40 point rule

General discussions about ratings.
Reg Clucas
Posts: 602
Joined: Mon May 16, 2011 3:45 pm

40 point rule

Post by Reg Clucas » Wed Nov 27, 2013 2:35 pm

In the explanation of grading calculations, the ECF grading help page states "There is a proviso that if your opponent's grade differs from yours by more than 40 points it is assumed to be exactly 40 above (or below) yours. This applies whatever the result."

It is the final sentence in italics that I am querying. I understand the rule in respect of a win against a player more than 40 points lower than yourself, and a loss against a player more than 40 points higher. It ensures you can't lose points by winning against the lower graded player, or gain points by losing against the higher graded player.

But I don't understand why it applies whatever the result. Surely if, say, player A (grade 150) beats player B (graded 200) then player A deserves to score 250, and player B deserves to score 100? Similar logic applies to a drawn game between these two players, and only if player B wins should the 40 point rule apply.

The only possible reason I can think of for this is that it is an attempt to mitigate the effect of artificially high/low grades, but I would be interested to know if this is the case, or if there is some other reason.

Roger de Coverly
Posts: 21301
Joined: Tue Apr 15, 2008 2:51 pm

Re: 40 point rule

Post by Roger de Coverly » Wed Nov 27, 2013 2:44 pm

Reg Clucas wrote: The only possible reason I can think of for this is that it is an attempt to mitigate the effect of artificially high/low grades, but I would be interested to know if this is the case, or if there is some other reason.
It's a rule that's been around since the days when grades were done with pen and paper or mechanical adding machines. I expect there's a theoretical justification that it minimises the impact of a really freak result, but a practical effect is that you can work out a grading performance by adding up the published grades, as adjusted, and then adding in 50 times the excess of wins over losses.

MartinCarpenter
Posts: 3044
Joined: Tue May 24, 2011 10:58 am

Re: 40 point rule

Post by MartinCarpenter » Wed Nov 27, 2013 3:05 pm

Well, the whole ECF system is based on the presumption that the chance of a win/loss/draw is a linear function of the difference in strength. That clearly works quite well in practice.

However, you can easily imagine it breaking down quite badly once the grade gap gets really big. After all, even with a huge strength gap, the stronger player still has the chance of doing something terminal like blundering a piece.

I doubt if keeping the +-40 rule working even when the weaker player wins is the theoretically 'ideal' solution to this but there's definitely some motivation to do something. Once you accept that then retaining the +-40 rule for all cases is much the easiest way to do it. It'll be a pretty rare event anyway.

User avatar
Christopher Kreuzer
Posts: 8806
Joined: Fri Aug 06, 2010 2:34 am
Location: London

Re: 40 point rule

Post by Christopher Kreuzer » Wed Nov 27, 2013 3:46 pm

One effect of this is to ensure that if you have a grade, you can't go up or down by more than 90 points in any one list. Not that this greatly impedes improving juniors. Since you have to adjust the grades anyway, making the adjustment dependent on the result is trivial.

User avatar
IM Jack Rudd
Posts: 4818
Joined: Tue Apr 17, 2007 1:13 am
Location: Bideford

Re: 40 point rule

Post by IM Jack Rudd » Wed Nov 27, 2013 4:31 pm

Christopher Kreuzer wrote:One effect of this is to ensure that if you have a grade, you can't go up or down by more than 90 points in any one list.
*QI sirens*

Check E Michael White's previous posts for how you can do this. (It isn't terribly likely, of course. Nor is his scenario where six 100s can reach 200 within four years without playing anybody but each other.)

User avatar
Christopher Kreuzer
Posts: 8806
Joined: Fri Aug 06, 2010 2:34 am
Location: London

Re: 40 point rule

Post by Christopher Kreuzer » Wed Nov 27, 2013 4:39 pm

IM Jack Rudd wrote:
Christopher Kreuzer wrote:One effect of this is to ensure that if you have a grade, you can't go up or down by more than 90 points in any one list.
*QI sirens*

Check E Michael White's previous posts for how you can do this. (It isn't terribly likely, of course. Nor is his scenario where six 100s can reach 200 within four years without playing anybody but each other.)
The most recent one that was relevant seems to be at the end of this thread?

http://www.ecforum.org.uk/viewtopic.php?f=4&t=4356

Looking back over some of his posts in grading threads, I think I see what you mean, but it doesn't help that the search facility on this board ignores "40" when I try and search for that. Is that any numbers or any two-character search terms that it is ignoring?

User avatar
IM Jack Rudd
Posts: 4818
Joined: Tue Apr 17, 2007 1:13 am
Location: Bideford

Re: 40 point rule

Post by IM Jack Rudd » Wed Nov 27, 2013 4:44 pm

It's in this thread. (I searched his posts for "grader" because I knew it was in the relevant post.)

User avatar
Christopher Kreuzer
Posts: 8806
Joined: Fri Aug 06, 2010 2:34 am
Location: London

Re: 40 point rule

Post by Christopher Kreuzer » Wed Nov 27, 2013 4:56 pm

IM Jack Rudd wrote:It's in this thread. (I searched his posts for "grader" because I knew it was in the relevant post.)
Fascinating. Not that plausible, but food for thought. And the bit about the local grader was hilarious!

