Page 1 of 2

Second Revision 05.09.2009

Posted: Sat Sep 05, 2009 4:55 pm
by Carl Hibbard
An early revision has been necessary following a programming bug which caused all Direct Members' foreign results, for both of the last two seasons, to be omitted from the calculations for the 30th-August list. This has been corrected. Changes to the Direct Members' grades, in some cases substantial, have led to changes in a number of other grades. Total changes to 2009 Standardplay grades: 302, of which 275 are by 1 point. Rapidplay: 32, of which 29 are by 1 point.

See the grading site:

http://grading.bcfservices.org.uk/

Re: Second Revision 05.09.2009

Posted: Sat Sep 05, 2009 5:05 pm
by Alex Holowczak
I'm beginning to think I should check my grade weekly. Who knows what it'll be the next time I look at it?! :lol:

Re: Second Revision 05.09.2009

Posted: Sat Sep 05, 2009 6:31 pm
by Roger de Coverly
Carl Hibbard wrote:An early revision has been necessary following a programming bug which caused all Direct Members' foreign results, for both of the last two seasons, to be omitted from the calculations for the 30th-August list.
I've run a comparison between v2 and v3

Observations

The ECF Marketing Director has gained 12 points (8 points on GRADELAST) without his game count (GRADEGAMES) changing :!:
IMs and GMs don't generally bother to submit their foreign results. The only IM in the changes list has lost a shedload of ECF points.

Re: Second Revision 05.09.2009

Posted: Sat Sep 05, 2009 6:50 pm
by Carl Hibbard
Roger de Coverly wrote:
Carl Hibbard wrote:An early revision has been necessary following a programming bug which caused all Direct Members' foreign results, for both of the last two seasons, to be omitted from the calculations for the 30th-August list.
I've run a comparison between v2 and v3

Observations

The ECF Marketing Director has gained 12 points (8 points on GRADELAST) without his game count (GRADEGAMES) changing :!:
IMs and GMs don't generally bother to submit their foreign results. The only IM in the changes list has lost a shedload of ECF points.
Could be last year, Richard can you clarify this one?

Re: Second Revision 05.09.2009

Posted: Sat Sep 05, 2009 7:01 pm
by Roger de Coverly
Just a quick follow up.

I've looked up the IM and the ECF director in the v1 download

The IM has gained dozens of extra games - from the history on the FIDE site this is plausible although his results look consistent with his international rating.

The ECF director has just reverted back to his v1 record.

Re: Second Revision 05.09.2009

Posted: Sat Sep 05, 2009 9:37 pm
by Richard Haddrell
Carl Hibbard wrote:
Roger de Coverly wrote:The ECF Marketing Director has gained 12 points (8 points on GRADELAST) without his game count (GRADEGAMES) changing :!:
IMs and GMs don't generally bother to submit their foreign results. The only IM in the changes list has lost a shedload of ECF points.
Could be last year, Richard can you clarify this one?
I'll try. The ECF Director has simply regained 9 foreign games from 2007-8 which had been incorrectly removed. His this-year game count hasn't changed, but his last-year one has. As for the IM, he sent 53 games from foreign events 2008-9, compared with only 19 “domestic” games which went in automatically. It’s not surprising if the 53 made a difference to his grade when they were added late.

Richard

Re: Second Revision 05.09.2009

Posted: Sat Sep 05, 2009 10:13 pm
by Roger de Coverly
Richard Haddrell wrote:His this-year game count hasn't changed, but his last-year one has.
Ok, the foreign games are a red herring. What we've got in v1 and v3 is 22 games with a start grade of 135 and an end grade of 143. In v2 we have the same 22 games but with a start grade of 127 and an end grade of 131. It still looks a bit odd but then I don't know whether the 22 games consist mostly of +40 and -40 games or what the "missing" 8 have been averaged in at.

What you are seeing is an 8 point difference in the start grade being magnified to a 12 point difference in the end grade whereas you would expect the current year performance to swamp the prior year.

