What is the baseline rating for someone who cannot play?

General discussions about grading.
John Upham
Posts: 4461
Joined: Wed Apr 04, 2007 10:29 am
Location: Cove, Hampshire, England.
Contact:

What is the baseline rating for someone who cannot play?

Post by John Upham » Wed Mar 04, 2009 11:08 am

For my own benefit I would like to know the expected rating under the new system of someone who cannot play chess.

This is to baseline my understanding of the inflationary measures.

For example, if an inanimate object played 30 games and lost them all against the weakest opposition would they end up with a rating of zero or would it be inflated to something higher?

J.
British Chess News : britishchessnews.com
Twitter: @BritishChess
Facebook: facebook.com/groups/britishchess :D

Roger de Coverly
Posts: 18521
Joined: Tue Apr 15, 2008 2:51 pm

Re: What is the baseline rating for someone who cannot play?

Post by Roger de Coverly » Wed Mar 04, 2009 1:15 pm

For example, if an inanimate object played 30 games and lost them all against the weakest opposition would they end up with a rating of zero or would it be inflated to something higher?
The problem is that zero doesn't represent any well-defined skill level. It's recorded in both the development of the Elo system and the Clarke system, that the numbers are essentially arbitrary and that both developers worked backwards from "master" level. So IM might have been about 2400 on the Elo and 225 on the Clarke. The players that IM's score 75% against would be 2200 Elo and 200 Clarke etc. Equally the systems could have been based around 3400/ 325 = IM without changing anything.

Rating systems attempt to measure "strength" by using results as a proxy. It's arguable whether you get sensible measurements for players well down the skill ladder.

The Clarke system chain is that a player of zero should score 10% against a player of 40 who scores 10% against a player of 80 who scores 10% against a player of 120 etc.

Looking at the top end of the chain, it's my contention that IM/GM's have got better. Therefore to score 10% (190 players) against GMs , you had to improve, and to maintain 10% against those who score 10% against GM's (150 players) you also needed to improve. Those that didn't improve lost grading points.

Thus the scale stretches out between novice and GMs because the GMs have got better. The effect of the ECF's changes is to attempt to compress the range between GMs and novices. If the range was correct all along, the net effect over time will just be to inflate the GMs grades because the current relative distances will be restored.

Sean Hewitt

Re: What is the baseline rating for someone who cannot play?

Post by Sean Hewitt » Wed Mar 04, 2009 1:27 pm

An interesting theory Roger. Is there any evidence to support this?

Supposing your theory is correct, why havent GMs grades gone up though? If a 240 GM has gotten better and instead of scoring 75% vs a 215 player he is now scoring 85% (say) - why isnt his resulant grade 250?

John Upham
Posts: 4461
Joined: Wed Apr 04, 2007 10:29 am
Location: Cove, Hampshire, England.
Contact:

Re: What is the baseline rating for someone who cannot play?

Post by John Upham » Wed Mar 04, 2009 1:45 pm

Sean Hewitt wrote:An interesting theory Roger. Is there any evidence to support this?

Supposing your theory is correct, why havent GMs grades gone up though? If a 240 GM has gotten better and instead of scoring 75% vs a 215 player he is now scoring 85% (say) - why isnt his resulant grade 250?
Could it be that most players of 220+ do not give a stuff about their Clarke rating and only care about their FIDE rating?
British Chess News : britishchessnews.com
Twitter: @BritishChess
Facebook: facebook.com/groups/britishchess :D

Roger de Coverly
Posts: 18521
Joined: Tue Apr 15, 2008 2:51 pm

Re: What is the baseline rating for someone who cannot play?

Post by Roger de Coverly » Wed Mar 04, 2009 2:10 pm

An interesting theory Roger. Is there any evidence to support this?
Well the opinions of the GMs themselves - Matthew Turner wrote something to the effect that GM's are better than they used to be. Also it's a reasonable explanation as to why some players have maintained the same grades and relative rankings for twenty years or more whereas others ("the lazy ones") have declined.

In this season's Torquay tournament I scored 2.5/5 against a field of (154,196,162,160,175) for a 169 perf.

All 5 opponents were in the 1994 list as well (173,204,165,161,180). Only one of these grades (the first) is materially different between 1994 and 2008. This was a player who didn't play between 1996 and 2003 who evidently was unable to get back to his previous standard.


