Database Upgrade

Request amendments or pass comments on this free service.
Richard Bates
Posts: 2767
Joined: Fri Nov 14, 2008 8:27 pm

Re: Database Upgrade

Post by Richard Bates » Sun Mar 08, 2009 6:49 pm

Howard Grist wrote:
Roger de Coverly wrote:Reality check - do you really believe in under 10 players of a 150 standard?
Good question - on the old basis, no - but on the new basis 150 is the old 130, and juniors have been undergraded for some considerable time. So it's really a question of changing your ides on these things. Unfortunately, I still don't have quite enough data to do some of the checks that I'd like to do - such as whether the 10 year-olds this year are significantly better than those of last year as not all the grades are calculated using three years worth of data. I don't think that they are as far out as you seem to think though.
[/quote]

This is slightly missing the point. The phrase "150 is the new 130" is in relation to the perceived deflation in recent years. You can go back any number of years you like (long before "deflation" is claimed to be apparent) and you won't find a pool of U10s at this sort of (relative) strength. It would be pretty rare to find many as high as 130. So the result is that the new grades are effectively saying that this is probably the strongest group of young children the UK has ever had. I am in no position to judge that, but i know of people who have been working in junior chess for many, many years who would say that is just wrong.. And additionally strange when the U18 age group downwards is nothing remarkable in historic grading terms at all.

Roger de Coverly
Posts: 17389
Joined: Tue Apr 15, 2008 2:51 pm

Re: Database Upgrade

Post by Roger de Coverly » Sun Mar 08, 2009 7:27 pm

and juniors have been undergraded for some considerable time.
Would you like to comment on the Rough case? - see above.

Could you please demonstrate how you add 50 points to a player by incorrectly flagging them as a junior without there being a serious error somewhere in the system?

Neill Cooper
Posts: 1212
Joined: Sun Feb 24, 2008 4:43 pm
Location: Croydon
Contact:

Re: Database Upgrade

Post by Neill Cooper » Sun Mar 08, 2009 7:59 pm

I am very concerned by what I see as junior grades being significantly too high.
This means that in mixed junior/adult teams (club, county) juniors will be forced to play in higher teams, and on higher boards than they should (opponents will object to board orders with juniors being moved down below players graded 20 points less than them). They will therefore either quickly improve or, more likely, get depressed after 5 successive losses to players graded equal to themselves and shy away from league and county chess.

I think that there are 3 possible explanations:
Possible explanation 1) I am wrong about the exageration. But why is it that the all the 31 players who have gained more than 75 grading points are aged 9 to 12 and have grade code B (1 player) to E? A couple played 1 only game last season, and their avegare is 10 games.
Possible explanation 2) The computer code to do the calculation is faulty. It should, of course, have been through a full QA checkwith appropriate functional specification, verification and validation procedures. Have these been documented? In view of the significant impact on junior chess of the new grades this should be shown to have been done.
Possible explanation 3) The mathematical basis of the change is unsound. Whilst it is clear that it works for the central core who regularly play 40+ matches per season, there appear to be subnetworks of chess players only losely connect to the core. Would this explain the Wallasey anomalies of negative grade changes. Also many junior players will have played most or all long play gamnes against other juniors and so would depend on the few players in their network who have played games against the core.

I accept that this season's junior grades are underestimates (which is why my club, Castles, is top of the Surrey U90 division with a team of juniors and ungraded parents). But it would be wrong to overgrade juniors, and we should err on the side of underestimates.

Leonard Barden
Posts: 1325
Joined: Wed Dec 24, 2008 11:21 am

Re: Database Upgrade

Post by Leonard Barden » Sun Mar 08, 2009 8:32 pm

Long ago when I was junior squad manager and responsible for a golden generation of future GMs and IMs, I quickly discarded official BCF annual ratings as useless for fast improvers. For junior invitation tournaments I started off U10s as 80-100 depending on what I knew about them and under-12s as 110-130. But I also kept track of as many results as possible, updated my ratings every week, and gave juniors their latest rating every time they were contacted.

For the top juniors I took this very seriously, and Professor Elo's seminal work The Rating of Chessplayers includes my graph of Julian Hodgson's progress over a period of several years with calculations every week or two. Elo liked it because there were dips in form and the graph illustrated the phenomenon of quantum jumps. I also offered him a graph of Short but he refused that because Nigel's progress was so consistent with no setbacks.

I think that's what is really needed, far more frequent updates for the best juniors. Instead of that we are seeing here a controversial psychologically, statistically and empirically dubious megabonus, for a load of weakies. I felt similarly when I read on some other thread that there was going to be a gold level for players who reached a grade of 70. It should be more like 170.

