PDA

View Full Version : Changes in FIDE Title and Rating Regulations



Bill Gletsos
19-03-2009, 09:05 PM
From the 1st July 2009 the following changes will apply to FIDE titles.

1) Only one player can having a rating uplift, instead of two.
2) The rating of the player uplifted has been reduced by 50 points. Hence the uplift for a GM norm is now 2200 (was 2250), IM norm 2050 (was 2100).
3) Performance requirements and requirements of the opponents’ ratings have been changed to even tens. i.e GM norm used to be 2601 performance, now 2600, minimum average rating of opponents used to be 2381, now 2380.
4) For any given title norm at least 1/3 of a players opponents must hold that title. i.e. For a GM norm 1/3 must hold GM title. Whilst in comparison to the old regulations this makes no change to a 9 round event, it would effect a 12 round event where 4 GM's would be required (see to be 3) However until the full regulations are listed it is whether it impacts events with 10 or 11 rounds will depend on whether requirement is rounded up or down. I suspect it will be up.
5) Only the following time controls are permitted for FIDE title events.
a) 90 minutes for 40 moves, followed by 30 minutes for remaining moves, with an increment of 30 seconds per move, starting from move 1
b) 100 minutes for 40 moves, then 50 minutes for 20 moves, then 15 minutes for remaining moves, with an increment of 30 seconds per move, starting from move 1
c) 90 minutes plus 30 seconds per move
d) 120 minutes for 40 moves, followed by 30 minutes for remaining moves
e) 120 minutes for 40 moves, then 60 minutes for 20 moves, then 30 minutes for remaining moves
f) 120 minutes for 40 moves plus 60 minutes for remaining moves



The following will apply from 1st July 2009 for FIDE ratings.

1) The rating floor is reduced from 1400 to 1200.
2) The rating lists will be bi-monthly.
3) The K factor for players who have ever been rated above 2400 will increase to 20 from 10.
4) The K factor for all other players will increase to 30 from 15.
5) The K factor of 25 for those with less than 30 games is abolished.
6) The maximum rating difference of players in ratings calulations is increased from 350 to 400.
7) Closing date of tournaments to be rated on a list is seven days before the list date, except for official FIDE tournaments listed in the Handbook, which are rated even if they end on the last day before the list date.



The proposal was to award direct titles only in World and Continental Championships. According to existing rules 2/3 of the points in a sub-zonal tournament is sufficient for an IM title.
Honorary President Campomanes informed, that the old idea of FIDE was to, give a title to all winners of a zonal tournament. At that time only seven zones existed.
After a long discussion the chairman summarized, that zonal tournaments should be kept with a direct title, not sub-zonal tournaments. The 2/3-rule should be deleted.
This summary was recommended to the QC.


Titles coming from sub-zonal tournaments will be accepted until the current rules have been changed.

The possibility to have a IM/WIM title by applying the 2/3-rule in a zonal or subzonal should be restricted only to one not depending on whether the tournament was played as a round-robin or Swiss, only 2 FM/WFM titles.From the above it appears that automatic zonal titles in sub zonals (Oceania Zonal) look like they may be no longer available, however wemay have to wait for the publication of the regulations to be certain.

Kevin Bonham
19-03-2009, 09:28 PM
3) The K factor for players who have ever been rated above 2400 will increase to 20 from 10.
4) The K factor for all other players will increase to 30 from 15.

Pleased to see FIDE making its K factors more dynamic.

eclectic
19-03-2009, 09:33 PM
will there be any changes to how long initial ratings blocks remain active?

Bill Gletsos
19-03-2009, 09:34 PM
will there be any changes to how long initial ratings blocks remain active?
From the Qualification Commission minutes


It was decided that the rating of a new player is changed to be cumulative until the required number of nine games to have a published rating is reached.
Games of rated players against unrated players in a tournament are rated for rated players, provided that the unrated player plays at least 3 games against rated players in the tournament, scores at least half a point and has achieved performance that is at least the rating floor. These performance rating is used for calculating the new ratings for the rated players.
New player can achieve the nine games required in a period of two years.

