December 2014 FIDE rating list

[LawrenceCooper]

England (Active players)

# Name Title Fed Rating G B-Year
1 Adams, Michael g ENG 2745 0 1971
2 McShane, Luke J g ENG 2672 2 1984
3 Howell, David W L g ENG 2670 2 1990
4 Jones, Gawain C B g ENG 2661 0 1987
4 Short, Nigel D g ENG 2661 0 1965
6 Sadler, Matthew D g ENG 2658 2 1974
7 Nunn, John D M g ENG 2601 11 1955
8 Pert, Nicholas g ENG 2562 2 1981
9 Hawkins, Jonathan m ENG 2552 2 1983
10 Haslinger, Stewart G g ENG 2531 0 1981
11 Hebden, Mark L g ENG 2523 22 1958
12 Parker, Jonathan F g ENG 2519 0 1976
13 Speelman, Jon S g ENG 2501 2 1956
14 Gordon, Stephen J g ENG 2500 2 1986
15 Gormally, Daniel W g ENG 2499 2 1976
16 Conquest, Stuart C g ENG 2491 2 1967
17 Arkell, Keith C g ENG 2489 22 1961
18 Zhou, Yang-Fan m ENG 2477 2 1994
19 Plaskett, H James g ENG 2475 7 1960
20 Ghasi, Ameet K m ENG 2471 1 1987
21 Trent, Lawrence m ENG 2470 0 1986
22 Wells, Peter K g ENG 2468 2 1965
23 Emms, John M g ENG 2462 0 1967
24 Flear, Glenn C g ENG 2460 2 1959
25 Palliser, Richard J D m ENG 2444 2 1981
26 Williams, Simon K g ENG 2443 2 1979
26 Pert, Richard G m ENG 2443 0 1981
28 Hunt, Harriet V m ENG 2438 0 1978
29 Hunt, Adam C m ENG 2437 2 1980
30 McDonald, Neil g ENG 2431 0 1967
31 D`Costa, Lorin A R m ENG 2430 2 1984
32 Adair, James R m ENG 2428 2 1992
33 Ward, Chris G g ENG 2423 0 1968
34 Eggleston, David J m ENG 2416 11 1987
34 Summerscale, Aaron P g ENG 2416 0 1969
36 Ferguson, Mark m ENG 2415 0 1977
36 Levitt, Jonathan P g ENG 2415 0 1963
38 Roberson, Peter T m ENG 2412 2 1989
39 Mah, Karl C C m ENG 2411 0 1980
40 Ashton, Adam G f ENG 2398 2 1980
41 Ledger, Andrew J m ENG 2396 2 1969
42 Lawton, Geoff W m ENG 2389 2 1960
43 Houska, Jovanka m ENG 2388 5 1980
44 Collinson, Adam R f ENG 2383 0 1972
45 Ansell, Simon T m ENG 2379 2 1975
46 Rendle, Thomas E m ENG 2378 2 1986
47 Martin, Andrew D m ENG 2375 2 1957
48 Quillan, Gary m ENG 2367 0 1970
49 Sowray, Peter J f ENG 2365 2 1959
50 Cox, John J m ENG 2361 2 1962
51 Jackson, James P f ENG 2356 2 1991
52 Kilpatrick, Callum f ENG 2350 0 1992
53 Bates, Richard A m ENG 2349 2 1979
54 Merry, Alan B f ENG 2347 2 1996
55 Buckley, Simon T ENG 2345 0 1985
56 Pein, Malcolm m ENG 2344 2 1960
57 Ciuksyte, Dagne m ENG 2340 2 1977
58 Longson, Alexander f ENG 2339 2 1982
59 Richardson, John R f ENG 2338 2 1964
59 Bellin, Robert m ENG 2338 0 1952
61 Duncan, Chris R f ENG 2336 2 1970
62 Basman, Michael J m ENG 2330 0 1946
63 Dorrington, Chris J ENG 2329 0 1987
64 Taylor, Martin R f ENG 2328 0 1981
65 Buckley, Graeme N m ENG 2327 2 1971
65 Cooper, Lawrence H m ENG 2327 1 1970
65 Berry, Stephen H f ENG 2327 0 1951
68 Crouch, Colin S m ENG 2325 0 1956
69 Povah, Nigel E m ENG 2324 2 1952
69 Stebbings, Anthony J f ENG 2324 2 1951
71 Sanders, Isaac B f ENG 2323 11 1998
72 Harvey, Marcus R f ENG 2322 2 1996
73 Bisby, Daniel L f ENG 2318 2 1978
74 Knott, Simon J B m ENG 2316 2 1958
75 Rose, Matthew ENG 2309 1 1962
76 Kirk, Ezra f ENG 2308 9 1996
77 Fernandez, Michael f ENG 2301 0 1999
77 Franklin, Samuel G A f ENG 2301 0 1994
79 Bigg, Andrew J f ENG 2298 0 1982
80 Walker, David J f ENG 2295 1 1964
81 Beaumont, Chris R m ENG 2292 0 1952
81 Cooksey, Paul S f ENG 2292 0 1971
83 Batchelor, Peter J ENG 2290 2 1996
84 Dickenson, Neil F f ENG 2289 1 1964
85 Kenworthy, Gary f ENG 2285 1 1957
86 Eckersley-Waites, Tom ENG 2282 2 1987
86 Haydon, David L f ENG 2282 0 1988
88 Rogers, Jonathan W f ENG 2277 2 1972
88 Alfred, Nathan S W f ENG 2277 0 1980
88 Carr, Neil L f ENG 2277 0 1968
91 Constantinou, Peter f ENG 2276 9 1988
92 Webb, Laurence E f ENG 2272 1 1970
93 Buckley, David E ENG 2269 2 1979
94 Wheeler, Darren P ENG 2268 2 1971
94 Thomas, Nicholas f ENG 2268 1 1966
94 Eckersley-Waites, Adam ENG 2268 0 1987
97 Mackle, Dominic ENG 2267 0 1962
98 Lewis, Andrew P f ENG 2266 2 1960
99 Savage, Ben D M f ENG 2265 0 1977
100 Lund, D Brett ENG 2264 2 1962