That standards have improved is an empirical observation that a certain amount of work is needed to stay in the 170-180 range. You need to know your openings to a greater depth than you used to. If you look at my last round from Uxbridge, you will see a game where my 140 opponent knew or found a tactical trick in the Sicilian that I was blind to.
If a 240 GM has gotten better and instead of scoring 75% vs a 215 player
It's inherent in the theory that the 215 player had to get better as well to avoid being crushed quite so often.
why isnt his resulant grade 250?
With all the extra points injected into the system with the ECF's changes, that's what I suspect will happen and why the relative changes "2007 existing" to "2008 existing" versus "2007 revalued" to "2008 revalued" should be published.

carstenpedersen
Posts: 47
Joined: Fri May 23, 2008 10:20 am

Re: What is the baseline rating for someone who cannot play?

Post by carstenpedersen » Wed Mar 04, 2009 2:36 pm

The Clarke system chain is that a player of zero should score 10% against a player of 40 who scores 10% against a player of 80 who scores 10% against a player of 120 etc.
If I've understood the whole debate correctly, the problem is that players of zero were scoring 18.4% against players of 40 who were scoring 18.4% against players of 80 who were scoring 18.4% against players of 120 etc, and the recalculated grades are designed to make your "should score 10%" work again.

I do agree the best players are stronger than they used to be, which is presumbably why their grades are now 260+ as opposed to 240+ 35 years ago. :wink:

Roger de Coverly
Posts: 18521
Joined: Tue Apr 15, 2008 2:51 pm

Re: What is the baseline rating for someone who cannot play?

Post by Roger de Coverly » Wed Mar 04, 2009 4:01 pm

the problem is that players of zero were scoring 18.4% against players of 40 who were scoring 18.4% against players of 80 who were scoring 18.4% against players of 120 etc, and the recalculated grades are designed to make your "should score 10%" work again.
They have never produced any evidence that directly showed this effect. In any event, if players of 80 scored 18.4% against players of 40 and players of 120 scored 18.4% against players of 80 and everyone played a field equal to their rating, there would be no effect since players of 80 would gain the points against those of 120 that they lost against those of 40. You would only see an effect on players right at the bottom (since they only play people above them) and those at the top (since they only play people below them).
I do agree the best players are stronger than they used to be, which is presumbably why their grades are now 260+ as opposed to 240+ 35 years ago.
As far as the domestic scene is concerned, I would regard the likes of Mark Hebden ( 236) as top of the food chain, rather than Short or Adams at 260 plus because they only play a couple of UK events a year. Is Mark as strong today as Miles, Nunn, Keene or Hartston 35 years ago? I suspect yes not least because of 35 years worth of opening theory.

John Hickman
Posts: 195
Joined: Sun Sep 07, 2008 8:35 pm

Re: What is the baseline rating for someone who cannot play?

Post by John Hickman » Wed Mar 04, 2009 8:42 pm

John Upham wrote:For my own benefit I would like to know the expected rating under the new system of someone who cannot play chess.

This is to baseline my understanding of the inflationary measures.

For example, if an inanimate object played 30 games and lost them all against the weakest opposition would they end up with a rating of zero or would it be inflated to something higher?

J.
You could test this theory by entering some inanimate objects into tournaments. I guess their problem is their tendency to lose on time a lot.

Hmm, lets see, enter them in a tournament with an odd number of players, give them a bye in the first round, they then lose a lot, get a 1 point bye, play each other where the one with White loses on time, and the other one goes on to play someone in the last round whos mobile goes off to reach a respecable 3.5 points and two rated wins!

E Michael White
Posts: 1362
Joined: Fri Jun 01, 2007 6:31 pm

Re: What is the baseline rating for someone who cannot play?

Post by E Michael White » Wed Mar 04, 2009 10:04 pm

John Upham

Your scenario needs more precise definition. As described it has a number of mathematical and metaphysical difficulties. For instance:-

Is there more than one inanimate object playing 30 games per year ?
Are the games played in leagues or congresses ?.
Is there a junior object club where many inanimate objects play each other and each object plays against just one live graded player ?
Do inanimate objects have mobile phones ?
Do any already have bona FIDE or mala FIDE grades ?
To be graded the inanimate object has to be registered with the ECF, what dates of birth are given to obtain the junior increments ?

Incidentally the ECF will be able to publish photographs of objects, dates of birth and other objectional details because the Data Protection Act only applies to living people.

A more interesting scenario is where absolute zero is defined as a player who has just learnt the moves and makes random legal moves very quickly. Come to think of it ..........

