INflated ratings?!

Sort:
johnnyboy1287

i've also increased in points alot the last few days.  I went from 1400 Blitz to 1700.  I just thought i was getting alot better but i don't know.

Jpatrick

There are many causes of rating deflation or depression, not the least of which are waves of new accounts, all of which start out at 1200.  Compensating for this is a bit problematic.  There are other complicating factors, however they should be discussed elsewhere.

bondocel

I think that the current rating system is simply wrong. After a game, if I win let's say 8 rating points, my opponent loses only 6. In this way, all the ratings will go up. Mine went up 500 points.

It doesn't matter if a rating system is inflated or deflated. Deflated compared to what? With OTB ratings? They cannot be compared. The purpose of a rating system online is to help people select their opponents. And I think that the old system you had in place was very good and didn't need a change.

vowles_23
bondocel wrote:

I think that the current rating system is simply wrong. After a game, if I win let's say 8 rating points, my opponent loses only 6. In this way, all the ratings will go up. Mine went up 500 points.

It doesn't matter if a rating system is inflated or deflated. Deflated compared to what? With OTB ratings? They cannot be compared. The purpose of a rating system online is to help people select their opponents. And I think that the old system you had in place was very good and didn't need a change.


 Rating changes are proportionate, I thought, to both players. I loss gives, say -8 to the loser and +8 to the victor.

bondocel

No vowles, not anymore. It's like -6 for the loser and +8 to the winner. That's how all the ratings will go up.

vowles_23
bondocel wrote:

No vowles, not anymore. It's like -6 for the loser and +8 to the winner. That's how all the ratings will go up.


 Really?? :O

That sucks! Why did they change it?!

So basically whoever is a reasonable player, and plays loads of games will increase their rating.

Atos

From some reason it seems to be mostly the Bullet ratings that have increased lately. The average Bullet rating in Live Chess is now 1237, higher than the start rating. The average Blitz rating is 1138 and the average Standard 1147, still below the start rating. I don''t have an explanation for this, just pointing out the fact.

Kacparov

I agree

Atos

Also, if you look at the top of the list, there are now four players rated above 2700 in Bullet, and another 10 or so rated above 2600. The top player in Blitz is rated 2618, while the top player in Standard is rated 2488.

I suppose that it may just be because Bullet players manage to play a lot of games in a short time.

kamileon

I agree with Frankdawg. Ratings should be eight digits to show the average strength of your opponent and you should only be allowed to play players of similar averages. otherwise I could decide to play players of,say, 1000 or less. Playing enough games, I suppose I could reach 2000+ rating, which is not a true reflection of my playing strength. However, The average of my opponents would be closer. Therefore my rating as an eight digit format would be 2000/1000 ( or less). On the other hand a player, Mr A. rated 1000 may only decide to play opponents of 1300+.  Therefore his rating will be 1000/1300( could be more). If we both had played 1000 games reulting in the above results, logic dictates that if I played Mr A. then Mr A. would most probably win as he has been playing players stronger than he is. Mr A.'s game is bound to be better. Any philosophy students out there who can challenge my logic? I'm sure there are many holes. However, the logic that the current rating system, as it stands, does not reflect a genuine representation of a players true strength and that a new rating system may be required should be sound.

Atos

Hmm, I notice that even today you have lost 3 blitz games to 1600 players, but maybe they were actually 1200s as you say. Perhaps we could avoid hyperbole.

psyduck
kamileon wrote:

I agree with Frankdawg. Ratings should be eight digits to show the average strength of your opponent and you should only be allowed to play players of similar averages. otherwise I could decide to play players of,say, 1000 or less. Playing enough games, I suppose I could reach 2000+ rating, which is not a true reflection of my playing strength. However, The average of my opponents would be closer. Therefore my rating as an eight digit format would be 2000/1000 ( or less). On the other hand a player, Mr A. rated 1000 may only decide to play opponents of 1300+.  Therefore his rating will be 1000/1300( could be more). If we both had played 1000 games reulting in the above results, logic dictates that if I played Mr A. then Mr A. would most probably win as he has been playing players stronger than he is. Mr A.'s game is bound to be better. Any philosophy students out there who can challenge my logic? I'm sure there are many holes. However, the logic that the current rating system, as it stands, does not reflect a genuine representation of a players true strength and that a new rating system may be required should be sound.