Re: Second Revision 05.09.2009

Posted: Sun Sep 06, 2009 8:14 am
by Matthew Turner
Can I just clarify something about the submission of foreign events which applies more widely to playing foreign players. Let's say that I play no games in England, but submit 9 games from a tournament in Imagistan (the Imagistanis never come to England). Here are my results

1 draw vs 2200
2 draw vs 1400
3 draw vs 1800
4 draw vs 1800
5 draw vs 1800
6 draw vs 1800
7 draw vs 1800
8 draw vs 1800
9 win vs 1200

A cursory look would suggest that I am about 1800, but is it true that I would come out with an ECF grade of infinty?
The draws are irrelevant because their performance will be whatever my grade is and hence I will get my grade for drawing with them. If I start at 150, then my round 9 opponent's performance is 100, but I must get at leat 160 for winning, so his performance is 110. This process just seems to continue until I and the first 8 Imagistanis reach infinity and my poor round nine opponent has to settle for an ECF grade of infinity - 50.

Re: Second Revision 05.09.2009

Posted: Sun Sep 06, 2009 8:41 am
by Roger de Coverly
Matthew Turner wrote:A cursory look would suggest that I am about 1800,
I suspect they would treat the Imagistan grades as valid, so the calculation would be a two stage process at worst ( without and with 40 point rule). It still leaves the open question as to the appropriate conversion factor from Imagistan national Elo (or Fide Elo for that matter). The only grade ever calculated for the Imagistan players is the converted one.

As I understood it, the sequence of events is (or was)

1) new player estimation
2) calculate grades for domestic games
3) adjust grades for overseas games

Re: Second Revision 05.09.2009

Posted: Sun Sep 06, 2009 9:07 am
by Matthew Turner
Roger,
As far as am aware the grading team have consistently said that they would not use conversions from national ELO or FIDE, but if they are, then perhaps I should mention that most of the Imagistanis are juniors :D

Re: Second Revision 05.09.2009

Posted: Sun Sep 06, 2009 9:13 am
by Matthew Turner
This came from Sean Hewitt

"I see! Yes, it that respect you're right. In fact that's the same for any FIDE rated player. Their actual rating is totally irrelevant."

So he is under the impression that FIDE rating are never used in the calculation of ECF grades.

Re: Second Revision 05.09.2009

Posted: Sun Sep 06, 2009 9:32 am
by Roger de Coverly
Matthew Turner wrote:As far as am aware the grading team have consistently said that they would not use conversions from national ELO or FIDE
They should add "except in the calculation of direct member's foreign results". I think Howard Grist confirmed that point a while back. http://www.ecforum.org.uk/viewtopic.php ... 2891#p2891
Howard Grist June 2008 wrote:Sean, it's not a myth. The FIDE conversion formula is used when ECF Members play in foreign tournaments and want to have their results included in their grade. It could be a source of error, but the number of games involved is very small compared to the number of games that are graded so is unlikely to throw the whole thing out of balance.
Think of the practicalities of using the estimation process. This would require submission and processing of the complete crosstable of the foreign event. They don't ask for this - just the personal list of results. I should know, my almost career best 188 (at the age of 58) was downgraded to a mere 186 because of foreign results.
Matthew Turner wrote:then perhaps I should mention that most of the Imagistanis are juniors


The ECF submission form asks for international Elo when it exists and national Elo when it doesn't. It does not ask whether any of the players are juniors.

Re: Second Revision 05.09.2009

Posted: Sun Sep 06, 2009 9:52 am
by Matthew Turner
Roger,
Many thanks, I've clearly misunderstood.
Matt

Re: Second Revision 05.09.2009

Posted: Sun Sep 06, 2009 2:34 pm
by Sean Hewitt
Matthew - My previous comment was in relation to a FIDE or otherwise nationally graded player playing games in England and therefore appearing in the ECF grading list themselves. In such cases their rating / grade is ignored and they are treated as a new ungraded player.

However, if an ECF member plays games abroad (in Imagistan for example) the Imagistanis never make it into the ECF grading list. The ECF work out the grading perfomance of the member based on first the opponents FIDE rating, then their national grade by using published conversion formula, and add that to your English games to calculate your new grade. I don't know what they do if the opponent has no such grade / rating. Ignore the game presumably.

Re: Second Revision 05.09.2009

Posted: Sun Sep 06, 2009 2:43 pm
by Neill Cooper
Carl Hibbard wrote:Total changes to 2009 Standardplay grades: 302, of which 275 are by 1 point. Rapidplay: 32, of which 29 are by 1 point.
Previously such changes post the 'published' list were highlighted, I think in a blue background. Could that be done this time? It makes it much easier for me to know which players I need to change in my spreadsheets.