Highest Possible Chess Rating!!!

Jump to forum:
« Previous | 1 2 3 | Next » | Last Post
30th May 2008, 08:33pm
#1
by Dr_Doc_MD
Imaginationland United States
Member Since: Mar 2008
Member Points: 662

On FIDE standards, what is the highest rating possible? My guess would be, for an absolutely perfect chess player, it could go to infinity, but I have no way of actually KNOWING that.

 

Anyone have any idea of what the max could be?


30th May 2008, 08:41pm
#2
by NickyFrooj
pittsburgh United States
Member Since: Apr 2008
Member Points: 5
there is no spoon
30th May 2008, 08:57pm
#3
by Dr_Doc_MD
Imaginationland United States
Member Since: Mar 2008
Member Points: 662
What?
30th May 2008, 09:12pm
#4
by phishcake5
California United States
Member Since: May 2008
Member Points: 575
lol
30th May 2008, 09:13pm
#5
by Azoth
Ecatepec Estado de Mexico Mexico
Member Since: Aug 2007
Member Points: 339
Rybka running on a 16 core pc have an estimated elo of 3186
30th May 2008, 09:13pm
#6
by Dr_Doc_MD
Imaginationland United States
Member Since: Mar 2008
Member Points: 662
OK then... I should delete this whole topic... no answers yet...
30th May 2008, 09:55pm
#7
by likesforests
United States
Member Since: May 2007
Member Points: 3181

I don't believe there is an upper-limit. Gary Kasparov earned the highest FIDE rating that's ever been achieved... 2851. But if, by the help of some angels or demons, someone had beat him 10 out of 10 times, they would have had an even higher rating.


30th May 2008, 10:19pm
#8
by Maradonna
Scotland
Member Since: Jan 2008
Member Points: 984

When Chucky was playing like a demon ealier this month, it was said that he had a performance rating of over 3500. I may be wrong about that.

 Anyway, the point I was going to make is, that it would depend upon the rating of the opponents. If I play a 500 player on here, I don't get no cookies for a win. If I played someone 2500+ and won, I get more treats than a Krufts winner.


31st May 2008, 12:46am
#9
by silentfilmstar13
Medford, OR United States
Member Since: Sep 2007
Member Points: 2080
Dr_Doc_MD wrote: OK then... I should delete this whole topic... no answers yet...

 What a thing to say.  Why should I bother answering you?


31st May 2008, 12:53am
#10
by oginschile
Salt Lake City, UT United States
Member Since: Aug 2007
Member Points: 955

Do not try to bend the spoon, thats impossible

Instead only try to realize the truth

There is no spoon

Then you will see that it is not the spoon that bends, but only yourself


1st June 2008, 07:12pm
#11
by Dr_Doc_MD
Imaginationland United States
Member Since: Mar 2008
Member Points: 662

I recognize that, oginschile, not sure where I first heard it (it was a good while ago), and thanks for the answers, everyone. To oginschile: I suppose this means that there is no definite answer and that you can only make the answer appear through your own work (only had a minute to ponder this, so correct me if I'm wrong).

 

I've seen websites with maximum ratings of 3500, but what if two 3500's played each other and 1 consistently won? In all win-lose games, especially with players of equal or approximately equal ratings, the winner's ratings has to increase somewhat. On this basis, if there was a hypothetical maximum rating, what the heck would happen to the winner???


1st June 2008, 07:15pm
#12
by grensley
Minnesota United States
Member Since: Mar 2008
Member Points: 284
why doesn't someone go find the formula and just set things to infinity and zero where applicable?
1st June 2008, 07:28pm
#13
by Dr_Doc_MD
Imaginationland United States
Member Since: Mar 2008
Member Points: 662
There could possibly maybe be different formulas for different types of ratings, I'm not sure if there's a definite set of formulas for any type of rating, though. If anyone knows, please tell about any formulas you may have heard of!
1st June 2008, 08:03pm
#14
by grensley
Minnesota United States
Member Since: Mar 2008
Member Points: 284

O.K I looked ELO over briefly.  