Realistically, our juniors now are as weak by international standards as they have ever been in my chess lifetime. Two potential IMs-Yang-Fan Zhou and Felix Ynojosa. A few steady improvers like Kilpatrick, Auckland, O'Toole and Franklin who may make 2300. At the youngest end where the rating points are about to be scattered like confetti, I currently see only Andreev and Wadsworth who are developing promisingly. This pair indeed do seem underrated, and the answer is a specially frequent individual calculation for these and any others who start to stand out.

Leonard Barden
former BCF international grader

John Upham
Posts: 4157
Joined: Wed Apr 04, 2007 10:29 am
Location: Cove, Hampshire, England.
Contact:

Re: Database Upgrade

Post by John Upham » Sun Mar 08, 2009 11:29 pm

Leonard,

I would like to add another AMCA squad member to your one-to-watch list (apart from Peter Andreev) viz : Tarun Malhotra.

We have just recruited Abigail Pritchard into the AMCA senior squad.

Is Peter A. Williams (labelled as a "Child Genius" by television) on your current radar?

Nick Thomas
Posts: 450
Joined: Sat Oct 11, 2008 9:56 pm

Re: Database Upgrade

Post by Nick Thomas » Mon Mar 09, 2009 8:18 am

Leonard Barden said:
Realistically, our juniors now are as weak by international standards as they have ever been in my chess lifetime. Two potential IMs-Yang-Fan Zhou and Felix Ynojosa. A few steady improvers like Kilpatrick, Auckland, O'Toole and Franklin who may make 2300. At the youngest end where the rating points are about to be scattered like confetti, I currently see only Andreev and Wadsworth who are developing promisingly. This pair indeed do seem underrated, and the answer is a specially frequent individual calculation for these and any others who start to stand out.
True but probably not the whole truth. There are others who haven't made the list who are certainly on course for 2300 in the next 2 or 3 years and potential Im's GM's e.g. Brandon Clarke. GM's such as Arkell, Hebden, Emms were not much more than 200 - 215 by the time they were about 20yo if I remember correctly.

Leonard Barden
Posts: 1325
Joined: Wed Dec 24, 2008 11:21 am

Re: Database Upgrade

Post by Leonard Barden » Mon Mar 09, 2009 12:35 pm

Peter Williams has been around for several years and has made rather slow progress. I know little about Brandon Clarke but after checking with his Fide card and recent events I would observe that if he is really a potential IM he should be playing more frequently and in tougher competition than the likes of the Leicestershire rating or the Eurounion U14. Why didn't he play in any of the British Championship, Hastings, Liverpool, or Cappelle?

In the late 1970s and early 1980s we had few rated and titled players so there was a bottleneck which until Lloyds Bank and Benedictine were in full swing slowed down players of the generation of Hebden, Arkell, Martin and Flear. But Hebden played Ramsgate 1979, an international APA, at 19. Arkell was in the team which played Spassky in January 1979 and I rated him about 210 at 18 then. Emms? At age 19, at Oakham 1986, he drew with Michael Adams and defeated Vishy Anand(!). You can view the games on chessgames.com.
Last edited by Leonard Barden on Mon Mar 09, 2009 2:38 pm, edited 1 time in total.

Matt Harrison
Posts: 129
Joined: Fri Apr 06, 2007 4:51 pm

Re: Database Upgrade

Post by Matt Harrison » Mon Mar 09, 2009 12:39 pm

Leonard may well be right that the current strength of junior chess in England isn't great - results at world and european level would bear that out. I do think however that there are another 10 or so juniors worth watching that haven't been mentioned so far.

But back to the changes. Having looked at v5 in some detail (and with quite a bit of knowledge of the older juniors at least), it does seem strange that some of the players who have just about dropped out in the last year or two are showing the largest increases in grade.

Confining the selection to players graded over 150 from England. Gain is defined as new grade minus current grade.

Age_____Sample___Ave____Max____Cat of_______Average gain
_________________gain____gain____max gain____A&B______C,D &E
8________1_______66______66______A__________66_______n/a
9________5_______64______82______D__________57_______68
10_______5_______59______69______E__________55_______65
11_______8_______48______65______C__________42_______65
12_______15______36______57______E__________35_______57
13_______22______30______53______D__________27_______44
14_______24______25______47______E__________23_______39
15_______25______15______30______C__________14_______22
16_______23______13______34______E__________12_______23
17_______27______11______35______E__________8________16

In all but one case (where there is only one 8 year old with a new grade of 150), the largest increases have gone to the players who have been least active in the last year. This alone leads me to query the whole foundation of this approach. Now I suppose, if these players aren't going to be playing much more this year/next year then they will leave the system taking their inflated grades with them which will remove the problem. My guess (and it is a pure guess) is that when games from earlier seasons have been brought into the grading calculation they have had the junior bonus added more than once, thus distorting the resultant new grade.