Bill Gletsos
19-03-2009, 09:41 PM
Pleased to see FIDE making its K factors more dynamic.About time.

Good to see they have also reduced the cut-off for rating non FIDE events to 7 days prior to the list publication date.

Desmond
20-03-2009, 06:58 AM
From the 1st July 2009 the following changes will apply to FIDE titles.

1) Only one player can having a rating uplift, instead of two.
2) The rating of the player uplifted has been reduced by 50 points. Hence the uplift for a GM norm is now 2200 (was 2250), IM norm 2050 (was 2100).
I would think (1) is going to make it a whole lot harder and (2) is not going to counteract that much, if at all.

eclectic
24-03-2009, 03:54 PM
i meant to ask this before but what effect does the change in K factors have on those trying to eek past a certain rating for example david smerdon?

Ian Rout
24-03-2009, 04:07 PM
i meant to ask this before but what effect does the change in K factors have on those trying to eek past a certain rating for example david smerdon?
On the one hand it helps since you only have to get past the limit once so if the amplitude of the swings is greater there's more chance of the upswing taking you past it. Moreover you will have to play fewer under-rated opponents as fast improvers will move quicker.

On the other hand if you are relying on beating over-rated downwardly-mobile players the above effect will be the opposite, though this doesn't apply in the case you have in mind.

So i'd say it should help, on first thoughts.

Kevin Bonham
24-03-2009, 09:42 PM
On the one hand it helps since you only have to get past the limit once so if the amplitude of the swings is greater there's more chance of the upswing taking you past it.

And also some of the players who are looking for ratings points to get titles are genuinely underrated and hence will move up and get their titles faster

A likely effect of the new system is still further acceleration in the numbers of titled players but that's been a growing issue for decades with little attempt to control it.

eclectic
29-03-2009, 06:04 PM
will events which are due to be rated for the july 1 2009 list be rated under the old or new system or will the new system apply to events played after july 1 2009?

Bill Gletsos
29-03-2009, 06:14 PM
will events which are due to be rated for the july 1 2009 list be rated under the old or new system or will the new system apply to events played after july 1 2009?I expect the July list to go down to the new limit of 1200, however as for the application of the new K-factors etc I expect they will only apply to events that start on or after 1st July 2009.

Bill Gletsos
09-04-2009, 10:58 PM
FIDE have posted the new Title Regulations and the new Rating Regulations that come into effect from 1st July 2009 on their website.

One item that was not mentioned previously is that the time control of 90 minutes with 30 seconds cumulative increment for each move starting from first move is only valid for title events until 30th June 2010.

Garvinator
10-04-2009, 09:55 AM
FIDE have posted the new Title Regulations and the new Rating Regulations that come into effect from 1st July 2009 on their website.

One item that was not mentioned previously is that the time control of 90 minutes with 30 seconds cumulative increment for each move starting from first move is only valid for title events until 30th June 2010.
What do fide define as title events?

Bill Gletsos
10-04-2009, 01:09 PM
What do fide define as title events?Events in which players can gain Title norms.

Garvinator
10-04-2009, 06:36 PM
Events in which players can gain Title norms.
So looks like the 40 moves in 90 minutes, with 30 minutes to finish and 30 seconds increment from move one will be the minimum fide time control approved.

Talk about hard yakka for some events :eek:

antichrist
10-04-2009, 07:27 PM
FIDE have posted the new Title Regulations and the new Rating Regulations that come into effect from 1st July 2009 on their website.

One item that was not mentioned previously is that the time control of 90 minutes with 30 seconds cumulative increment for each move starting from first move is only valid for title events until 30th June 2010.

what will happen after that? To the increment I mean?

eclectic
10-04-2009, 07:29 PM
what will happen after that? To the increment I mean?

the increment is for each move from the beginning

antichrist
10-04-2009, 07:57 PM
the increment is for each move from the beginning

but will it still be on special after that close off date? what will happen then?

eclectic
10-04-2009, 07:59 PM
but will it still be on special after that close off date? what will happen then?