England (Active players) Only Women

# Name Title Fed Rating G B-Year
1 Hunt, Harriet V m ENG 2438 0 1978
2 Houska, Jovanka m ENG 2388 5 1980
3 Ciuksyte, Dagne m ENG 2340 2 1977
4 Chevannes, Sabrina L wm ENG 2173 2 1986
5 Sarakauskiene, Zivile wm ENG 2144 2 1978
6 Jackson, Sheila wg ENG 2140 2 1957
7 Atkins, Rita wm ENG 2130 2 1969
8 Maroroa, Sue wm ENG 2096 1 1991
9 Bellin, Jana M wg ENG 2090 2 1947
10 Kalaiyalahan, Akshaya ENG 2080 5 2001
11 Lauterbach, Ingrid wm ENG 2078 12 1960
11 Hegarty, Sarah N wf ENG 2078 2 1988
13 Kueh, Audrey ENG 2071 0 1987
14 James, Ann-Marie wf ENG 2068 2 1974
15 Regan, Natasha K wm ENG 2051 0 1971
16 Hoare, Amy B ENG 2041 0 1996
17 Bhatia, Kanwal K wf ENG 2028 2 1981
17 Grigoryan, Meri wf ENG 2028 0 1977
17 Pass, Deborah J wf ENG 2028 0 1967
20 Head, Louise ENG 1940 2 1997
21 Summerscale, Claire E ENG 1912 2 1975
22 Martin, Katie A ENG 1889 0 1989
23 Pakhomova, Tuyaara ENG 1863 0 1986
24 Zhu, Yao Yao ENG 1848 0 1984
25 Giles, Yasmin ENG 1838 0 1999
26 Moore, Gillian A ENG 1831 0 1944
27 Norman, Dinah M wc ENG 1820 5 1946
28 Foster, Chantelle L ENG 1807 0 1996
28 Shepherd, Katherine M ENG 1807 0 1997
30 Keen, Cosima ENG 1805 2 1999
31 Kourtseva, Julie ENG 1774 0 1987
32 Wei, Naomi ENG 1762 0 1999
33 Needham, Sabrina ENG 1754 0 1972
34 Varney, Zoe ENG 1747 1 1999
35 Sainbayar, Anuurai ENG 1676 0 1989
36 Milson, Amy F ENG 1674 0 1995
37 Graham, Cassie M ENG 1643 1 2001
38 Sit, Victoria ENG 1618 3 1998
39 Wen, Jessica ENG 1600 0 1998
40 Green, Emily ENG 1590 1 2001
41 Imykshenova, Eugenia ENG 1589 2 1978
42 Sheremetyeva, Elizaveta ENG 1585 1 2001
43 Robson, Caroline J ENG 1573 0 1963
44 Gardner, Joan ENG 1566 0 1942
45 Neil, Jennifer ENG 1557 2 1997
46 Ivanov, Elizabeth ENG 1554 2 2001
47 Jina, Asha ENG 1545 1 2001
48 Ressel, Eva ENG 1527 2 2000
49 Dwarakanathan, Srinidhi ENG 1512 0 2005
49 Gelain-Sohn, Emmanuelle ENG 1512 0 2003
51 Weaver, Lauren ENG 1505 1 1999
52 Boztas, Lana ENG 1497 4 1947
53 Hapeshi, Eleanor ENG 1492 0 1999
54 Mihajlova, Diana ENG 1478 0 1958
55 Somton, Anita ENG 1464 0 2004
56 Raghavendra, Mahima ENG 1451 0 2003
57 Ragavan, Soumiya ENG 1429 2 2000
58 Purvis, Anna ENG 1422 2 2000
59 Davidson, Laura ENG 1384 1 2003
60 Rahulan, Thivyaa ENG 1381 0 2004
60 Strong, Zoe ENG 1381 0 1998
62 Daniel, Sharon ENG 1375 0 2002
63 Chadwick, Susan E ENG 1370 0 1950
64 Hansen, Annalise ENG 1191 0 2004
65 Haridas, Navieinaah ENG 1154 0 2006
66 Hastilow, Sarah ENG 1152 3 1966

