't know whether there's been inflation or deflation over the years... I know chess.com has artificially boosted rapid and daily ratings in the past... that's likely had the biggest effect out of anything.

However over the last few years, as far as I can tell, the point-to-skill ratio has stayed the same, and in any case it is the nature of the math that new users on their own do not inflate or deflate ratings.

You can see from my graph when they boosted my rapid rating, and that the boost was higher than I deserved.

I started playing much more rapid when 10 minute chess was moved into that category. I like ten minute, but won't play it when it affects my 3 0 rating. I understand the resistance from those who think 10 0 is too fast for rapid.

IIRC higher ratings were boosted more. Looks like you gained ~250 points. IIRC some gained double that.

I didn't realize it at the time, but this was an ignorant strategy. Assuming for a moment only active players exist, then it makes no difference whether you give 1 player 10,000 points, or 10,000 players 1 point each. The final shape of your rating distribution will remain unchanged because the skill of your players hasn't changed. Adding points only alters the skill-to-point ratio.

So what can we expect from giving a player like you 250 points? We can expect that your rating will subsequently go down... which is exactly what happened... the fact that it returned back to where it started means the total number of points chess.com added wasn't enough and/or more points than they expected were lost to "inactive" players.

Ok, now bringing back inactive players... these players can trap or remove the points you add by beating players you gave points to and then not playing games... so what's the best strategy to prevent this?

The answer is simple... apply a simple horizontal linear shift i.e. add the same number of points to absolutely every user in your database. There is no drawback to this strategy. Those at or near the rating floor can be talked about, but this post is already getting long and I have reasons for why it doesn't change my answer.

What's annoying to me is if you gave this problem to anyone with any kind of engineering degree (I assume chess.com employs a non-zero number of such people), they should be able to go from knowing nothing about ratings to understanding all of this in a week or less... it took me only a few days and I wasn't working on it full time... so I don't understand how chess.com screwed it up as badly as they have in the past.

Here's @blueemu talking about the daily rating change that was quickly (1 or 2 days IIRC) reverted.

https://www.chess.com/forum/view/general/the-daily-rating-rebalancing-change-has-been-reverted?page=2#comment-56892182

-

As I said, it's grossly ignorant to add points in this way.

And yeah, on top of this to only boost active players (which is how they did it with rapid ratings as I recall)... I mean... tell me you don't understand anything about the rating system without telling me you don't understand anything. Completely embarrassing for them

Thank you for all of your comments on this. You explained it all very well, in a way that makes complete sense to both noobs and old people (me)!

Anyway... as I said before, the easy way to understand it is thinking of an economy.

Imagine an economy where they only printed money when something of value was added i.e. a traveler comes to the country with $1000 worth of gold, so you print $1000 of bills and exchange that for the gold... and when value is lost, let's say a chair is destroyed in a fire, you also destroy an equal amount of your currency.

Under such a system there is no inflation or deflation. The ratio of currency to value is constant. Easy to understand right?

And this is what RD is doing all the time. It creates and destroys rating points out of nothing... if new players are quickly given correct ratings, then new players don't cause inflation or deflation.

I was boosted 400 and it went down 300