Re: Hastings


"Andy Walker" <anw@xxxxxxxxx> wrote in message
news:flhf1p$haq$1$8300dec7@xxxxxxxxxxxxxxxxxxx
In article <jOudndzzY_dNtOHanZ2dnUVZ_jednZ2d@xxxxxxxxxxx>,
Chess One <OneChess@xxxxxxxxxxx> wrote:
ECF have agreed to be interviewed by Chessville - would you like to
compose
a question based on what follows, that I can present to them in your name?

Not really. ECF are "onside" on this, more-or-less; see
"http://www.bcf.org.uk/grading/2006/stat-gs_oct06.htm";. There is
some later info floating around, but I haven't seen it on the Web.

Thank you for this resource; I see that the 2 initial statements were

"David Welch came up with the hypothesis that the oddities in the conversion
from FIDE to ECF were due to anomalies in the ECF grading system.
Subsequently analysis by Dave Thomas showed that:
1) The actual performance of players did not appear to correspond to the
theoretical performance. This means that whereas for a grading difference of
25 points the stronger player should score 75%, his actual score is more
like 68%.

2) The problem appears to have been present at least since the introduction
of the submission of individual results in 2000 and probably since 1995,
which is the earliest date for which an electronic record of grades is
available to the team. It has been suggested that it stems from the
extension of the grading system to players graded below 175 (in present
terms) in the 1960s."

Which are then followed by 10 other points to which most answers are of the
'unknown' or indeterminable type.

It strikes me Andy, that no-one has suggested any benchmarks for on-going
calibration of ECF ratings, and many questions cannot therefore be answered
using available data, since successive increments in that data cannot be
pegged to other factors. Thereby all the data 'floats' and individual
factors cannot de differentiated from other factors. I wonder if there is
not some means to distinguish one thing from another by sensibly isolating
these factors from each other, then observing their respective progress over
time?

In the US the cause of rating inflation is clear in 2 respects;- here there
are 'awards' which fix ratings so that on attaining 2200, eg, that becomes a
permanent 'rating-floor' for that player, despite the fact that 3 years
later actual performance has slipped to 1950 on the same scale, the result
of games is still scored against the 2200 number.

This seems to make a monkey out of Elo's idea by deliberately removing
ratings from any mathematical basis whatever!

Secondly, popularity of tournament chess can place players within the same
approximate rating range, consistently either at the top or the bottom of
it: A friend has played in half a dozen money tournaments now where he is in
the top 10% of players in a [eg] 1500-1900 class, and he often wins money
thereby, but on the broader scene he must play higher rated players than his
nominal 1850 rating, and so loses the points he gained against the lower
rated ones - and he is quite happy about this because then he can keep
winning money for his wins!

I do, as an aside, credit David with some sense that there are more GMs in
the pool now, and this would therefore stabilise or substantialise the
high-end of the pool against statistical-widow data.

"More GMs" does nothing to stabilise the pool if it is merely
a reflexion of ratings inflation. FIDE [and ECF, and USCF etc if
they are seeing the same phenomena] need to identify the underlying
causes of ratings drifts, and cure them. I have my own suspicions,
but it would be a major research exercise to do the statistics on
the available data. FWIW, my impression is that the "good club
player" level, at around 200 ECF, 2200 FIDE, is not *yet* very much
affected, at least in the UK, which could [if confirmed] serve to
"anchor" any re-rating exercise.

Khalifman was a good analyst of the 'SuperGM' group - writing about 9 years
ago he noted that SuperGMs do not play sufficiently with plain old everyday
GMs, thus maintaining their own very high ratings, while repressing the
chance of other GMs to gain sufficient points and join the SuperGM group.

At the time he thought this situation was the one that SuperGMs wanted,
since it also convenienced the big money Euro-Tour Invitational organisers
to promote these big 'names' in chess to credit their own tournament.

Another challenge to the current system is from Adorjan who does not think
that if the top 300 players were to fight it out in a huge tournament that
the results would be predicted by ELO.

Anyway - I see Dave has sensibly transposed this increasingly
rating-oriented thread to one specifically on that subject.

Cordially, Phil Innes

--
Andy Walker
Nottingham


.