[Roger de Coverly]

There used to be a facility to rank by country. This would calculate the average of the top ten active players. When Matthew Sadler returned to active play, this boosted ENG's ranking. I was expecting to see the same effect with the return of John Nunn. This ranking list by country has either been withdrawn or made very difficult to uncover.

[David Sedgwick]

There used to be a facility to rank by country. This would calculate the average of the top ten active players. When Matthew Sadler returned to active play, this boosted ENG's ranking. I was expecting to see the same effect with the return of John Nunn. This ranking list by country has either been withdrawn or made very difficult to uncover.

I didn't have any trouble finding it.

https://ratings.fide.com/topfed.phtml

John Nunn's return has moved England up from 14th to 12th.

[Keith Arkell]

When are FIDE going to listen to the mathematicians rather than the top players, and do something about the 'K-factor' ? 10 worked very well when there was a list every 6 months, because it was easy to play 72 games during that period, and thus end it with a representative rating.

With a list now appearing every month it is very easy to demonstrate why, with a k-factor of 10, it will take a very long time for a player's strength and rating to align, when his strength is on the increase or decrease.

One solution, in order to protect the stability/status quo at the top might be to, eg, have k-factor 10 for those who have reached 2700, 15 for those who have reached 2600, and 20 for everyone else; while maintaining the necessary anti- deflationary device of k-factor 40 for juniors until they hit 2300.

[NickFaulks]

Keith, you want your FIDE rating to be a performance rating, which is understandable given that you have been doing well lately. However, that is not its motivation. The system is devised to optimise its predictive power, and by inference incorporates an element of reversion to the mean. The radical changes made this year were the result of detailed analysis including every rated game played since we started getting reliable game-by-game results, about eight years ago. For players rated above 2400, k=10 was found to be satisfactory, which is why it was not changed.

As a very surprising by-product ( to me, at least ) the analysis also put to bed the long running argument over "rating inflation".

[E Michael White]

....... The radical changes made this year were the result of detailed analysis including every rated game played since .......".

Hello Nick, is any of this analysis available ?

The system is devised to optimise its predictive power...............

How is the predictive power defined and tested ?

[NickFaulks]

How is the predictive power defined and tested ?

A good place to start is here.

https://www.kaggle.com/c/ChessRatings2

How can I do those neat hyperlinks that other people use?

For what it's worth, there are only two unbiased loss functions ( that must be well known, but I couldn't find it in the literature so I proved it myself ) and I prefer the other one. It turns out that it doesn't make much difference.

[NickFaulks]

Hello Nick, is any of this analysis available ?

This is an excellent summary.

http://en.chessbase.com/post/sonas-over ... tem-220813

[Ian Thompson]

https://www.kaggle.com/c/ChessRatings2

How can I do those neat hyperlinks that other people use?

Use the URL button - Kaggle Chess Ratings

(Quote this post to see what it looks like.)

[NickFaulks]

https://www.kaggle.com/c/ChessRatings2

How can I do those neat hyperlinks that other people use?

Use the URL button - Kaggle Chess Ratings

(Quote this post to see what it looks like.)

Thanks!

[Keith Arkell]

[quote="NickFaulks"]Keith, you want your FIDE rating to be a performance rating, which is understandable given that you have been doing well lately.quote]

Not at all Nick, I've been arguing for 5 years, on forums, on facebook and to anybody who would listen to me, that k-factor 10 has been too sluggish from the moment the rating lists began being calculated every 2 months, in September 2009.

In fact I'd say that someone my age ought to be arguing for small k-factors, or even a factor of zero out of fear of future decline!

Let me try to demonstrate what I mean:

Imagine a 2500 rated junior who is in reality 2700 strength, and for further simplicity, let's imagine that he plays every game at exactly 2700 strength.