ELO k=32 

Turns out that if you have more than a certain number of points over your opponent (1000, I believe) your probability of victory is 100% according to the system, and you get no points for winning.  

Therefore the conclusion is that a person can't have a rating that is more than 1000 above that of the next highest player.

It is theoretically possible (say for computer chess) that all of the "players" will rise as they become more powerful.  Since the scale stays fairly constant for humans, what will happen as we are compared to computers, and when these computers beat each other?

I think that computers will max out at around 3800, where they will have 100% at beating humans, but not at going much past each other.  It is possible that one computer will always win at this level and hit 4800.

Technically infinite however.  If a 50,000 beats a 50,000 the new ratings are 49,984 and 50,016


1st June 2008, 08:05pm
#15
by grensley
Minnesota United States
Member Since: Mar 2008
Member Points: 284

Btw, the glicko system is a nightmare to compute.  If someone wants to give that a go, hats off to you.  The basis of the glicko standard is that number of games played factors in, so there is an additional value for that.  This number represents possible deviation from the expected result due to less reliability (fewer games played).  This is further complicated by lower end threshold limits, which prevent stabilization (ratings don't stop changing, because the system prevents it to allow for improvement).  I speculate that the maximums are about the same. 


1st June 2008, 08:08pm
#16
by Dr_Doc_MD
Imaginationland United States
Member Since: Mar 2008
Member Points: 662

Thanks a lot, grensley, that clears some things up for me!

 

But wouldn't it take an extremely large amount of games to reach a rating of, say, 50000 as grensley suggested? And in that case, wouldn't rating rise extremely slowly? The rating wouldn't rise as sharply for a player of rating 50000 that has played, say, 10000 games and beat someone of the same rating, as would for one who played 1000 games and beat someone of an equal rating, so would ELO always be accurate in that manner?


1st June 2008, 08:14pm
#17
by grensley
Minnesota United States
Member Since: Mar 2008
Member Points: 284
Actually the number of games only factors in in the glicko system.  the rating change would be the same in both instances.  That is the idea behind the lower limit threshold in glicko, to prevent a stop due to a huge number of games played.
1st June 2008, 08:16pm
#18
by Dr_Doc_MD
Imaginationland United States
Member Since: Mar 2008
Member Points: 662
In that case, would the glicko system have a maximum rating? ;D Basically almost back to where we were when I started this post!
1st June 2008, 08:49pm
#19
by Niven42
West Lafayette, Indiana United States
Member Since: Feb 2008
Member Points: 291

About the Elo system:  It's a popular misconception that Elo is an acronym, like "E.L.O." - it's actually the name of the guy that invented it, Arpad Elo.

 

Anyways, based on the math, there's apparently an upper limit and a lower limit to what you can achieve, but since it's based on the population of all chess players that have a rating, it changes over a period of years, and what someone could attain at the moment is different from what was possible in the past.  40 years ago, it was unusual for a GM to have a rating over 2500, but now that is pretty common.  Wikipedia has a nice article on it, as well as describing ways to combat rating inflation (50,000 is not a plausible rating - the total population of chess players can't support that number) and why deflation is a real concern at:

 

http://en.wikipedia.org/wiki/Chess_rating


1st June 2008, 09:01pm
#20
by Niven42
West Lafayette, Indiana United States
Member Since: Feb 2008
Member Points: 291

Glicko was invented by Mark Glickman, who says at his website:

"In 1995, I created the Glicko rating system in response to a particular deficiency in the Elo system which I describe below. My system was derived by considering a statistical model for chess game outcomes, and then making mathematical approximations that would enable simple computation. The Elo system, coincidentally, turns out to be a special case of my system..."

The deficiency turns out to be the reliability factor of a player's rating, which could be very high if the player has not played a lot of games.  So Glicko is really the same thing as Elo (Mr. Glickman called it a "special case") except that Glicko corrects for the volatility of newcomers.


« Previous | 1 2 3 | Next » | Last Post

Add your comment:

Join Chess.com for free to add your comment! Already a member? Then login now to comment.