EACH MOVE

Garvinator
10-04-2009, 08:03 PM
EACH MOVE
ec, you are making two fatal mistakes:
1) Attempting to debate with ac at all
2) In attempting to debate or discuss anything with ac, you are under the delusion you are dealing with someone interested in a logical conversation/debate.

You have committed two mistakes in a row here. You should be ashamed of yourself :whistle: Now apologise to everyone else for dragging this thread off course :uhoh:

eclectic
10-04-2009, 08:08 PM
i thought close off date was code for when the 40/90 and the 30min extra had been added not as in can the 90 30 control still be used by lesser tournaments as a/c seems to be implying

Bill Gletsos
10-04-2009, 09:44 PM
what will happen after that? To the increment I mean?It means that after 30th June 2010 that the time control cannot be used for FIDE title events.

It can still be used for other events.

eclectic
10-04-2009, 09:54 PM
in the early section of the regulations it has time controls labelled d) e) f) the ones with no increment

are these de jure or de facto ie only for use when a digital clock is unavailable

i would have thought that for title events especially there ought not ever be a stage in a game where a move is not required to be recorded

antichrist
10-04-2009, 10:10 PM
It means that after 30th June 2010 that the time control cannot be used for FIDE title events.

It can still be used for other events.

so will any increment be used for title events after thta date?

eclectic
10-04-2009, 10:11 PM
antichrist

read post 1

antichrist
10-04-2009, 10:19 PM
a quick look I did not notice that cut off date that Bll mentions of 2010

eclectic
10-04-2009, 10:20 PM
post 1 was a draft; final regulations at www.fide.com

antichrist
11-04-2009, 08:03 PM
Bill, after I was rudely interrupted last night did you give an answer in shoutbox if there will be any increment after that closing date? I don't go chasing back ages ago.

Or was GG estimation correct?

pappubahry
11-04-2009, 08:52 PM
Bill, after I was rudely interrupted last night did you give an answer in shoutbox if there will be any increment after that closing date? I don't go chasing back ages ago.

From the new regulations (http://www.fide.com/images/stories/news2009/title__rating_regulations/title_regulations_1.7.2009.pdf):


The tournament must be played by using one of the following rates of play:
90 minutes with 30 seconds cumulative increment for each move starting from first move
(This time control is valid only until 30.6.2010.)
90 minutes for 40 moves + 30 minutes with 30 seconds cumulative increment for each move starting from the first move
100 minutes for 40 moves followed by 50 minutes for 20 moves, then 15 minutes for the remaining moves with 30 seconds cumulative increment for each move starting from first move
40 moves in 2 hours followed by 30 minutes for the rest of the game
40 moves in 2 hours followed by 60 minutes for the rest of the game
40 moves in 2 hours followed by 20 moves in 1 hour followed by 30 minutes for the rest of the game

So there will still be increments allowed after 30/6/10.

Denis_Jessop
11-04-2009, 09:11 PM
From the new regulations (http://www.fide.com/images/stories/news2009/title__rating_regulations/title_regulations_1.7.2009.pdf):



So there will still be increments allowed after 30/6/10.

Also, as pointed out by Bill, above, those time limits are only mandatory for events in which title norms may be gained. The time limits for events that are simply to be FIDE-rated are in the rating regulations. Events that are neither for FIDE title norms nor FIDE rating can use whatever they like bearing in mind the ACF rating requirements if the event is to be ACF-rated.

DJ

antichrist
11-04-2009, 09:21 PM
From the new regulations (http://www.fide.com/images/stories/news2009/title__rating_regulations/title_regulations_1.7.2009.pdf):



So there will still be increments allowed after 30/6/10.

thank you very much for the bad news, but I never really expected they would go backwards.