I don't know about that lol. Players actually do improve, so if they were playing at a lower level for a year, then their average opponent number would be small even if they played higher opponents. A lot of ppl would keep cycling through new accounts. Also, you can't get to 2000 playing 1000 level opponents. You don't get any points beating someone 400 pts (or something) lower than you. In fact, it's easier to raise your rating playing higher rated opponents. So Mr A would have about 50/50 shot at winning. The rating system is good. Also, it's online chess, so you can't expect it to reflect any other ranking system.

kamileon
psyduck wrote:
kamileon wrote:

I agree with Frankdawg. Ratings should be eight digits to show the average strength of your opponent and you should only be allowed to play players of similar averages. otherwise I could decide to play players of,say, 1000 or less. Playing enough games, I suppose I could reach 2000+ rating, which is not a true reflection of my playing strength. However, The average of my opponents would be closer. Therefore my rating as an eight digit format would be 2000/1000 ( or less). On the other hand a player, Mr A. rated 1000 may only decide to play opponents of 1300+.  Therefore his rating will be 1000/1300( could be more). If we both had played 1000 games reulting in the above results, logic dictates that if I played Mr A. then Mr A. would most probably win as he has been playing players stronger than he is. Mr A.'s game is bound to be better. Any philosophy students out there who can challenge my logic? I'm sure there are many holes. However, the logic that the current rating system, as it stands, does not reflect a genuine representation of a players true strength and that a new rating system may be required should be sound.


I don't know about that lol. Players actually do improve, so if they were playing at a lower level for a year, then their average opponent number would be small even if they played higher opponents. A lot of ppl would keep cycling through new accounts. Also, you can't get to 2000 playing 1000 level opponents. You don't get any points beating someone 400 pts (or something) lower than you. In fact, it's easier to raise your rating playing higher rated opponents. So Mr A would have about 50/50 shot at winning. The rating system is good. Also, it's online chess, so you can't expect it to reflect any other ranking system.


 Thanks for the comments psyduck.  What you say does make sense!

vowles_23
uhohspaghettio wrote:
vowles_23 wrote:
bondocel wrote:

No vowles, not anymore. It's like -6 for the loser and +8 to the winner. That's how all the ratings will go up.


 Really?? :O

That sucks! Why did they change it?!

So basically whoever is a reasonable player, and plays loads of games will increase their rating.


 

omg, don't you get it? That is HOW they're increasing the ratings. That's a temporary thing. That's how they're inflating the ratings. 


 Yes I do get it. That is what I am saying.

bondocel

It seems this is no longer the case, but the effects of the rating change will take time to disappear. It's bad because before this change I knew that, if X has a blitz rating over 2200, most likely he's a bot.

lovablekash

Inflation is rising? Maybe I should invest while we're in a boom. Funny... I never knew chess had economics

Atos

I went out to play blitz today armed with Spaghettio's notion that today's 1600s are weaker than yesterday's 1200s and, as a result, my rating took some beating (spanking ?).

games2play

?

vowles_23
uhohspaghettio wrote:
lovablekash wrote:

Inflation is rising? Maybe I should invest while we're in a boom. Funny... I never knew chess had economics


No, inflation is falling now as the increase is levelling out. Sell!!! Sell!!!


 Haha, are we going to be entering the GCC? And yes, you can get what that stands for! :D

coairrob777

I was struggling to break 1800 in bullet, now I'm clobbering 1900s in bullet and I'm over 2000+  and it was out of nowhere this was happening.  I know I got a lot better out of nowhere in actual tournaments, but still. >_>