User avatar
John Saunders
Posts: 1255
Joined: Thu Apr 19, 2007 3:10 pm
Location: Kingston-upon-Thames
Contact:

Re: What is the baseline rating for someone who cannot play?

Post by John Saunders » Thu Mar 05, 2009 1:58 pm

What a pity Jan Hein Donner is not alive and a contributor to this forum. Had he been so, his answer to the question posed by this thread would surely have been "whatever Lod Prins' rating is at the moment".
Personal Twitter @johnchess / Personal Website http://www.saund.co.uk / Britbase http://www.britbase.co.uk

Stewart Reuben
Posts: 4101
Joined: Tue Apr 03, 2007 11:04 pm
Location: writer

Re: What is the baseline rating for someone who cannot play?

Post by Stewart Reuben » Sun Jun 21, 2009 6:05 pm

E Michael White is quite correct. To find the true base line you need to program a computer to play random legal moves. Then enter it into tournaments and found out what its Rating (FIDE) and Grade (ECF) is, measured against established players. They would have to be young children, older players would not put up with such nonsense.
It would certainly be lower than 0 ECF and possibly lower than 0 ELO. USCF Ratings go down to 300 I believe.
Negative grades have been used in the ECF. They are highly undesirable if published. Very young children do not understand the concept of negative numbers.
Stewart Reuben

Ian Thompson
Posts: 2307
Joined: Wed Jul 02, 2008 4:31 pm
Location: Awbridge, Hampshire

Re: What is the baseline rating for someone who cannot play?

Post by Ian Thompson » Sun Jun 21, 2009 8:52 pm

Stewart Reuben wrote:E Michael White is quite correct. To find the true base line you need to program a computer to play random legal moves. Then enter it into tournaments and found out what its Rating (FIDE) and Grade (ECF) is, measured against established players. They would have to be young children, older players would not put up with such nonsense.
It would certainly be lower than 0 ECF and possibly lower than 0 ELO. USCF Ratings go down to 300 I believe.
Negative grades have been used in the ECF. They are highly undesirable if published. Very young children do not understand the concept of negative numbers.
Stewart Reuben
I think there are easier ways to the same end:

1. You could take all tournaments where there were an odd number of players and therefore a bye. You work out what the grading and rating performance of the bye would be if he was a real person who had lost all those games.

2. You work out the grading and rating performance of a player who loses all his games in many fictitious tournaments against graded players.

Out of curiosity I spent 5 minutes doing 2. for the Surrey Border League this season. There were just over 1000 real games submitted for standard play grading this season. I copied all these games twice and changed the white player to a beginner in the first copy and the black player to a beginner in the second copy. I changed all the results to losses for the beginner. The beginner 'played' and lost just over 2000 games against opponents with grades ranging from 18 to 240. The ECF Result File Checker program tells me his grading performance was 26. Using the average conversion formula from old to new grades, this would make the beginner's new grade 65.

Obviously, if the beginner did the same thing again in subsequent years his grade would go down by at least 10 points each year, heading towards minus infinity.

Mike Gunn
Posts: 729
Joined: Wed Apr 11, 2007 4:45 pm

Re: What is the baseline rating for someone who cannot play?

Post by Mike Gunn » Sun Jun 21, 2009 10:10 pm

The problem with this beginner/ inanimate object idea is that if he plays in another league or set of tournaments he will end up with a different grade. This tells us something quite important about grading systems.

Stewart Reuben
Posts: 4101
Joined: Tue Apr 03, 2007 11:04 pm
Location: writer

Re: What is the baseline rating for someone who cannot play?

Post by Stewart Reuben » Sun Jun 21, 2009 10:52 pm

That is a deeply flawed system. The player's start grade would depend on the average grade of his opponents and be -50 from that, or possibly -40. That is why in FIDE the player must score at least 1 point in their first game.
It must be against extremely weak opponents yet who have a grade. Then what makes you think that a player playing random legal moves will lose all his games? Theoretically the random player could play perfectly. To play 30 perfect moves would only be about 30 to the power 30. Quite a small number in terms of Britain's likely debt measured in pence.

Stewart Reuben

Kevin Thurlow
Posts: 3396
Joined: Wed Apr 30, 2008 12:28 pm

Re: What is the baseline rating for someone who cannot play?

Post by Kevin Thurlow » Mon Jun 22, 2009 12:22 pm

"That is why in FIDE the player must score at least 1 point in their first game."

Er, should that be first tournament? I'm not complaining, just confused!
"Kevin was the arbiter and was very patient. " Nick Grey

Post Reply