Go back
Call for ratings floor revisited

Call for ratings floor revisited

General

R
Out of drinks

On Clique Beach

Joined
06 Feb 05
Moves
64036
Clock
15 Nov 05
Vote Up
Vote Down

The recent nonsense involving the poor sportsmanship of TRACKHEAD entering a low banded tournament has convinced me to revive my call for Russ to implement a rating floor. See my previous site idea thread at http://www.redhotpawn.com/board/showthread.php?threadid=22232

I realize I'm not the first to suggest this. I was annoyed at the time of the original thread by what I felt was unsporting behavior by jira, who after a long absence was in my group in a banded 1400-1599 tournament. See http://www.redhotpawn.com/tournament/view.php?tid=524

jira has predictably gone on to win that tournament without a loss (he has won though it's not officially complete) and has a current rating of 2230. While jira was quiet and TRACKHEAD is public and obnoxious about this, it's the same unsporting behavior. Since we can't count on people to police themselves (and skeeter has suggested they shouldn't be expected to), I think it's time to solve this (and a few others issues in the process).

What I've suggested is that everyone, in addition to having a rating, also have a new statistic called a "rating floor". The rating floor will be far more stable than their rating (it generally will be 200 points below the highest rating ever). Only when your actual rating falls below your rating floor, your rating floor will be used to:

1) Determine whether you are eligible for banded tournaments.
2) Calculate the rating change for your opponent at the conclusion of a game.

An added benefit is that this will tend to stabilizing ratings. The best rating system is one which bears the closest correlation to reality. When you play someone who has an artificially low rating, the resulting rating change for you is inaccurate no matter the game result.

Two questions I'd anticipate:

Question 1) So, that means my rating floor would never decrease?

Answer 1) Generally no. But it is possible someone might end up with an artificially high rating floor. You might have brain damage which permanently reduces your chess abilities. Or a bunch of people on RHP might play a practical joke on you and all resign their games with you at the same time. If this became a problem, and petitioning Russ with reasonable requests to lower your rating floor was impractical, I
have a scheme in mind to allow some rating floor "decay" if you continue to play below it.

Question 2) What about the resulting "ratings inflation"?

Answer 2) If ratings inflation results, there are a few simple things to be done to mitigate the effect. However, I have reason to suspect that any ratings inflation would be negligible. Currently (without this scheme) I now have 9 months worth a actual data showing there is no ratings inflation. yet there is the occassional thread complaining about what people perceive as evidence of inflation. It's probably time for me to update my thread on ratings analysis with the latest data. The reason I believe we would not see much inflation is that the gain/loss of points to the system is influenced to a far greater degree by the gain/loss of players. Hundreds of people join RHP every day, adding 1200 points into the system. Dozens stop playing, some permanently (thereby taking their points with them), some temporarily (locking their points away). The flow of people and their points will dominate the occassional inequitable flow of ratings points due to a floor.

I know this belong in site ideas (where the original appeared), but I've posted it here so get the reaction of a larger audience.

-- the supposed newbie

a

THORNINYOURSIDE

Joined
04 Sep 04
Moves
245624
Clock
15 Nov 05
Vote Up
Vote Down

Whilst a good idea in theory, due to the way that ratings are calculated if someone with a 2200+ rating suddenly dropped to sat 1200 due to timeouts then entered a rated tournament for say players rated 2000+ then each victory would bring them a large points increase and also, unfairly penalise the player they beat who would lose a lot of rating points.

In theory if three players were to be at say a rating of 1200 or so, and they all won against a 2000 rated player, playing as black and white then that player rated at 2000 would lose a ton of points and this would not be justifiable.

It really boils down to an individual players morals and unfortunately some people have none.

😠

BigDogg
Secret RHP coder

on the payroll

Joined
26 Nov 04
Moves
155080
Clock
15 Nov 05
Vote Up
Vote Down

Originally posted by adramforall
Whilst a good idea in theory, due to the way that ratings are calculated if someone with a 2200+ rating suddenly dropped to sat 1200 due to timeouts then entered a rated tournament for say players rated 2000+ then each victory would bring them a large points increase and also, unfairly penalise the player they beat who would lose a lot of rating points. ...[text shortened]... really boils down to an individual players morals and unfortunately some people have none.

😠
For this reason, the rating should not be allowed to drop below the floor. At least, not until a significant number of games are played at a performance at or below the floor level.

In other a words, a 2200+ (established) should not be allowed to fall below 2000. If they get stuck on that floor long enough, then it can be dropped to 1800, and so on.

