When is the new grading list coming out?

[Alex Holowczak]

So assume a player with a grade of 100 based on performances over 3 seasons. They can now achieve 150 standard results but only play 5 games a season. Their own grade will lag their strength but it only affects 5 people a year so doesn't do much harm to the aggregate system.

But given enough of them, with the juniors playing each other in an undergraded state, the net result is to pull any adult opposition down as the juniors work their way through. As a result, you get the stretching cited by the ECF graders. This is why accurate grading of juniors is so important. When you're a junior, you can often play in entire tournaments against people of the same age. That doesn't happen when you age. So the grading points get passed around juniors, all of which are undergraded. Then any results against adults which would show a natural sign of improvement have a reduced effect, because the grade is a mean. So the juniors remain undergraded. If such a process repeats itself for long enough, everyone ends up undergraded down at the bottom of the ladder.

[Paul McKeown]

Alex,
Better still would be not to grade those for whom insufficient results are available for a reliable grade.
P.

[Roger de Coverly]

Alex,
Better still would be not to grade those for whom insufficient results are available for a reliable grade.

There is a practical issue though that some/many tournaments won't allow ungraded entrants into their rated limited sections or if they allow them in, they limit the prize money. So there's pressure to hand out grades to as many as possible. As a sort of compromise you could have a grade which entitles you to enter restricted tournaments but won't always be used as the input or 40 point base for the next season. Treating all juniors as unrated is over the top but a rule which said that E rated juniors were treated as unrated is plausible. As a modification apply it only if the current season's performance is x points better than the published grade.

In the US of A it's regarded as a financial advantage to be able to enter restricted competitions with a low rating because of the seriously interesting prizemoney. In the UK whilst there are players who have been identified as baggers of sand, it's never been totally clear whether these were strong players trying to under perform or patzers capable of talking the talk but not playing the game.

[Paul McKeown]

There is a practical issue though that some/many tournaments won't allow ungraded entrants into their rated limited sections or if they allow them in, they limit the prize money. So there's pressure to hand out grades to as many as possible. As a sort of compromise you could have a grade which entitles you to enter restricted tournaments but won't always be used as the input or 40 point base for the next season. Treating all juniors as unrated is over the top but a rule which said that E rated juniors were treated as unrated is plausible. As a modification apply it only if the current season's performance is x points better than the published grade.

Seems a reasonable compromise; I would say D and E grades should be treated as ungraded, whether junior or indeed adult, unless there is strong evidence of a previous stable grading level (e.g. adult playing less frequently than before). Don't let such grades be used to modify other grades.

[Matthew Turner]

I feel that we are sometimes falling into the same trap as the grading team and treating the grading system as an academic exercise. We are now talking about players who are improving from 100 to 150 on playing 5 games. I would have thought this was very difficult, as Richard Bates has said on numerous occasions you improve by playing.
I would suggest that if players are improving by 50 points on a few games then it is because they had a very low grade to start off with. The answer as I see it is not to start treating juniors, or anybody else, as ungraded, but to treat ungraded players as graded ones! Give juniors a nominal grade when they start, let's say 80, and let's get back to a simple grading system that works.

[E Michael White]

I guess the stubbornness of the ECF and grading team is probably down to having invested too much time and money in the iterative programs in the past and want to appear to be gaining some benefit. To put it right means starting by asking what the grading system should do as it cant meet all current expectations.

[Brian Valentine]

Here are a few statistics that tend tosupport recent comments here:

These are the average movement in grades by categories for juniors. So 2008 Category E who became category A in 2009 improved 19.6 points - Thereare only 13 players in this group

Code: Select all

[img]	2009	A	B	C	D	E
2008	A	   10.4 	    5.2 	    6.8 	-   5.3 	    3.3 
	B	   10.0 	    7.6 	-   5.1 	-   5.4 	-   7.5 
	C	    5.4 	   11.2 	    9.5 	    2.3 	-   3.6 
	D	   18.2 	   10.0 	    5.3 	    1.0 	-   1.3 
	E	   19.6 	    3.4 	   16.1 	   12.8 	-   1.3 
[/img]

Similar results for Non Juniors are as follows:

Code: Select all

[img]	2009	A	B	C	D	E
2008	A	    0.5 	-  0.9 	-  1.7 	-  0.3 	   0.1 
	B	    0.9 	   0.0 	-  1.1 	   0.9 	-  0.2 
	C	    2.6 	   0.7 	-  0.8 	-  1.1 	-  1.1 
	D	    5.2 	   0.8 	   0.4 	-  0.3 	-  1.2 
	E	    6.7 	   2.3 	-  0.1 	   0.4 	-  0.4 
[/img]

Given these players don't get the 5/10 point junior addition maybe the solution needs to be more universal.

[E Michael White]

Brians tables show what might be expected in some cases.

Adult A graders who reduce there playing activity to a B C or D grade lose grading points; D and E graders who increase their activity to become A graders increase their grading tally.

These are conditions, which unless accompanied by highly skewed data composing the averages, indicate the likely presence of internal deflation caused by a correlation between increasing activity rates and increasing performance.

For juniors, in previous years from about 1992 to 2008 I noticed an approximate relationship between grade, age and top ten/twenty.

(Age x 10) +10 usually ensured a place in the top 10 , whereas (Age x 10) usually ensured a grade in the top 20 for each age group. This year the +10 is more like +30 at some ages.

[John Upham]

invested too much money

Do you really think so?

It would be a first if true...

[E Michael White]

Do you really think so?
It would be a first if true...

Are not 1 or more of the grading team paid ?

[John Upham]

Do you really think so?
It would be a first if true...

Are not 1 or more of the grading team paid ?

I believe both Howard Grist and Richard Haddrell are paid £3k per annum.

Not sure if I'd consider that too much or not but by ECF standards it is a fortune.

[E Michael White]

Not sure if I'd consider that too much or not but by ECF standards it is a fortune.

Do you know who was paid what for developing the iterative process ? This might not have been Howard but if it has been developed over a number of years it notches up.

[Paul McKeown]

I would have thought the grading team open to a complete and fundamental overhaul: they must be surely aware that the current status is unsatisfactory. I suspect that they would not be prepared to undertake such an idea through fear that the ECF would fail to provide sufficient resources (financial, moral, management, time) for such an activity to be carried out to a successful conclusion.

[Roger de Coverly]

developing the iterative process ?

The archive on the SCCU website is one of the best sources on "history of grading".

Here's 2003

http://www.sccu.ndo.co.uk/0203/grad.htm

This is when the iterative process went completely live. ( Previously new players had been processed event by event) By a curious coincidence this is when the problem of negative grades is first raised.

With at least two different descriptions of the iterative process having been published, there's nothing to say that the 2009 process was the same as the 2003 process.

[John Upham]

I'm sure HG will be delighted to reveal the development history of the algorithm and its coding. I've asked David Thomas to have access to the source code so as I can reverse engineer it but have had no reply. I've not bothered to contact Cyril Johnson for obvious reasons.

I suspect that the rating team will go to ground on this and hope that the problems go away.

J.