New Grades

Request amendments or pass comments on this free service.
User avatar
Carl Hibbard
Posts: 5618
Joined: Fri Dec 08, 2006 8:05 pm
Location: Evesham

New Grades

Post by Carl Hibbard » Sun Aug 31, 2008 2:33 pm

The start-of-season revision which includes the new grades amendment for next year has just gone live
Cheers
Carl Hibbard

John Upham
Posts: 4130
Joined: Wed Apr 04, 2007 10:29 am
Location: Cove, Hampshire, England.
Contact:

Re: New Grades

Post by John Upham » Sun Aug 31, 2008 4:10 pm

Carl,
Can you list the 10 biggest gainers and the 10 biggest lowest(!) gainers please?

I've already spotted a 57 -> 113 transition for a thireteen year old whose previous two ratings were 57 & 54.

This is a huge rise c/w with her county team mates. She should now be on board one!

john

Howard Grist
Posts: 82
Joined: Tue Dec 12, 2006 1:14 pm
Location: Southend-on-Sea
Contact:

Re: New Grades

Post by Howard Grist » Sun Aug 31, 2008 8:42 pm

John,

The 11 people with the largest differences between their New and old standard play grades are:-

253399G Chan, Isla
259038E Powney, Elliot
252776F Tso, Benedict
253552L Eyles, Alan
268794L Marsh, Matthew
256553F Linsley-Parrish, James C
271011A Rajaratnam, Thushaan
243382F Thrumble, Milo
252115F Mitchell, Jason F
263764K Tothill, Christopher
263795K Gryn, Isaac

who all have differences of 73 or more points.

If you bother to look these people up you will discover that they are all under 12 and don't have very many graded games.

As 'everyone knows', juniors are all significantly under graded, and my investigations actually bore this out. This could also be seen as the cause of the deflation of recent years. Juniors actually need to be over graded so that the system does not deflate - the junior enhancements are currently the only way that points can be put in to the system.

I'm not sure what a biggest lowest gainer is, but if you want to know who comes off worst by comparing the New and old grades, then I suggest that you look at the top player list.

Howard
ECF grading system programmer
ECF Grading System Programmer

John Upham
Posts: 4130
Joined: Wed Apr 04, 2007 10:29 am
Location: Cove, Hampshire, England.
Contact:

Re: New Grades

Post by John Upham » Sun Aug 31, 2008 11:10 pm

Howard Grist wrote: juniors are all significantly under graded, and my investigations actually bore this out.
Is this actually true and scientifically accurate?

I'm sure there are many juniors (u-18) whose rating has not improved or has worsened.

Contrariwise there are many who are improving (their rating is increasing at least) at a large rate of knots.

I'm not sure I'd be so bold as to claim that " juniors are all significantly under graded".

I guess HG is closer to the statistical / deterministic coal face than the rest of us.

John

Paul Stimpson
Posts: 109
Joined: Mon Feb 25, 2008 10:52 pm
Location: Essex
Contact:

Re: New Grades

Post by Paul Stimpson » Sun Aug 31, 2008 11:41 pm

Something doesn't seem right to me!

Just looked at a bunch of Essex Junior Graded Players I know really well and the grades are just boggling.

I am seeing players jump 40 points, now if I could see that all players have moved accordingly then I might just believe
what I am seeing, but if this is to correct a deflation problem then it blows out of the water the rough model we have in our heads over the years
to what a 100 or a 140 or a 160 or a 200 player would be.

These new grades don't look correct to me, especially when compared to adult grades. Looking at the revised grades and the grades of known adult players
and the results I know they would get against these players, it looks like mass over inflation.

Also the new increment on Junior Rapid grades also looks small to me. I know it's compounded but 6 points over a year for an u11 but 15 at Standard grade?? That just seems fishy. The majority of u11's play lots of rapidplay and I would suggest perhaps the increment should be linked to number of games played. 3 points for those who play 5 games or less, but those playing 30+ ought to be getting 10 at least surely.

Also after following this grading debate over the last year and the need to re-adjust, I find it strange that in the end retrospective grading has occurred. I am somewhat bewildered like John that established known and very much real strength order is now revised out of thin air. Players who were No1 are now 5th or 6th
in the ranking, something looks like it might have gone wrong here somehow.

carstenpedersen
Posts: 46
Joined: Fri May 23, 2008 10:20 am

Re: New Grades

Post by carstenpedersen » Mon Sep 01, 2008 11:59 am

The 10 biggest "losers" are

264038H Zhu, Chen F D 233 239 -6
246407L Cramling, Pia F C 233 239 -6
259049K Georgiev, Kiril M D 259 266 -7
248317J Stefanova, AntoanetaF D 239 246 -7
245630J Fridman, Daniel M D 238 245 -7
137069J Devereaux, Maxim LM B 216 223 -7
118942G Short, Nigel D M D 254 262 -8
263277K Kobalia, Mikhail M D 250 258 -8
261296D Ghaem Maghami, EhsanM D 233 241 -8
127577L King, Daniel J M C 229 238 -9

Howard Grist
Posts: 82
Joined: Tue Dec 12, 2006 1:14 pm
Location: Southend-on-Sea
Contact:

Re: New Grades

Post by Howard Grist » Mon Sep 01, 2008 1:02 pm

John Upham wrote:
Howard Grist wrote: juniors are all significantly under graded, and my investigations actually bore this out.