R
Out of drinks

On Clique Beach

Joined
06 Feb 05
Moves
64036
Clock
15 Nov 05
Vote Up
Vote Down

Originally posted by adramforall
Whilst a good idea in theory, due to the way that ratings are calculated if someone with a 2200+ rating suddenly dropped to sat 1200 due to timeouts then entered a rated tournament for say players rated 2000+ then each victory would bring them a large points increase and also, unfairly penalise the player they beat who would lose a lot of rating points. ...[text shortened]... really boils down to an individual players morals and unfortunately some people have none.

😠
No. If Player A, a 2200 player dropped to 1200, then beat some 2000+ players, their resulting ratings get calculated as if they lost to a 2000-rated player (2000 being the rating floor for Player A).

My ratings floor suggestion totally *prevents* the problems you describe.

a
Enola Straight

mouse mouse mouse

Joined
16 Jan 05
Moves
12804
Clock
16 Nov 05
Vote Up
Vote Down

This should be on the top of the upcoming additions to the site. It's an excellent idea.

The only problem I see is the possibility for rating inflation. Suppose a 2200 starts timing out a bunch of games and drops to his rating floor of 2000 for example. If he continues to time out games, people will gain points, without any being lost by the player in question, thus increasing the total number of points in circulation without adding more players.

A
D_U_N_E

Arrakis

Joined
01 May 04
Moves
64653
Clock
16 Nov 05
1 edit
Vote Up
Vote Down

Originally posted by adramforall
Whilst a good idea in theory, due to the way that ratings are calculated if someone with a 2200+ rating suddenly dropped to sat 1200 due to timeouts then entered a rated tournament for say players rated 2000+ then each victory would bring them a large points increase and also, unfairly penalise the player they beat who would lose a lot of rating points. ...[text shortened]... really boils down to an individual players morals and unfortunately some people have none.

😠
Take a look at fierytorment's profile. I got matched up with this bozo when he was rated around 1275. If you check his stats you'll see that he resigned a ton of games and dropped his rating from near expert down to around 950!

I personally recommend that nobody play this jerk who informed me that he's a USCF expert. He might be rated as an expert... but what gives him the right to drop 1000 points here and start playing with a 950 rating!?

X
Cancerous Bus Crash

p^2.sin(phi)

Joined
06 Sep 04
Moves
25076
Clock
16 Nov 05
Vote Up
Vote Down

Originally posted by RookRAK
The recent nonsense involving the poor sportsmanship of TRACKHEAD entering a low banded tournament has convinced me to revive my call for Russ to implement a rating floor. See my previous site idea thread at http://www.redhotpawn.com/board/showthread.php?threadid=22232

I realize I'm not the first to suggest this. I was annoyed at the time of the original ...[text shortened]... d), but I've posted it here so get the reaction of a larger audience.

-- the supposed newbie
Everytime a player is far below their rating floor wins a game points enter the RHP pool (as less points are lost than won). This could lead to rating inflation (which despite some claims to the contrary isn't occuring now). This would need to be addressed before I would be happy with introducing a rating floor.

A
D_U_N_E

Arrakis

Joined
01 May 04
Moves
64653
Clock
16 Nov 05
Vote Up
Vote Down

Originally posted by XanthosNZ
Everytime a player is far below their rating floor wins a game points enter the RHP pool (as less points are lost than won). This could lead to rating inflation (which despite some claims to the contrary isn't occuring now). This would need to be addressed before I would be happy with introducing a rating floor.
Who cares what you think? You've already established yourself as an imbecile kid who knows nothing, criticizes everybody, and thinks only your ideas are any good. We don't need people like you.

R
Out of drinks

On Clique Beach

Joined
06 Feb 05
Moves
64036
Clock
16 Nov 05
Vote Up
Vote Down

Originally posted by XanthosNZ
Everytime a player is far below their rating floor wins a game points enter the RHP pool (as less points are lost than won). This could lead to rating inflation (which despite some claims to the contrary isn't occuring now). This would need to be addressed before I would be happy with introducing a rating floor.
Point inequalities are introduced into the system today, every time players with different K values play, or when provisional plays non-prov.

My hypothesis was that the inflation would be negligible. But even if it wasn't, there an inelegant solution. Every New Years Day Russ could compute the total inflation, and dock everyone the overage. Happy New Years, you all lose 2 points.

Like I said, many more points come and go via players every day, than would be accounted for by rating floor inaccuracies.

T
Total Domination

Wilmington, NC

Joined
11 Jul 04
Moves
24569
Clock
16 Nov 05
Vote Up
Vote Down

Originally posted by XanthosNZ
Everytime a player is far below their rating floor wins a game points enter the RHP pool (as less points are lost than won). This could lead to rating inflation (which despite some claims to the contrary isn't occuring now). This would need to be addressed before I would be happy with introducing a rating floor.
I agree with Xan here. Plus there are just too many other problems with this. It's nice on paper, although so is communism. The main problem is the one mentioned by the first response. And then your reply saying the 1200 player who was formally 2000 will gain gain points as if it were two 2000 players playing, then that wouldnt be fair to that player. There are jusr far too many variables here. Personally I think the entire rating system should be changed but you guys voted on something like that and killed it off.

