K-Factor of 20

The very latest International round up of English news.
Tim Spanton
Posts: 1205
Joined: Thu May 01, 2008 11:35 am

K-Factor of 20

Post by Tim Spanton » Fri Aug 29, 2014 4:46 pm

Fide has just introduced this, seemingly to replace 15. I clearly missed the announcement, and cannot find it. Anyone know the reasoning?

Roger de Coverly
Posts: 21301
Joined: Tue Apr 15, 2008 2:51 pm

Re: K-Factor of 20

Post by Roger de Coverly » Fri Aug 29, 2014 5:04 pm

Tim Spanton wrote: Anyone know the reasoning?
It's an attempt to deal with the ever increasing number of players whose rating bears no resemblance to their current playing strength. For under 18s it's K=40 if their rating is under 2300.

http://www.chessblog.com/2013/07/fide-o ... rt-of.html

Tim Spanton
Posts: 1205
Joined: Thu May 01, 2008 11:35 am

Re: K-Factor of 20

Post by Tim Spanton » Fri Aug 29, 2014 6:01 pm

It will add to the excitement at Paignton, starting this weekend!

NickFaulks
Posts: 8453
Joined: Sat Jan 02, 2010 1:28 pm

Re: K-Factor of 20

Post by NickFaulks » Fri Aug 29, 2014 6:10 pm

Roger de Coverly wrote: It's an attempt to deal with the ever increasing number of players whose rating bears no resemblance to their current playing strength.
Needless to say, there is no evidence to support that statement.
If you want a picture of the future, imagine a QR code stamped on a human face — forever.

Roger de Coverly
Posts: 21301
Joined: Tue Apr 15, 2008 2:51 pm

Re: K-Factor of 20

Post by Roger de Coverly » Fri Aug 29, 2014 6:38 pm

NickFaulks wrote: Needless to say, there is no evidence to support that statement.
There are certainly players whose International rating bears no relationship to their strength. I played one at the weekend. Their number increased with the extension downwards of the International ratings. Evidence is where International ratings are out of line with domestic ratings or grades.

NickFaulks
Posts: 8453
Joined: Sat Jan 02, 2010 1:28 pm

Re: K-Factor of 20

Post by NickFaulks » Fri Aug 29, 2014 6:57 pm

Roger de Coverly wrote:
NickFaulks wrote: Needless to say, there is no evidence to support that statement.
There are certainly players whose International rating bears no relationship to their strength. I played one at the weekend. Their number increased with the extension downwards of the International ratings. Evidence is where International ratings are out of line with domestic ratings or grades.
Oh, you're talking about federations where very few games are submitted for rating. There's not much that any system can do with games it doesn't know about.
If you want a picture of the future, imagine a QR code stamped on a human face — forever.

Roger de Coverly
Posts: 21301
Joined: Tue Apr 15, 2008 2:51 pm

Re: K-Factor of 20

Post by Roger de Coverly » Fri Aug 29, 2014 7:04 pm

NickFaulks wrote:[
Oh, you're talking about federations where very few games are submitted for rating. There's not much that any system can do with games it doesn't know about.
The ECF system for example can almost totally forget about a previous grade. I thought the intent of K=40 for juniors was to emulate that behaviour so as to get games at an earlier lower standard forgotten about more quickly. It's a feature of Elo based systems that equal performances don't give equal ratings where the players have previous ratings well out of line with their current strength. Only where there is no previous rating are equal performances equally rewarded.

Brian Valentine
Posts: 574
Joined: Fri Apr 03, 2009 1:30 pm

Re: K-Factor of 20

Post by Brian Valentine » Fri Aug 29, 2014 7:16 pm

There is clear evidence that either the FIDE rating for juniors or the ECF grading for the same players is way out of line. The ECF method is one I'd prefer of the two (declaring an interest as the manager of the system). I'd suggest one advantage of the ECF system is that it grades games that FIDE will not touch (e.g. for being too quick or for at least one player not having a rating, or even that they has not played their subs.

MartinCarpenter
Posts: 3044
Joined: Tue May 24, 2011 10:58 am

Re: K-Factor of 20

Post by MartinCarpenter » Fri Aug 29, 2014 7:56 pm

Given how they've kept over performing in recent British champs, the ECF ones for the juniors look to be rather more accurate :) Some of them were playing a reasonable amount of FIDE chess a year too, it just took a long time to correct when the initial rating was notably out of line.

There are also, I presume, a bunch of people with basically pseudo random FIDE grades arising from playing a few games a year in 4NCL Div3 and the like. Can't see how you could do anything to improve those.
(It would be awfully dull to do unless automated but you could presumably check which set of grades predicted 4NCL results better.).