Bill Gletsos
14-04-2009, 02:33 PM
I expect the July list to go down to the new limit of 1200, however as for the application of the new K-factors etc I expect they will only apply to events that start on or after 1st July 2009.The new rating floor of 1200 will come into effect on the September 2009 list, not the July 2009 list.

Bill Gletsos
14-04-2009, 02:50 PM
In the past the regulations required that for a GM norm at least 3 opponents needed to be GM's

As of 1st July for norms 1/3 of opponents must already have the applicable title. i.e for GM norms 1/3 must be GM's.

What this means is that for 9 round events there is no change but for 10 and 11 round events you now need 4 GM opponents for a GM norm.


1.45 Titles of opponents
1.45a At least 50% of the opponents shall be title-holders (TH) as in 0.31, excluding CM and WCM. For exact numbers see table in 1.7.
1.45b For a GM norm at least 1/3 with a minimum 3 of the opponents (MO) must be GMs. For exact numbers see table in 1.7.
1.45c For an IM norm, at least 1/3 with a minimum 3 of the opponents (MO) must be IMs (GM counts as if 1.5IM). For exact numbers see table in 1.7.
1.45d For a WGM norm, at least 1/3 with a minimum 3 of the opponents (MO) must be WGMs (GM=1.5WGM, IM=1WGM, FM=1WGM). For exact numbers see table in 1.7.
1.45e For a WIM norm, at least 1/3 with a minimum 3 of the opponents (MO) must be WIMs (GM=2WIM, WGM=1.5WIM, IM, FM=1.5WIM). For exact numbers see table in 1.7.
1.45f Double round-robin tournaments need a min 1.45b-e shall be counted only once.


1.7 Summary of Requirements for the Number of Opponents


Number of games 7 8 9 10 11 12 13 14 15 16 17 18 19
Number of MO 3 3 3 4 4 4 5 5 5 6 6 6 7
Minimum title holders TH 4 4 5 5 6 6 7 7 8 8 9 9 10
Minimum rated 6 7 7 8 9 10 11 11 12 13 14 15 15
Maximum players from applicant’s federation - - 5 6 6 7 7 8 9 9 10 10 11
Maximum players from one federation - - 6 6 7 8 8 9 10 10 11 12 12

Bill Gletsos
25-04-2009, 01:52 AM
Latest information from FIDE regarding the new K-factors in rating calculations.


Following the publication of the new Rating Regulations which will come into effect from 1 July 2009, concerns have been raised by several players and organisers. During the 10-day working visit of Deputy President Georgios Makropoulos in Singapore with General Secretary Ignatius Leong, the following points were noted and recommended:

1. That especially for major high-level competitions, organisers normally would have an invitation list of players based on their ratings at the time of consideration and invitation. Organisers normally would have a contract with the invited players sometimes as long as one year in advance. Some players would have agreed to participate based on the ratings of the other invited players. The dynamic K=20 factor could have a major effect on the ratings of the invited players one year later.

2. The rating lists are used to seed players for the world championship cycle -Candidates Tournament, World Cup etc... The radical change in the K-factor is a major concern especially among the top players. It was regretted that parallel lists were not calculated before the decision to adopt the new K-factors.

3. The radical change in the K-factor for players rated under 2400 is also a serious concern as this could lead to an inflation for norm categories and title purposes. The potential inflation among players under 2400 will also cause a ripple effect as more players move upwards and cross the level of 2400. A study and review on the title regulations would be necessary as a result of the new K-factor - the rating scales may have to be increased.

4. Qualification Commission Chairman & FIDE Rating Administrator Mikko Markkula will produce a parallel list which uses the new K-factors (30 & 20) for tournament reports submitted from the rating lists starting 1 April 2008 till 1 July 2009. FIDE will publish 2 lists for 1 July 2009 - the normal list based on the K-factors being unchanged (25, 15 & 10) and a parallel list (known as the K-List) using the K-factors (30 & 20). The K-List will not be used but only as a comparison.