Is this actually true and scientifically accurate?
Under the current system Juniors were, on the whole, out-performing their grades. The New grades correct this, which is why juniors have increased more than a similarly graded adult has.
Paul Stimpson wrote:Also the new increment on Junior Rapid grades also looks small to me
The Junior increment performs two functions. One is to acknowledge that juniors are, on the whole, improving. The other is to put points in to the system that are lost due to the deflation that occurs 'naturally' due to the fact that the better players play more often. The composition of the standard and rapid play lists are very different. The rapid play lists consist of many more juniors, and junior games make up a far higher percentage of the total number of games. This result of this is that the component of the increment required to stop the system deflating is much smaller.
ECF Grading System Programmer

Roger de Coverly
Posts: 17255
Joined: Tue Apr 15, 2008 2:51 pm

Re: New Grades

Post by Roger de Coverly » Mon Sep 01, 2008 1:22 pm

the fact that the better players play more often.
I've seen this stated before. This was in the context of trying to overturn the argument that it shouldn't matter if a 170 player scored 168 against 160's because he would score 172 against 180s.

Could the terms "better" and "often" therefore be defined please? There are any number of players with grades below mine who play more - equally there are players with higher grades that play fewer games.

carstenpedersen
Posts: 46
Joined: Fri May 23, 2008 10:20 am

Re: New Grades

Post by carstenpedersen » Mon Sep 01, 2008 1:47 pm

Roger

Not that hard to do. From the downloaded text file I get:

Category........ Average "old" grade.........Average Adjustment
A..................127.4 ..........................21.5
B..................107.7............................24.6
C.................105.6.............................25.1
D.................101.8.............................26.3
E..................96.6..............................27.3

A preadjustment difference of 30 points (and 24 postadjustment) between the most and least active categories I think confirms the assertion that stronger players tend to play more.

Roger de Coverly
Posts: 17255
Joined: Tue Apr 15, 2008 2:51 pm

Re: New Grades

Post by Roger de Coverly » Mon Sep 01, 2008 2:05 pm

A preadjustment difference of 30 points (and 24 postadjustment) between the most and least active categories I think confirms the assertion that stronger players tend to play more.
I think your statistic just shows that active players have higher ratings. Not a total surprise since you need to retain some degree of match fitness to get decent results. Do you want to support an entire theory of grading deflation on those averages?

On a different topic, does anybody know whether the "new" grades are going to be checkable by players? By this I mean the ECF grading printout sent to some categories of direct members. Last year I dropped 2 points from 174 to 172. The "new" grade came out at 180. Was this a rise or a fall?

Howard Grist
Posts: 82
Joined: Tue Dec 12, 2006 1:14 pm
Location: Southend-on-Sea
Contact:

Re: New Grades

Post by Howard Grist » Mon Sep 01, 2008 2:34 pm

Roger,

Carsten's stats illustrate what I was referring to with my comment. An alternative one is that if you take the arithmetic mean of all players' individual percentages over the course of a season it comes to just under 40% - not the 50% that the grading system is expecting. With more active players having higher grades and, perhaps more significantly, an increase in activity usually resulting in an increase in grade, the result is that the system would deflate unless you did something to prevent it.

You ask if there is any way to 'check' your New grade. As mentioned in the explantory text, New grades have been calculated going back to 2006 and these figures could be published. However, wouldn't doing this simply move the problem back to 2006?
ECF Grading System Programmer

carstenpedersen
Posts: 46
Joined: Fri May 23, 2008 10:20 am

Re: New Grades

Post by carstenpedersen » Mon Sep 01, 2008 2:40 pm

I think your statistic just shows that active players have higher ratings.
Which, to me, is the same as saying that stronger players are more active.

The question you originally asked was
Could the terms "better" and "often" therefore be defined please?
My interpretation would be:

Better = Higher grade
Often = plays more games per year.

Therefore I would take Howard Grist's original "the fact that the better players play more often." to mean that, on average, higher graded players play more games than lower graded players.

All I've shown is that this statement is correct, which you seemed to doubt.

I had no intention to make any judgement regarding the overall issue of deflation and adjustments.

Carsten

User avatar
IM Jack Rudd
Posts: 3706
Joined: Tue Apr 17, 2007 1:13 am
Location: Bideford

Re: New Grades

Post by IM Jack Rudd » Mon Sep 01, 2008 2:48 pm

I checked my new grade. Exactly the same as the old grade.

All my Barnstaple team-mates' grades have gone up a fair amount, though.

