Page 1 of 1

Elo vs ECF grading systems, take 2

Posted: Sat Aug 27, 2011 12:31 am
by Rob Thompson
hopefully we could actually have a discussion on the topic at hand this time? It would be appreciated. Also, as another thing to throw into the mix, dare i ask why Glicko isn't more accepted?

Re: Elo vs ECF grading systems, take 2

Posted: Sat Aug 27, 2011 12:55 am
by Roger de Coverly
Rob Thompson wrote:Also, as another thing to throw into the mix, dare i ask why Glicko isn't more accepted?
Glicko in my opinion is total nonsense for a domestic/national system. It makes the arrogant assumption that just because you haven't played any games subject to its rating, that that means you have become inactive. I've been told a story that an ex-pat player was able to become one of the highest rated players in his home country by virtue of returning for a month's play and exploiting good form and an extremely high Glicko driven k factor. In fact his form was not greatly different from his play elsewhere which had been rated with a somewhat saner K factor.

Re: Elo vs ECF grading systems, take 2

Posted: Sat Aug 27, 2011 12:31 pm
by Rob Thompson
Roger de Coverly wrote:
Rob Thompson wrote:Also, as another thing to throw into the mix, dare i ask why Glicko isn't more accepted?
Glicko in my opinion is total nonsense for a domestic/national system. It makes the arrogant assumption that just because you haven't played any games subject to its rating, that that means you have become inactive. I've been told a story that an ex-pat player was able to become one of the highest rated players in his home country by virtue of returning for a month's play and exploiting good form and an extremely high Glicko driven k factor. In fact his form was not greatly different from his play elsewhere which had been rated with a somewhat saner K factor.
I would assume that this person must have scored a high percentage to do this? If so, is this any different to those gaining extremely high ECF games from scoring a high percentage as an unrated?

Re: Elo vs ECF grading systems, take 2

Posted: Sat Aug 27, 2011 2:46 pm
by Roger de Coverly
Rob Thompson wrote: I would assume that this person must have scored a high percentage to do this? If so, is this any different to those gaining extremely high ECF games from scoring a high percentage as an unrated?
Yes, the player already had a local domestic rating. The problem with Glicko systems is that they assume if you haven't used your rating for a while, that it means it has become "unreliable". I think the not reliable parameter kicks in within a year. In a British context, if you hadn't played for nearly a year, you would still be graded. I think, but don't know, that you would have to be absent for 36 months before you came back as a new player.

One of its other sillier features is that once you become active again, your rating is again treated as reliable. So there are effect where a player of a 2000 standard takes a break, plays at 1500 for their first comeback event and 2000 again afterwards. Their rating takes a very big hit even if they outperform in later events.

The Aussies use Glicko, their forum is www.chesschat.org.au. Every so often you get a moan about it.

Re: Elo vs ECF grading systems, take 2

Posted: Sat Aug 27, 2011 3:47 pm
by Ian Thompson
Roger de Coverly wrote:In a British context, if you hadn't played for nearly a year, you would still be graded. I think, but don't know, that you would have to be absent for 36 months before you came back as a new player.
Unless its changed recently, that's not the way it works. I played no graded games in the 2006-07 season, only playing abroad. When I resumed graded play in 2007-08 I was treated as an ungraded player for calculating both my, and my opponents', grading performances for the current season. If I'd played less than 30 games in 2007-08 then my 2005-06 grade would have been used in my grading calculation to bring my number of games up to 30, but that's all that my previous grading record would have been used for.

Re: Elo vs ECF grading systems, take 2

Posted: Sat Aug 27, 2011 9:33 pm
by Christopher Kreuzer
Ian Thompson wrote:
Roger de Coverly wrote:In a British context, if you hadn't played for nearly a year, you would still be graded. I think, but don't know, that you would have to be absent for 36 months before you came back as a new player.
Unless its changed recently, that's not the way it works. I played no graded games in the 2006-07 season, only playing abroad. When I resumed graded play in 2007-08 I was treated as an ungraded player for calculating both my, and my opponents', grading performances for the current season. If I'd played less than 30 games in 2007-08 then my 2005-06 grade would have been used in my grading calculation to bring my number of games up to 30, but that's all that my previous grading record would have been used for.
That sounds correct to me. All the current grade categories require at least one graded game from the season at the end of which the list is being published to calculate a grade. Ian, during the year when you were not listed with a grade, were you listed with a grade under 'inactive' players?

http://grading.bcfservices.org.uk/getre ... ef=120322J

The FAQ says this about inactive players:

http://grading.bcfservices.org.uk/help.php#inactive

"By default, only players with a current Grade are shown. However a check box ("include inactive players") allows you to include players who are not currently graded but have been graded in the past."

All active players who play more than 30 games a year effectively rewrite their grades each year. Those who play less than 30 games a year get a moving average over their latest 30 games (if within 3 years) at the point the grading calculations are made. With the changes to a six-month list, more people will end up in the 'moving average over last 30 games' system than in the 'average of this (half-)season's performances' system.