5. Until 1 July 2010, FIDE shall continue to publish 2 lists using the different K-factors. FIDE shall then review the 2 lists after 1 July 2010 - based on the recommendations of the Qualification Commission. FIDE would then decide whether to extend the trials, abort or to adopt the new K-factors with effect from 1 July 2011 as The Rating List.
Looks like they have changed their mind about the new lists being the official rating lists.

eclectic
25-04-2009, 12:38 PM
Looks like they have changed their mind about the new lists being the official rating lists.

why don't they adopt glicko 2 and be done with it :rolleyes:

sorry, i couldn't resist taking that cheap shot :whistle:

of course some ratings pedants would then want a glicko 1 list too :hand:

Oepty
25-04-2009, 09:09 PM
What on earth are FIDE doing? They made what seemed to be a reasonable decision and then seem to not be able to defend it. They can't even make making the right decision right.
Scott

William AS
26-04-2009, 02:22 PM
What on earth are FIDE doing? They made what seemed to be a reasonable decision and then seem to not be able to defend it. They can't even make making the right decision right.
Scott
Classic post! Very good description of how FIDE has been working recently. :rolleyes:

eclectic
26-04-2009, 02:33 PM
what i have found ironic here is that often lower rated players such as myself get told by higher rated players not to worry about ratings yet here we have high rated prima donna players complaining that they don't want their ratings affected by the new k factor as they accepted tournament invitations some months ago based on what they expected their opponents' ratings to be ie none would be so low such that they might get burnt ratings wise by an unexpected result

shan_siddiqi
01-05-2009, 02:40 AM
what i have found ironic here is that often lower rated players such as myself get told by higher rated players not to worry about ratings yet here we have high rated prima donna players complaining that they don't want their ratings affected by the new k factor as they accepted tournament invitations some months ago based on what they expected their opponents' ratings to be ie none would be so low such that they might get burnt ratings wise by an unexpected result

I still think it's reasonable to say that low-rated players shouldn't worry about their ratings, but it's excusable for high-rated players. When your playing level is at least 2100 or so, you're a legitimate contender for the FM title - in other words, your rating is actually meaningful. There's a big difference between a player rated 2299 and a player rated 2300... that extra rating point is enough to make you a "chess master", which can help you get a great deal of credibility if you want to write chess-related articles/books or teach lessons.

When you're around 1700-1800 or lower, on the other hand, your rating is simply a representation of your performance. In this case, having a higher rating doesn't really come with any practical benefits. If you think you're playing at (or near) master level, then it's OK to worry about your rating; otherwise, a higher rating will only make it harder for you to win a class prize.

Vlad
04-05-2009, 10:54 AM
There is an interesting debate on chessbase regarding K=10 vs K=24 factor. I find all the people defending 24 are putting up very interesting and persuasive arguments indeed. Nunn and the last person from Turkey should really go back to the drawing board.

On the other hand, maybe chessbase is just biased and they select very bad comments defending K=10 on purpose? That way they can form a specific opinion.:)

Bill Gletsos
05-05-2009, 02:20 PM
There is an interesting debate on chessbase regarding K=10 vs K=24 factor. I find all the people defending 24 are putting up very interesting and persuasive arguments indeed. Nunn and the last person from Turkey should really go back to the drawing board.Exactly.

Nunn's comment quoted below is simply wrong.

=John Nunn] The K-factor and the frequency of rating lists are unrelated to one another. Rating change depends on the number of games you have played. If you have played 40 games in 6 months, it doesn't make any difference whether FIDE publishes one rating list at the end of six months or one every day; you've still played the same number of games and the change in your rating should be the same.If a player rated 2500 plays 40 games in a 6 month period and performs at the 2600 level then with k=10 if the list is published every six months then that players new rating will be 2556 on the next list.
If the list is published every 3 months with the player playing 20 games in each 3 month period then after 3 months his rating will be 2528. The folloing 20 games will then use this 2528 rating in comparison to his performance of 2600 in the next 20 games and after the next 3 months his rating will be 2548. This is a difference of 8 points.