Roger de Coverly
Posts: 21301
Joined: Tue Apr 15, 2008 2:51 pm

Re: 40 point rule

Post by Roger de Coverly » Wed Nov 27, 2013 5:09 pm

IM Jack Rudd wrote:It's in this thread. (I searched his posts for "grader" because I knew it was in the relevant post.)
It's really just a demonstration of how a rating or grade fraud could be perpetuated. Some less scrupulous Federations or their local organisers have been alleged to have done something similar using the Elo system.

They abolished a problem for Juniors with the fix of treating them as new players every season. If you had an influx of new inexperienced adult players, the forty point rule is liable to cause their grade to be understated for a season or two if they play a lot of graded games before they have the necessary experience to gain decent results.

User avatar
IM Jack Rudd
Posts: 4818
Joined: Tue Apr 17, 2007 1:13 am
Location: Bideford

Re: 40 point rule

Post by IM Jack Rudd » Wed Nov 27, 2013 5:15 pm

Roger de Coverly wrote: It's really just a demonstration of how a rating or grade fraud could be perpetuated.
It's that, but it's also a demonstration of one of the features of the ECF grading system that isn't readily apparent to someone not familiar with it: if there's a correlation between activity and form in either direction, this will tend to be inflationary or deflationary (depending on which way the correlation goes).

Matt Fletcher
Posts: 271
Joined: Wed Mar 09, 2011 9:42 pm

Re: 40 point rule

Post by Matt Fletcher » Wed Nov 27, 2013 5:21 pm

IM Jack Rudd wrote:
Roger de Coverly wrote: It's really just a demonstration of how a rating or grade fraud could be perpetuated.
It's that, but it's also a demonstration of one of the features of the ECF grading system that isn't readily apparent to someone not familiar with it: if there's a correlation between activity and form in either direction, this will tend to be inflationary or deflationary (depending on which way the correlation goes).
The point about the hypothetical 159 grade playing lots of under-160 tournaments one season, then playing far less next season because he's now graded 174 and is less likely to win prizes to offset travel costs was an interesting one - I wonder to what extent that actually happens?

Roger de Coverly
Posts: 21301
Joined: Tue Apr 15, 2008 2:51 pm

Re: 40 point rule

Post by Roger de Coverly » Wed Nov 27, 2013 5:23 pm

IM Jack Rudd wrote: if there's a correlation between activity and form in either direction, this will tend to be inflationary or deflationary (depending on which way the correlation goes).
I would suspect the forces balance out. It might not be relevant, but why is it that some of the most active players never seem to get any better as measured by their grades? I suppose if you play every weekend, there's never the time to sit down with a database and a chess engine to figure out why you don't win 90% to 100% of the time.

Roger de Coverly
Posts: 21301
Joined: Tue Apr 15, 2008 2:51 pm

Re: 40 point rule

Post by Roger de Coverly » Wed Nov 27, 2013 5:35 pm

Matt Fletcher wrote: I wonder to what extent that actually happens?
The ECF sort of "knows" through the grading data. What it has never published is an analysis of the head count in each category as in played zero standard play Congresses, played 1, played 2 etc. The guesswork using averages, as seen in the latest ChessMoves is no real substitute for a proper analysis.

Congress organisers should avoid using the same grading boundaries for every event, and generally speaking they maintain a variety. The boundary between first and second sections is always a tricky one. If set high, you have the risk of a very small top section. Set too low and the players with grades at the bottom end of the range may feel discouraged from entering.

User avatar
Matt Mackenzie
Posts: 5206
Joined: Tue Mar 31, 2009 11:51 pm
Location: Millom, Cumbria

Re: 40 point rule

Post by Matt Mackenzie » Wed Nov 27, 2013 6:20 pm

I do believe that back in the 1980s some graders only applied the 40 point rule when the result was the "expected" one.

Neither they nor I knew at the time that they were doing the wrong thing.......
"Set up your attacks so that when the fire is out, it isn't out!" (H N Pillsbury)

User avatar
Greg Breed
Posts: 723
Joined: Thu Apr 05, 2007 8:30 am
Location: Aylesbury, Bucks, UK

Re: 40 point rule

Post by Greg Breed » Thu Nov 28, 2013 9:06 am

Roger de Coverly wrote:...They abolished a problem for Juniors with the fix of treating them as new players every season. If you had an influx of new inexperienced adult players, the forty point rule is liable to cause their grade to be understated for a season or two if they play a lot of graded games before they have the necessary experience to gain decent results.
The 40 point rule is not applicable to new players whether junior or not.
ECF Grading Database Help Page wrote:Estimating a starting Grade for an ungraded player

A Rapid grade, where available, will be used in default of a Standard grade; and vice versa. If the player has no grade at all, a starting grade is calculated as follows, using all his games in the latest three years (for adults) or one year (for juniors), inclusive of the current year.

Stage 1 is to calculate a 'grade' for each ungraded player on all his games against graded opponents in the relevant period. The 40-point rule is not used. If all his opponents are graded, it stops there and the result will be used as his starting grade.

Stage 2 brings in games between the ungraded players. Once again the 40-point rule is not used. The players are 'graded' on all their games, using as starting grades the figures obtained from Stage 1.

The resulting 'grades' will not be very accurate. So they are fed in again as new starting grades, and Stage 2 is repeated. This continues, with increasing accuracy each time, until the figures (more or less) stop changing. The starting grades can then be considered accurate.

These starting grades are then used in the grading proper.
Hatch End A Captain (Hillingdon League)
Controller (Hillingdon League)