Roger de Coverly
Posts: 17255
Joined: Tue Apr 15, 2008 2:51 pm

Re: New Grades

Post by Roger de Coverly » Mon Sep 01, 2008 2:54 pm

You ask if there is any way to 'check' your New grade. As mentioned in the explantory text, New grades have been calculated going back to 2006 and these figures could be published.
My grade has gone up by 8 points. I'd like to know where the extra points came from. Empirical checks on some of my local opponents suggests that those who keep their 175 ish grades by bashing 140s and 150s have gone up by more than those who score 50% against a 175 field or 25% against a 200 field. I also suspect without proof that you've injected a lot of inflation into the system. The results have been cooked so that for example Jack Rudd is unchanged at 215 but if you increase the grades of people that the likes of Jack play and beat then the top end will inflate. By way of a small scale example, I scored 2.5 from 6 in my most recent tournament for an exact par score of 172. On the new grades it was 182 which is +2 on my "new" grade.

David Robertson
Posts: 1951
Joined: Tue Apr 03, 2007 6:24 pm
Contact:

Re: New Grades

Post by David Robertson » Mon Sep 01, 2008 3:03 pm

Doesn't seem quite right to me either. It's not just eye-popping jumps in junior grades. At my own club, there are some frankly laughable increases lower down among adults. One very weak player leaps from 31 to 67. He hasn't won a game in three years, and I can assure you, 31 wasn't deflated. Another rockets from 95 to 121; again 95 was nearer the mark. I'm all for supporting the science in these matters, but it's hard when one's credibility is tested.

For example, as professional policy-makers will tell you, it's not the scientific facts, but the relativities that count. People do not assess themselves against groups far removed from them; they make comparisons with those of similar characteristics. Take salary levels. People on middle incomes worry far less about wage hikes for the low paid, but fret hugely if the bloke next door, or in the next office, gets a wage rise. It changes relative standings between people, causing unease or resentment.

The same is likely to apply with these new grades unless a convincing explanation is forthcoming. Take my own case. I move from 167 to 170, a below par uplift compared with my peers. So I wonder why and look at my grading peers. My close club colleague, Steve Connor, and I are roughly the same standard. I play above him generally, but our grades have tracked each other identically over the past few years (177-176-171) until this past year when I drift to 167 and Steve to 162. But in the new grades, I go to 170 while Steve jumps to 172. The only apparent difference is that Steve plays weaker opponents in our league, and a few more games in the 4NCL. This alters his grading 'gene pool' a little from mine, and presumably holds the key to his higher new grade.

Then I conducted a more or less random comparison by looking at all the people named 'Robertson', by happy chance finding the next on the ECF list, Stuart, has also moved down a little this year from 171 to 160. By further happy chance, he has played the same number of games as me too. Yet his new grade jumps from 160 to 171, and I'm feeling a bit rueful. Presumably the explanation in this case is that Stuart (who I've never met) is a student (I deduce), and has been playing in a more junior 'gene pool' picking up the deflation virus.

I've done other 'controls' and comparisons, and a similar pattern emerges: the new grades seem to reflect, not one's own specific performance, but the prior performance of one's opponents and the extent of their exposure to deflated juniors. In short, new grades have a strong element of luck in them, depending on how fortunate one has been in meeting opponents with juniors in their playing network.

Finally, I checked out the new grades of the juniors I played in the recent Major Open. Not having played a junior in years, like the No 72 bus, a load came along all at once. :) And the grade shifts? More eye-popping stuff! :shock:

* Rd 1 - 13 year old (164); new grade 192. Good, better than me probably or certainly will be shortly, but not yet that good in my opinion. (I lost, close)

* Rd 3 - 17 year old (131); new grade 153. About right (I won, close)

* Rd 5 - 15 year old (141); new grade 162. Much too high on the evidence before me (I won, easy)

* Rd 6 - 17 year old (172); new grade 190. Not far off. I played well; opponent played better. (I lost, close)

* Rd 8 - 9 year old (103); new grade 147. Absurd. Very promising young player, but even so... (I won, easy)

* Rd 9 - 15 year old (131); new grade 160. Much too high on the evidence (I won, easy)

* Rd 10 - 15 year old (131); new grade 164. Too high (draw, dull)

* Rd 11 - 17 year old (135); new grade 156. About right (I won, prettily)

From this I conclude, if not scientifically: there looks to be an element of inflation in some junior grades, less so at the quality end though, or with older juniors. I lost to both neo-190s, but had little trouble winning against juniors below that. From these benchmarks, and from my own 'feel' based on nearly fifty years playing chess, I'd put myself closer to 180 - not far short of my grade from this sample of eight games, in fact.

It goes without saying that I'm personally not that bothered. But I fear some others in a similar position may not be so phlegmatic as these new grades start to make an appearance. Raised eye-brows may lead to raised voices. Our hard-working graders need to have their explanations ready.

David
Atticus CC

Locked