14 years ago, when there was a rating list every 6 months, he could quite comfortably have played 77 games during that period, after which he would have ended up being correctly rated at 2700 in the following list, using a K-factor of 10 ( ie the number by which you multiply the difference between the player's expected score and his actual score).

Let us see what happens to his rating now that we have a list published every month, while maintaining a k-factor of 10:

After 1 month 2531
After 2 months 2560
After 3 months 2585
After 4 months 2606
After 5 months 2623
After 6 months 2637

What a difference! Suddenly, despite playing 77 games at 2700 level, his rating is still 63 points less than his playing strength!

Let's continue, with him playing at 2700 strength for another 6 months:

2649, 2658, 2666, 2672, 2677, 2681.

So, after a full year his rating finally gets within about 20 of his playing strength.

Next, let us see what happens if, as I am recommending, K-factor is increased to 20 now that we publish the ratings every month:

2562, 2611, 2642, 2663, 2676, 2684, and bingo, we have the player's rating and playing strength reasonably aligned ( just 16 points short) after 6 months.

This is what I meant when I said the current k-factor is too sluggish.

To sum up, of course a player's rating shouldn't be his performance over just one or two recent events, but equally it would be nice if the rating reflected his performance over a longer time, such as 6 months of active chess, as in the example I gave.

[Richard Bates]

It doesn't seem quite so "sluggish" when you're losing points though!

I think you're missing the point a bit though Keith. Your argument is motivated by setting "K"-factors to accommodate individuals whose rating is at signficant variance to their true strength. You are also to some extent assuming that the system was perfect before, and therefore the motivation of the current K-factors under frequent lists should be replicate what happened with old K-factors under infrequent ones. However it takes no account of individuals whose rating is a pretty accurate reflection of their true strength (logically a person whose rating is absolutely 'correct' should have a "K"-factor of 0). As Nick i think is saying, if the system aims to make/is the dependent on the predictive element as accurate as possible, K-factors need to be set to balance out these two competing extremes (whilst factoring in all the people in the middle, as well as accounting for the fact that some people play very small numbers of games, and some people large numbers). Whilst as you have suggested, a K-factor of 10 might not "work" for one isolated individual who is 200pts off their 'true' strength playing 77 games per year, the question for those deciding K-factors is how much such an individual (or number of such individuals) will affect the predictive value of the system as a whole. Clearly the rarer such individuals are within the rating pool the less effect they will have. And as such individuals are likely to be rarer (relative to the overall pool) at higher rating levels, it makes sense for the K-factors at those levels to be much lower, to avoid random noise from the majority of individuals having a material effect on the system.

[Roger de Coverly]

As Nick i think is saying, if the system aims to make/is the dependent on the predictive element as accurate as possible

A less ambitious objective would be to make the ranking order as correct as possible. That is after all the primary use of rating lists. For British players under around 2100, that is close to being totally broken, essentially because of far too many players having ratings several hundred points adrift from their true strength. Many national rating systems have mechanisms where a rating can be forgotten or nearly so if a player is performing at a couple of hundred Elo points or more above their previous rating. The International system is trying to compensate for this by having the K=40 for lower rated young players. This has been running for less than six months, so it remains to be seen whether it resolves the problems or just creates even more rating instability.

The case for not having a higher K factor with more frequent lists is the stability one. With six monthly lists, a player stands still if they play half their games well above their rating and the other half well below. With monthly lists, the rating will fluctuate and by more depending on how high is the k factor.

[Christopher Kreuzer]

I think Keith's example above is for those playing about 12 games a month. Richard mentioned those playing a small number of games. Typically if you only play 4NCL you will get at most 2 games rated a month, and 3 in May. Has the move to monthly calculations had a big effect here? One effect, pointed out by David Sedgwick, is being unable to remember your rating as it fluctuates by small amounts.

[NickFaulks]

Keith, you want your FIDE rating to be a performance rating, which is understandable given that you have been doing well lately.

Not at all Nick.

Sorry, but that is exactly what you are arguing for in your post. I'm not saying it's daft, but it isn't how the system works.

Imagine a 2500 rated junior who is in reality 2700 strength

I can, but only just. If it was ever possible for such a beast to exist, under the new rules, with k=40 for juniors up to 2300 and then k=20 up to 2400, I don't see how anyone's 2500 rating could be so wrong. The rating system is devised to be as accurate as possible for all players, and cannot be tailored specifically for cases which are somewhere between once-a-decade and mythical.

but equally it would be nice if the rating reflected his performance over a longer time, such as 6 months of active chess

No, it wouldn't. Even over a six month period, if a player's performance varies from his rating, this may be assumed to reflect some combination of change in playing strength and random fluctuation. You wish to throw out the second input, but it has been shown ( and, frankly, is obvious ) that this does not produce the best results.