If the list is published monthly with the number of games in the following six months being 7, 7, 7, 7, 6, 6 then after each month the players rating will be 2510, 2518, 2526, 2533, 2538, 2543. This is a difference of 13 points compared to the 6 monthly list.

Kevin Bonham
06-05-2009, 02:07 AM
Indeed. I was surprised that Nunn made his "where is the proof?" claim in response to Sonas' claim that K=24 is better when Sonas has prominently defended 24 at length in the past. If Nunn had some issue with Sonas' reasoning that would be one thing, but to just go "where is the proof?" as if Sonas had just plucked 24 out of nowhere when Sonas even gave a link to it, suggests Nunn hasn't even done the most basic research on Sonas' views and is just being lazy. That plus Nunn's incorrect comment mentioned by Bill above just shows that Nunn isn't across what he is talking about. It's not the first time and arguing like that is strange given his chess reputation as a patient, objective and detailed analyst - and even stranger given that he is a mathematician and should be able to evaluate Sonas' argument. Will be interesting to see his promised follow-up.

One laziness deserves another so rather than even consider any other possibility, I'll just assume Nunn's comments are the standard line you get from some inactive and well-over-the-hill players who believe that even if their rating is decades old they should be able to come back, play a few games, perform waaaaaay below it and still have a rating that says they're 99.5% the player they were when everybody knows this can't be true. And, of course, a system with a too-low k-factor and that doesn't factor in inactivity at all is exactly what they are looking for.

Kevin Bonham
12-05-2009, 02:39 AM
It gets worse. Sheesh, talk about digging a hole deeper ...

Chessbase has now published Nunn's final reply (http://chessbase.com/newsdetail.asp?newsid=5418) and canned the debate.

But Nunn's last reply contains more nonsense. He writes concerning Sonas:


I did read the 2002 ChessBase article by Jeff Sonas. In this article Sonas analyses the results of his rating system and provides evidence that it better predicts future results than the Elo system. It may be true that his system is better in this respect than the Elo system; it’s a complex question. However, it’s important to realise that the Sonas system is fundamentally different to the Elo system – for example, it is based on a linear function rather than a normal probability distribution. What FIDE are going to do is not to adopt the Sonas system in its entirety, but to simply increase the K-factor in their current system. Therefore the possible merits of the Sonas system are not relevant to the current discussion. I stick to my original point, that there’s no proof that increasing K to 20 will improve the current system.

But actually if you read Sonas' article (here (http://www.chessbase.com/newsdetail.asp?newsid=562)) it is quite clear that he is saying (about halfway down) that K=24 works best for ELO ratings specifically - the x-axis of his graph refers to ELO ratings not Chessmetrics ratings.

Also I suspect where Nunn says "it may be true ... it's a complex question" what he means is of course it's true, and he's handwaving to avoid admitting it.

Vlad
12-05-2009, 10:00 AM
I agree that Nunn just makes himself look silly by making most of his comments. However, my interpretation of what Nunn implied about the complexity of the question is that Sonas did not really present a proof.

All what Sonas did was to explain what could have happened if a different rating system was used in application to the data that realized. By a proof mathematicians mean such a result that shows that for all possible realizations for all chess players Sonas system with K=24 dominates the current system with K=10. Ok, one can say that it is impossible to do and I agree with that. However, until that was done Sonas should not really call his simulations a proof. It can be called the best effort to make simulations look like a proof. :lol:

1min_grandmaster
12-05-2009, 10:42 AM
I sent the following message to Chessbase a couple of weeks ago, at a time when people were saying there is no relationship between K-factor and frequency of publishing ratings lists. For example, John Nunn said that "K-factor and the frequency of rating lists are unrelated to one another" [http://www.chessbase.com/newsdetail.asp?newsid=5392]. They have not bothered to publish what I sent, and since they have closed the debate for now, I reproduce it here:


