I play on both lichess and chess.com and I have noticed that either I just play better on lichess or the rating here are inflated quite a lot. Can anyone tell me roughly how much inflation is there???
It is because of the initial rating. I guess on chess.com it is 1200 and on lichess.org it is 1500. Beside you have played much.
Rating inflation doesnt exist because ELO or any similar system is not an objective measurement of strength. The rating pool is accurate to the players and conditions existing within each pool system.
OMG Thank you Willy_Wonka. You explained that very well. Of course, you might notice no one will realize why that is so correct.
Rating inflation within a pool does exist based on natural factors within the system when you add or subtract from the given pool. For example: Games. If your average number of games increases over time, you can find a ratings inflation depending on how the increase effects areas either geographically or in specific rating groups. This is over long periods of time in stable systems.
An example of a ratings deflation is an increase in players where they start off low and rapidly gain experience over time. USCF had this recently when they had a large influx of children join the USCF. The average rating went down, and there was a noticeable deflation over a 20 year time period.
The misconception comes when you start believing that the number presented is a strict representation of strength, and then you believing that all numbers should match regardless of pool. When in reality, you don't compare outside the pool, and you have to realize that it's not strict representation of strength. It's relative representation based on past performance. And it's purpose is to get an idea of gradual improvement over time, and to pair "like" skilled opposition.
Rating inflation within a pool does exist based on natural factors within the system when you add or subtract from the given pool. For example: Games. If your average number of games increases over time, you can find a ratings inflation depending on how the increase effects areas either geographically or in specific rating groups. This is over long periods of time in stable systems.
An example of a ratings deflation is an increase in players where they start off low and rapidly gain experience over time. USCF had this recently when they had a large influx of children join the USCF. The average rating went down, and there was a noticeable deflation over a 20 year time period.
The misconception comes when you start believing that the number presented is a strict representation of strength, and then you believing that all numbers should match regardless of pool. When in reality, you don't compare outside the pool, and you have to realize that it's not strict representation of strength. It's relative representation based on past performance. And it's purpose is to get an idea of gradual improvement over time, and to pair "like" skilled opposition.
Gradual improvement over time is not a good usage of a rating since they are relative values only to other current active ratings in the same pool. Comparing ratings from different times, even in the same pool, runs into similar problems as comparing ratings from different pools. Mainly because the pools *are* actually different from one time to another.
Simply imagine a pool where the ratings are deflating over a period of years (as the one you mentioned). If a player is getting slowly better over that time period, his rating might actually even *decrease*.
Simply imagine a pool where the ratings are deflating over a period of years (as the one you mentioned). If a player is getting slowly better over that time period, his rating might actually even *decrease*.
What I mean by gradual improvement over time is your own personal ruler of improvement.
Here is the basis:
Lets say you are an active player. We can use Lichess.
You play daily. You rating easily gets established in 3-4 days. It is stable between 1200-1400.
Now lets say you train daily. You find that with this training you are starting to see a rating increase of approximately 100 points.. Because it's based on 100 point system to be somewhat accurate. So now you are three months training and your average results are in the 1300-1500 range. But you keep training.. 4 months later you notice you're winning more and you can average 1500-1700. You might not notice the improvement right away, but then you look back on your play when you were two classes lower and understand why you were in that category.
This is common. And it's how ratings are suppose to be used as an improvement tool. You don't simply spout, :Well I am a 1740 player. You're a 1560 player. I should beat you because of that. You use it as a tool to gauge general improvement over time. You're not going to get a 1700 rating for nothing, just like you're really not going to get a 1200 rating for nothing. If you stay 1200 or 1700 for many months or many years, you are not improving.
Honestly though I am unsure why I had to explain that. Maybe you read too much into the statement?
If a player is slowly getting better over a period of ratings deflation which is about a 20 year period, he is not slowly improving if he decreases. Sorry.. You must gauge improvement based on activity not based on flaws of a system over long periods of time.
I can give you mine as an example I suppose from USCF.
I was in 2005 a 1200 player. I played at around a 1500 level at the time. I found out because I won several tournaments and boosted my rating very quickly. I consider that an inactive status to renew activity period. I played in about 8 tournaments a year to find out what my strength was. It settled around 1700 for a while. Then I trained a little and it took a sharp boost to 1850. Then again to 1980. This did this in overall a matter of about 4 years. This is a good measurement of improvement over time in USCF because the activity is a lot lower than online play. You can track similar improvements faster in lets say "Lichess". Where I took 4 years to make this track, you can track this in months on here.
Here is the basis:
Lets say you are an active player. We can use Lichess.
You play daily. You rating easily gets established in 3-4 days. It is stable between 1200-1400.
Now lets say you train daily. You find that with this training you are starting to see a rating increase of approximately 100 points.. Because it's based on 100 point system to be somewhat accurate. So now you are three months training and your average results are in the 1300-1500 range. But you keep training.. 4 months later you notice you're winning more and you can average 1500-1700. You might not notice the improvement right away, but then you look back on your play when you were two classes lower and understand why you were in that category.
This is common. And it's how ratings are suppose to be used as an improvement tool. You don't simply spout, :Well I am a 1740 player. You're a 1560 player. I should beat you because of that. You use it as a tool to gauge general improvement over time. You're not going to get a 1700 rating for nothing, just like you're really not going to get a 1200 rating for nothing. If you stay 1200 or 1700 for many months or many years, you are not improving.
Honestly though I am unsure why I had to explain that. Maybe you read too much into the statement?
If a player is slowly getting better over a period of ratings deflation which is about a 20 year period, he is not slowly improving if he decreases. Sorry.. You must gauge improvement based on activity not based on flaws of a system over long periods of time.
I can give you mine as an example I suppose from USCF.
I was in 2005 a 1200 player. I played at around a 1500 level at the time. I found out because I won several tournaments and boosted my rating very quickly. I consider that an inactive status to renew activity period. I played in about 8 tournaments a year to find out what my strength was. It settled around 1700 for a while. Then I trained a little and it took a sharp boost to 1850. Then again to 1980. This did this in overall a matter of about 4 years. This is a good measurement of improvement over time in USCF because the activity is a lot lower than online play. You can track similar improvements faster in lets say "Lichess". Where I took 4 years to make this track, you can track this in months on here.
in chess.com the ratings are more and more closer to fide ratings. lichess.org are VERY infleted.
look at me,for rapid games 15+10 in lichess.org i've a ~1700 rating....in chess.com (where i've another nickname, FabioFI17) im about ~1500 player.
look at me,for rapid games 15+10 in lichess.org i've a ~1700 rating....in chess.com (where i've another nickname, FabioFI17) im about ~1500 player.
Yes, and on chess.com, you'll quickly notice that my rating is below 1500, and here, 1900+
Yes yes.. You did happen to see how inflation doesn't exist for those reasons right @ScaccoLento from Willy's post? Those aren't just random words. They are truth. You can't compare ratings that way. It isn't possible.
sigh
This topic has been archived and can no longer be replied to.