R
Out of drinks

On Clique Beach

Joined
06 Feb 05
Moves
64036
Clock
16 Nov 05
Vote Up
Vote Down

Originally posted by TRACKHEAD21
...The main problem is the one mentioned by the first response. And then your reply saying the 1200 player who was formally 2000 will gain gain points as if it were two 2000 players playing, then that wouldnt be fair to that player.
No, either you misread or I was unclear. Remember the goal is not to "punish" players, but instead for ratings to be closest to reality.

Player A rating = 1200, rating floor = 2000 (so he's like really 2200)
Player B rating = 2000, rating floor = 1800 (he's at his actual rating)

If Player A wins:
Player A gains points based on a 1200 beating 2000, i.e. we want that gain because we want Player A to approach his real rating
Player B loses as if 2000 losing to 2000

If Player B wins:
Player A loses as if 1200 losing to 2000
Player B gains as if 2000 beating 2000

T
Total Domination

Wilmington, NC

Joined
11 Jul 04
Moves
24569
Clock
16 Nov 05
Vote Up
Vote Down

Originally posted by RookRAK
No, either you misread or I was unclear. Remember the goal is not to "punish" players, but instead for ratings to be closest to reality.

Player A rating = 1200, rating floor = 2000 (so he's like really 2200)
Player B rating = 2000, rating floor = 1800 (he's at his actual rating)

If Player A wins:
Player A gains points based on a 1200 beating 20 ...[text shortened]... Player B wins:
Player A loses as if 1200 losing to 2000
Player B gains as if 2000 beating 2000
thats the part of why I was agreeing with Xan. Too many points being brought into the system out of nowhere. That would cause a huge inflation in the ratings. I think this is somewhat happening already but something like that would make it far worse.

g

Joined
02 Feb 05
Moves
28913
Clock
16 Nov 05
Vote Up
Vote Down

Originally posted by RookRAK
The recent nonsense involving the poor sportsmanship of TRACKHEAD entering a low banded tournament has convinced me to revive my call for Russ to implement a rating floor. See my previous site idea thread at http://www.redhotpawn.com/board/showthread.php?threadid=22232

I realize I'm not the first to suggest this. I was annoyed at the time of the original ...[text shortened]... d), but I've posted it here so get the reaction of a larger audience.

-- the supposed newbie
I agree with you on that. and he started a thread on the tournament forum stating that all tourneys are ruined because someone higher in ranking is playing them.

Ragnorak
For RHP addons...

tinyurl.com/yssp6g

Joined
16 Mar 04
Moves
15013
Clock
16 Nov 05
Vote Up
Vote Down

Originally posted by arrakis
Take a look at fierytorment's profile. I got matched up with this bozo when he was rated around 1275. If you check his stats you'll see that he resigned a ton of games and dropped his rating from near expert down to around 950!

I personally recommend that nobody play this jerk who informed me that he's a USCF expert. He might be rated as an expert... but what gives him the right to drop 1000 points here and start playing with a 950 rating!?
And his options would be???

D

Suzianne
Misfit Queen

Isle of Misfit Toys

Joined
08 Aug 03
Moves
37387
Clock
16 Nov 05
Vote Up
Vote Down

Originally posted by TRACKHEAD21
thats the part of why I was agreeing with Xan. Too many points being brought into the system out of nowhere. That would cause a huge inflation in the ratings. I think this is somewhat happening already but something like that would make it far worse.
These ratings points aren't "coming from nowhere" since (using the previous example) the now 1200 player *should* be 2200 anyway. That's the whole point of this system.

And you forget that there is already a DEflation of points in the system by the 2000 (now 1200) player beating the stuffing out of the 1400-1700 players. The current system promotes the flow of points going from the 1400-1700 players to the 2000 players just so they can get back to where they were. It trashes the ratings of these 1400-1700 players for nothing because when all is said and done, the 2000 player is now 2000 and the only one who has lost points are the 1400-1700 players the 2000 player screwed to get back where he was.

Since there is no way to *punish* these apes doing this, then RookRAK's suggestion is a good way to prevent the wholesale *taking* of points from the 1400-1700 crowd for nothing.

But then maybe it would be harder for the high levels to win tournaments on the backs of the mid levels. It would also eliminate the widespread sandbagging going on in clan matches.

Yes, I can see why the high levels might not like this plan.

Cookies help us deliver our Services. By using our Services or clicking I agree, you agree to our use of cookies. Learn More.