"There is in fact a relationship between the K-factor and the frequency at which ratings lists are published. An increase in K-factor results in an increase in the rate of change of players' ratings. A decrease in the frequency at which the ratings lists are published (i.e. an increase in the period of time between the publishing of ratings lists) can also result in a greater rate of change of players' ratings. This is because there will be more time before the next ratings list is published, and therefore, it is possible to play more games during that period before the rating of a player has been updated to more accurately reflect their true strength.

For example, consider a player who is rated 2300. During a ratings period, this player performs at 2400. This player will earn more points if he/she plays 30 games instead of 20 games during the ratings period. However, if the frequency of publishing ratings lists is increased from every 3 months to every 2 months, and say, the last 10 of those games would therefore belong in the next ratings period, then that player's rating change for those 10 games would be calculated based on that player's newly-published rating. Since their rating would have gone up based on the player's performance on the first 20 games, they will therefore earn less points on the last 10 games than if those 10 games had also been counted in the earlier ratings period.

Therefore, it does indeed make sense to increase the frequency of ratings to counteract the increase in K-factor. It would be good to see some concrete simulation data, but one can expect the following to occur if these two proposed changes are made. The increased K-factor should result in faster rating changes, resulting in more accurate ratings, particularly for under-rated players. The increased frequency of publishing ratings should reduce the chances of a player's rating "overshooting" their performance rating during a period (e.g. a player with an old rating of 2300 performing at 2400 and therefore getting a new rating of 2450).

Finally, I would like to point out that now, compared with a few decades ago, there are many more chess players, games and tournaments. In this new information age, many more players are studying the game and improving quickly, and so it makes sense to adjust the ratings system to accommodate these changes, even if the ratings system has performed well and has been respected in the past. Increasing the K-factor as well as the frequency of publishing ratings lists does exactly this.

Jason Chan.
Sydney, Australia."


In John Nunn's most recent article, he in fact includes a table for the ratings of a player based on different K-values and rating list publication frequencies which confirms exactly what I wrote. But you cannot observe the benefits of these changes to the system unless you analyze the results to all players; I expect that you will find it "fixes" the problems with under/over-rated players while not making much change to "static" players.

The good news is that at the end of the most recent article [http://www.chessbase.com/newsdetail.asp?newsid=5418], it appears that "Sonas will be performing some practical experiments with past FIDE ratings data". This should result in a "proof" of something, one way or another.

Ian Rout
12-05-2009, 01:52 PM
I'm a little lost - is it now the case that official lists (to be used for titles, tournament rankings and cut-offs etc) will continue to be produced as per the status quo, that this will continue until at least July 2011, and any other lists generated are for research purposes?

Bill Gletsos
12-05-2009, 02:39 PM
I'm a little lost - is it now the case that official lists (to be used for titles, tournament rankings and cut-offs etc) will continue to be produced as per the status quo, that this will continue until at least July 2011, and any other lists generated are for research purposes?What is actually happening is unclear. The following appears to be the case:

1) The FIDE General Assembly in Dresden 2008 approved the changes when it accepted the report of the FIDE Qualifications Commission Cjhairman Mikko Markkula.

2) FIDE reported on their website that "During the 10-day working visit of Deputy President Georgios Makropoulos in Singapore with General Secretary Ignatius Leong, the following points were noted and recommended:"

One of the points was:
"Until 1 July 2010, FIDE shall continue to publish 2 lists using the different K-factors. FIDE shall then review the 2 lists after 1 July 2010 - based on the recommendations of the Qualification Commission. FIDE would then decide whether to extend the trials, abort or to adopt the new K-factors with effect from 1 July 2011 as The Rating List."

The report ended noting it was approved by FIDE President Kirsan Ilyumzhinov.

