Relationship of # of games played and rating?

I'm really curious if there has been any analysis of the relationship (if there is one) between the number of games played and rating. Like an equation for the line of best fit or something?

Or even some info like "the average rating of people who have 1000 games played is XXXX and the average rating of people at 2000 games is XXXX"

The reason I'm asking is i'm wondering how i'm stacking up compared to people with comparable number of games. Of course I dont have all my games played on chess.com but I have a general idea of how many i've played.

An interesting question. I think number of games is insufficient though, because there is a world of difference between a slow game (say 90 30 time control) and a blitz game. A more accurate metric would probably be some kind of measure based on how much time you've spent playing chess, with time spent playing slow chess weighted more heavily than blitz chess, which would in turn be weighted more heavily than bullet.

While that would be ideal I would think that would be hard to measure. I know you can't really sum up all the time a person has spent so I just thought the easy to find parameters "number of games played" and "rating" would suffice.

Even though we can't control for those other things, I still think a relatively accurate positive correlation would arise.

I wish there was a magic number to reach because right now I feel like I'm just haphazardly moving my pieces.  I just started playing because my teen daughters want to join chess club at school.  I know how to play chess but I've never really played chess seriously.  Right now I feel like the best thing I can do is play as many games as I can and learn from every game I lose.

I've always played like this and, in the end, it's just a game.  I just play for fun and improve while doing so. Playing more definitely is equatable with better perfomance.

No one has collected any sort of stats? Not even any chess.com staff?

I'd guess this comparison would only be relevant for the first few hundred games and the first hundred or so rating points (after whatever beginner would be).  After a person gets used to the game it takes practice to get better not just playing games.  In fact playing hordes of games can have the opposite effect if you're merely practicing mistakes over and over.

There are a lot (you could easily find 50 I'd bet) of players on Yahoo and FICS with over 30,000 games who are also at or below 1500 OTB strength.  The same level I think many could reach in a year or two if they worked hard / had a coach / free time and desire / all that stuff.

Why would you think that there is a linear relationship between number of games and rating? If that applied, the world champ would be an old guy.

What I don't get about this site is how common it seems that a member has an online rating way above their live rating. You would think that there must be players with low online ratings, but I never seem to come across them

Ratings don't measure your raw chess skill but your chess performance against other players in the same set.  So chess.com online ratings, chess.com blitz ratings, FICS ratings, over the board ratings etc will all be different because the player is being compared to a different set of players each time.

I didn't say it would be linear at all levels and I didn't say it would have to accurate at the upper and lower ends.

You've played 627 games on chess dot com.  How many or how much have you played OTB? 

If you just compare people with a lower number of games to people with a higher number of games, you’ll likely end up with some selection bias. If after 500 games you’re rated 1600, then you’ll be more likely to want to keep playing than if you were rated only 1200. A better method would be to compare the same players rating at different points. So for instance, look at only people with 5,000+ games. Look at their rating after 500, 1000, 2500, and 5000 games.

I've played two otb tournaments with about 4-5 games each, I also played a some games in highschool (all of them except one player were incredibly weak). The bulk of my playing was on FICS where i have 1505 games played

See how they're connected?

Not sure what you are saying, but if you're pointing out the fact that we can't get all the information i already addressed that

"While that would be ideal I would think that would be hard to measure. I know you can't really sum up all the time a person has spent so I just thought the easy to find parameters "number of games played" and "rating" would suffice.

Even though we can't control for those other things, I still think a relatively accurate positive correlation would arise."

As far as highest number of games played on chess.com this is for online games http://www.chess.com/echess/players.html?sortby=total_game_count

For blitz http://www.chess.com/livechess/players.html?sortby=total_game_count

for bullet i have played against someone who has played more than 86000 games and that was some months back.

For live standard i have come across atleast 4 people who have played more than 15000 standard games the highest being more than 17000 games.

Generally these people who play lot of games don't take active part in the forums so we don't know them.

If you take a look at those two lists you will find players with all sorts of rating so obviously number of games and rating are not related. 

But how could chess.com keep statistics involving number of games played if they only know how many games were played on chess.com?  e.g. as you said the bulk of your games so far haven't been here.

In my other post I was saying if they simply took the number of games played  and their rating then a correlation would arise. You were supposed to assume that they took it from chess.com games because them taking all games into account is too hard to do.

That would screw up the stats though whenever they were looking at an ID of a non-beginner.  I.e. a person who has gained playing strength through games not on chess.com.  Especially considering they use the glicko rating where new player's ratings can gain hundreds of points per game.  e.g. a 2000 rating with 20 games played.

Yes I realize there would be people who screw with the trend but i still think i somewhat accurate trend would exist. I said this in my quote.