View previous topic :: View next topic |
Author |
Message |
Mike Scott King
Joined: 01 Feb 2007 Posts: 676 Location: Edinburgh
|
Posted: Mon Aug 08, 2011 1:51 pm Post subject: ECF Grade Conversion |
|
|
Has there been any further look at the CS conversion formula of ECF grades (10*ECF+180) to judge who accurate it is compared to the old one still used by the ECF? |
|
Back to top |
|
|
DMB King
Joined: 08 Mar 2007 Posts: 267
|
Posted: Wed Aug 10, 2011 9:14 am Post subject: |
|
|
If two grading systems had exactly equal values then you would anticipate that the actual points scored should equal the expected points scored if you had the correct conversion factor between the two systems. We had thought "old" ECF grades were fairly comparable with CS.
To reflect the changes made to ECF grades in 2009 we have used the new conversion factor of *10+180 over the last two seasons.
You can see how many points were actually scored and expected by looking up the crosstables http://www.chessscotland.com/event2012/events2012.htm
Select "Outside Scotland Events", "UK Events" and then you get a list of all the mainly English events processed. The columns show the actual and expected points for each player based on the new conversion factor. (However the opponent grades are not pure ECF. If the opponent was known to have a CS or FIDE that was used first).
In season 2009-2010 there were 287 games processed with 136 actual points and 120.954 expected.
In season 2010-2011 there were 198 games processed with 105.5 actual points and 85.976 expected.
So despite the new conversion factor making ECF grades significantly less valuable Scots are still scoring more than expected.
**************
With regard to FIDE grades Scots are also outscoring what is expected. Over the last season there 927 FIDE results processed for members. The actual points scored was 412 and expected 373.813.
However this outperformance varies with grade. If you just look at the 284 games played by Scots with a CS of 2200+ then actual v FIDE was 156.5 compared to expected 149.648. ie very close.
If you look at the 367 games by sub 2000 CS players then the actual v FIDE was 125.5 compared to 96.167 expected.
Conclusion CS and FIDE grades are closer in value at the higher levels.
***************
20 years ago we regularly compared values CS and FIDE and found no significant difference between the grades. However that was at a time when FIDE was restricted to a minimum rating of 2200. Over the last 20 years FIDE have been gradually reducing the level, next year it stops at 1001. The vast majority of Scots who have a FIDE will find that it is higher than their domestic CS rating (unless they are over 2200 CS when the grades are still broadly similar).
FIDE grades may not really have had enough time to settle yet. It may not actually be that Scots are too low in relation to FIDE once the FIDE system has had more time to exchange points throughout the whole system. |
|
Back to top |
|
|
|