As a side note, the only player in the whole sample of 155 juniors who loses grading points is David Howell.
Last edited by Matt Harrison on Mon Mar 09, 2009 12:59 pm, edited 2 times in total.

John Upham
Posts: 4157
Joined: Wed Apr 04, 2007 10:29 am
Location: Cove, Hampshire, England.
Contact:

Re: Database Upgrade

Post by John Upham » Mon Mar 09, 2009 12:46 pm

Leonard,

Do you still have influence amongst the ECF junior selectors : I assume yes?

According to the ECF site the current selectors are :

International Selection: Stewart Reuben (Chairman), Allan Beardsworth, Jana Bellin, Ray Edwards, Glenn Flear, Harriet Hunt, Jonathan Parker, Richard Palliser.

Senior Selection: Stewart Reuben (Chairman), David Anderton, Ray Edwards

is there another panel for Junior selection or is that covered by the top most group?

Matt Harrison
Posts: 129
Joined: Fri Apr 06, 2007 4:51 pm

Re: Database Upgrade

Post by Matt Harrison » Mon Mar 09, 2009 12:48 pm

John,

Junior selection policy is described on the following page of the ECF website:

http://www.bcf.org.uk/junior/international-js.htm

It uses a combination of major junior tournaments and a trial (which was apparently this last weekend). Note it contains the line:

"Grades are irrelevant to this procedure!"

Leonard Barden
Posts: 1325
Joined: Wed Dec 24, 2008 11:21 am

Re: Database Upgrade

Post by Leonard Barden » Mon Mar 09, 2009 12:53 pm

I have had no influence on junior selection for the best part of a decade.

Nick Thomas
Posts: 450
Joined: Sat Oct 11, 2008 9:56 pm

Re: Database Upgrade

Post by Nick Thomas » Mon Mar 09, 2009 3:25 pm

Leonard Barden wrote:Peter Williams has been around for several years and has made rather slow progress. I know little about Brandon Clarke but after checking with his Fide card and recent events I would observe that if he is really a potential IM he should be playing more frequently and in tougher competition than the likes of the Leicestershire rating or the Eurounion U14. Why didn't he play in any of the British Championship, Hastings, Liverpool, or Cappelle?

In the late 1970s and early 1980s we had few rated and titled players so there was a bottleneck which until Lloyds Bank and Benedictine were in full swing slowed down players of the generation of Hebden, Arkell, Martin and Flear. But Hebden played Ramsgate 1979, an international APA, at 19. Arkell was in the team which played Spassky in January 1979 and I rated him about 210 at 18 then. Emms? At age 19, at Oakham 1986, he drew with Michael Adams and defeated Vishy Anand(!). You can view the games on chessgames.com.
Brandon played 126 (!!) graded games last season (and 54 rapid games). His rapid grade is 187 and he is only 12. He played at the British but in the junior sections. I'm certain that he will play higher quality tournaments from now on but he is still young and has only just made the transition into playing more "serious" forms of chess (rated games, 4NCL etc). You mention players who are far older but hardly any better.....

Leonard Barden
Posts: 1325
Joined: Wed Dec 24, 2008 11:21 am

Re: Database Upgrade

Post by Leonard Barden » Mon Mar 09, 2009 5:00 pm

Okay, I'll watch Brandon's future progress with interest.

Matthew Turner
Posts: 2680
Joined: Fri May 16, 2008 11:54 am

Re: Database Upgrade

Post by Matthew Turner » Mon Mar 09, 2009 5:35 pm

To monitor the progress of top juniors you certainly require information that is more up-to-date than the ECF grading list. However, the problem with 'Barden grades' as they were known, was that they were subjective. Leonard chose the juniors that he thought were good and then updated their grades regularly with information that he was supplied with (through the grand prix and other sources). This tended to favour juniors from the South East. It led to situations where juniors who were one of Leonard's selection were judged on their up-to-date performances and those who had not caught his eye (like Brandon Clarke) were judged on their ECF grade, which could be 18 months out of date.
It seems to me that in this electronic age we should be publishing grades more often than once a year. If we did so, I suspect many of the problems associated with the grading system would soon dissipate.

Sean Hewitt

Re: Database Upgrade

Post by Sean Hewitt » Mon Mar 09, 2009 5:57 pm

Matthew Turner wrote: It seems to me that in this electronic age we should be publishing grades more often than once a year. If we did so, I suspect many of the problems associated with the grading system would soon dissipate.
Absolutely right - half yearly minimum although I would favour quarterly.

Unfortunately, the ECF actually wanted to do this a couple of years ago but Chris Majers proposal to that effect was voted out by the ECF council!

Locked