3) It has been pointed out that the FIDE President does not have the authority to overturn a decision of the FIDE General Assembly.

ER
12-05-2009, 03:10 PM
...of Deputy President Georgios Makropoulos
One of the very rare occassions of his name spelled the right way!!! :clap: :clap: :clap:
PS That is as opposed to other international fora! Bill, Kev, Peter, Garvin, Jono and Denis, always spell it right here!

Kevin Bonham
13-05-2009, 12:06 AM
I agree that Nunn just makes himself look silly by making most of his comments. However, my interpretation of what Nunn implied about the complexity of the question is that Sonas did not really present a proof.

No, not in a pure mathematical sense of "proof". At best, what he's doing is reporting what would have worked best for one set of players empirically over the observed time frame. He's not even constructing a model to show why 24 should be best. Even if he did construct a model it would still have to be tested. It looks from Sonas' findings that ELO's model does not match reality perfectly in terms of the percentage expectancy curve.

Another criticism I've encountered and agree with (I recall Bill mentioning this) is that when Sonas says K=24 is better, he isn't actually testing a hypothesis with fresh data and giving it a chance to be falsified. OK, he's falsified the idea that K=10 works best - at least for a very large sample he looked at. What he should do is say "ok, it looks like K=24 is best based on the data up to 2002. Now I'm going to look at the data from 2003-2009 and see if that still holds up for that set or if I get something really different like 16 or 33 instead."


All what Sonas did was to explain what could have happened if a different rating system was used in application to the data that realized. By a proof mathematicians mean such a result that shows that for all possible realizations for all chess players Sonas system with K=24 dominates the current system with K=10. Ok, one can say that it is impossible to do and I agree with that. However, until that was done Sonas should not really call his simulations a proof. It can be called the best effort to make simulations look like a proof. :lol:

Agreed.

antichrist
14-05-2009, 07:25 PM
is it okay if I avoid this thread it is a little too heavy and deep for my simple soul

ER
15-05-2009, 01:38 AM
One of the very rare occassions of his name spelled the right way!!! :clap: :clap: :clap:
Talking about spelling :eek: :hmm:

Garvinator
10-06-2010, 11:56 AM
Ratings Meeting in Athens
The meeting was held in Athens on June 2nd and 3rd. The participants were Mikko Markkula, Stewart Reuben and Nick Faulks from the Qualification Commission, Jeff Sonas and Bartlomiej Macieja who provided the statistical support and David Jarrett.

The meeting concentrated on the current rating system and whether any changes were necessary before the major review of the regulations at the 2012 Congress. At last year’s meeting much concern was expressed about inflation in the system. Although there were differing explanations, it was generally felt that many complex factors were involved. It was agreed that there was no necessity for remedial action at this time.

Several recommendations will be made to the Presidential Board for their consideration, including that there were no technical impediments to a reduction in the rating floor to 1000 or the publishing of monthly lists if this was decided by FIDE. It was agreed that the regulations should be clarified regarding when tournaments must be received for rating in current list.

The current requirement of 9 games for an initial published rating was examined. Statistically there did not seem to be any reason to increase the requirement to 10 or more games. However, there were some reservations for players achieving initial ratings above 2200. It was recommended to leave the requirement of at least 1 point from 3 games against rated players from the first result submitted for unrated players.

New players entering the rating list with unusually high ratings, tend to be overrated and new players entering the list with unusually low ratings, tend to be underrated. It was recommended that the K factor for ‘new’ players (those that have not yet played 30 games) should be increased from 25 to 30. The initial rating would not be affected but it would impact on their rating until they had played 30 games and improve the predictive power of their rating in that period. The effect on the rating list was minimal.

It was recommended that if FIDE decides to go ahead, then a separate rating list for Blitz and a separate list for Rapid should be produced. FIDE should decide on need and cost structure.

A list of items that everyone thought should be considered in the major review for 2012 was made, including whether FIDE should stay with the Elo system or consider alternatives.