Lewis Martin
Posts: 296
Joined: Wed May 28, 2014 11:45 am

Re: K-Factor of 20

Post by Lewis Martin » Fri Aug 29, 2014 8:40 pm

You also may need to consider the fact that some players simply do better in longer time control than shorter ones or vice versa. A minority admittedly, but it does happen.

NickFaulks
Posts: 8453
Joined: Sat Jan 02, 2010 1:28 pm

Re: K-Factor of 20

Post by NickFaulks » Sat Aug 30, 2014 1:05 am

Brian Valentine wrote: I'd suggest one advantage of the ECF system is that it grades games that FIDE will not touch (e.g. for being too quick or for at least one player not having a rating, or even that they has not played their subs.
What? You charge your own people for the privilege of having a FIDE rating, and if they refuse you say "FIDE won't touch them".

As for one player not having a rating, they only need five rated games to get one now, so that shouldn't last long.

Finally, in events for players rated under 2200, which covers most juniors ( in England, at least ) games need to be only three hours long. Many federations are using this rule to populate the middle level of their rating list. The point is that they are looking for ways to get their players into the list, whereas in England the desire seems to be to keep them out of it.
If you want a picture of the future, imagine a QR code stamped on a human face — forever.

Brian Valentine
Posts: 574
Joined: Fri Apr 03, 2009 1:30 pm

Re: K-Factor of 20

Post by Brian Valentine » Sat Aug 30, 2014 7:33 am

Nick,
I think that you have to make more of a case on why the English should bother with FIDE ratings. As much as my post was blind to the FIDE attempts to be more inclusive, your original response to Roger ignores some real problems with how FIDE has implemented Elo.

Roger de Coverly
Posts: 21301
Joined: Tue Apr 15, 2008 2:51 pm

Re: K-Factor of 20

Post by Roger de Coverly » Sat Aug 30, 2014 8:35 am

NickFaulks wrote: The point is that they are looking for ways to get their players into the list, whereas in England the desire seems to be to keep them out of it.
I rather thought it was FIDE devising regulations which have the effect of deterring organisers of existing events from having them rated.

It's well known that young players can go from weak club player to strong amateur within a period of a couple of years. Most national Elo systems have had to adopt various ad hoc hacks to deal with this. Scotland have a rule along the lines that if the current season performance is 200 points better than the previous season, the previous rating is ignored. The k=40 hack might work, but the effect of extending the ratings to 1000 without having such a hack can be seen in the comparative rankings of players using the International list as opposed to the domestic grades.

MartinCarpenter
Posts: 3044
Joined: Tue May 24, 2011 10:58 am

Re: K-Factor of 20

Post by MartinCarpenter » Sat Aug 30, 2014 9:06 am

The UK juniors with really lagging FIDE grades tended to be starting rather under 2200 and shooting up to that level/above it. Playing the British/a full 4NCL season is 20-25 rated games a year but it takes quite a while to catch up while they're still improving fast.

Starting people after 5 games actually sounds like a fairly bad idea to me as those grades will, as a population, inevitably be really very unreliable indeed. That will obviously then randomise everyone else's grades.

Roger de Coverly
Posts: 21301
Joined: Tue Apr 15, 2008 2:51 pm

Re: K-Factor of 20

Post by Roger de Coverly » Sat Aug 30, 2014 9:21 am

MartinCarpenter wrote: Starting people after 5 games actually sounds like a fairly bad idea to me as those grades will, as a population, inevitably be really very unreliable indeed. That will obviously then randomise everyone else's grades.
I believe the expectation was that by increasing the number of rated games, it improved the reliability for everyone. That does make a certain amount of sense. You might see it in the 4NCL, that if say a player has accumulated enough games for a rating over three or four weekends, then their opponents in the final weekend will be playing against a rated player rather than an unrated one.

Stewart Reuben once suggested that if you wanted random Swiss pairings, that what you should do was to randomise the seeding order, but then use seeded pairings in the normal manner. Where you have weekend Swiss tournaments using FIDE rankings, at least in part you can get that effect.

With the apparent demise of e2e4 and relatively few local Congresses wanting the additional headaches of being FIDE rated, there are likely to be fewer FIDE rated games played in 2014-15 than in 2013-14.

(edit) Without really checking, I get the impression Adam Raoof is running more rated events than he used to. I think he may be the only organiser to exclude players over 2200 in the interests of having the events FIDE rated. In the context of being London based, they have plenty of alternatives. (/edit)