GRADING ANOMALIES

General discussions about ratings.
Roger de Coverly
Posts: 21301
Joined: Tue Apr 15, 2008 2:51 pm

Re: GRADING ANOMALIES

Post by Roger de Coverly » Sun Dec 20, 2009 4:41 pm

Robert Jurjevic wrote:and more rapidly than GS for players who played more than 30 games in the season.
I'm not sure I follow why it's more rapid, unless you are proposing to publish more often than annually.

If the ECF grading system, a player playing 60 games a season has the results measured over the same period as a player with 30 games. The ECF has experimented in the past with just grading the most recent 30 games, however this suffers from the drawback that players do not necessarily improve at a continuous rate or even perform at a consistent level for an entire season. If effect a grade published on the most recent 30 for a player with 60 games in a season will ignore half the experience since the last published annual list. This may or not be a better estimate of current strength. If you are publishing top 10s by age based on the published grade, this approach can give a poor ranking to anyone who started well and then faded.

User avatar
Robert Jurjevic
Posts: 207
Joined: Wed May 16, 2007 1:31 pm
Location: Surrey

Re: GRADING ANOMALIES

Post by Robert Jurjevic » Mon Dec 21, 2009 11:02 am

Hello Roger,
Roger de Coverly wrote:
Robert Jurjevic wrote:and more rapidly than GS for players who played more than 30 games in the season.
I'm not sure I follow why it's more rapid, unless you are proposing to publish more often than annually.
Say for a 140 player who scores 50% against a 160 opposition in 60 games in the season a new FGS (FIDE-like Grading System mentioned in my earlier post) grade would be 180 (not 160), if the player played 30 games in the season (also scoring 50% against a 160 opposition) his new FGS grade would be 160 (the same as it would be assigned by GS), if the player played 15 games in the season (also scoring 50% against a 160 opposition) his new grade would be 150 (the same as it would be assigned by AGS3), if the player played 1 game in the season (also scoring 50% against a 160 opposition) his new grade would be 150.67, basically a 140 player is awarded (160-140)/30=0.67 grading points for a draw against a 160 player, and the more games he or she draws the more grading points he or she is awarded.
Roger de Coverly wrote:In the ECF grading system, a player playing 60 games a season has the results measured over the same period as a player with 30 games. The ECF has experimented in the past with just grading the most recent 30 games, however this suffers from the drawback that players do not necessarily improve at a continuous rate or even perform at a consistent level for an entire season. If effect a grade published on the most recent 30 for a player with 60 games in a season will ignore half the experience since the last published annual list. This may or not be a better estimate of current strength.
No, with FGS you would still take into account all of the games a player played in the season (be it 0, 1, 15, 30, 60, 90, 120 etc.) and grade and publish grades only once a season.

The difference is in the formula used, which for non-FIDE-like (or current ECF-like) grading systems is

Code: Select all

g2i = gi + Sum_j ((kij*(qij - pij))/ni)
where 'ni' is the number of games 'i'th player played in the season, and for FIDE-like systems

Code: Select all

g2i = gi + Sum_j ((kij*(qij - pij))/n0)
where 'n0>0' is some constant number of games (in FGS 'n0=30').

The difference between FIDE-like and non-FIDE-like (i.e. current ECF-like) grading systems is that the grade change in current ECF-like grading systems is independent of the number of games the player played in the season (providing it is 30 or more) while the grade change in FIDE-like grading systems is dependent of the number of games the player played in the season (say GS would assign a grade of 160 to a 140 player who scored 50% against a 160 opposition regardless if the player played 30, 60, 90, 120, etc. games in the season, and FGS's grade would depend on the number of games the player played in the season).

My proposed FIDE-like formulae are exactly the same formulae FIDE uses for their ratings but scaled to ECF grades (well also maybe FIDE refers to win as 1 and ECF and me as 100, etc.), basically you correct one's grade for a single game and does that for every game played in the season, you always take grades from the previous season, though, live grades would be more accurate, i.e. you could correct grades after every game (not publish them but use them internally in the calculation).

Ungraded players:
One could set 'kij=0' and 'kji=1' if 'i'th and 'j'th players played at least one game in the season and if 'i'th player is graded and 'j'th player is ungraded (ungraded players' grades change with normal pace and graded players' games against ungraded players do not affect graded players' grades), or one could set 'kij=1' and 'kji>1' if 'i'th and 'j'th players played at least one game in the season and if 'i'th player is graded and 'j'th player is ungraded (ungraded players' grades change faster than graded players' grades).

Juniors (or fast improving or worsening players):
To further help with the "junior problem" one could set 'kij=1' and 'kji>1' if 'i'th and 'j'th players played at least one game in the season and if 'i'th player is not a junior and 'j'th player is a junior (juniors' grades change faster than non-juniors' grades).

Kind regards,
Robert Jurjevic
Vafra

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

Re: GRADING ANOMALIES

Post by Roger de Coverly » Mon Dec 21, 2009 11:15 am

Robert Jurjevic wrote:Say for a 140 player who scores 50% against a 160 opposition in 60 games in the season a new FGS (FIDE-like Grading System mentioned in my earlier post) grade would be 180 (not 160),
I see. You have in fact just reinvented the overshoot problem which is a design "feature" of Elo style systems with infrequent publication. In days gone by with annual publication of the international list, it was possible to overshoot your "true" strength by playing a lot. So if you had a 2300 rating and could play to 2390, you could overshoot to 2400 ( and get a title) by playing lots of games. More frequent publication of rating lists reduces or eliminates this effect because the 2300 gets revalued to something higher and thus the reduced gain per game makes it more difficult to overshoot.

User avatar
Robert Jurjevic
Posts: 207
Joined: Wed May 16, 2007 1:31 pm
Location: Surrey

Re: GRADING ANOMALIES

Post by Robert Jurjevic » Mon Dec 21, 2009 11:45 am

Dear all,
Roger de Coverly wrote:
Robert Jurjevic wrote:Say for a 140 player who scores 50% against a 160 opposition in 60 games in the season a new FGS (FIDE-like Grading System mentioned in my earlier post) grade would be 180 (not 160),
You have in fact just reinvented the overshoot problem which is a design "feature" of Elo style systems with infrequent publication. In days gone by with annual publication of the international list, it was possible to overshoot your "true" strength by playing a lot. So if you had a 2300 rating and could play to 2390, you could overshoot to 2400 ( and get a title) by playing lots of games. More frequent publication of rating lists reduces or eliminates this effect because the 2300 gets revalued to something higher and thus the reduced gain per game makes it more difficult to overshoot.
If you would internally (for the calculation purposes) grade (but not publish grades) after every game, one should eliminate the 'overshoot problem' (say then the 140 player would after every game become approximately a 140.67, 141.96, 143.76, 145.93,... player, and he should never 'overshoot' 160),

So, you could correct grades after every game (not publish them but use them internally in calculation). I wonder how difficult would that be to implement (actually one would need date and possibly time when each game ended)?.

If you would ever be tempted to choose a better ECF grading system and try to implement it in practice, I would be very happy to provide free help with the computer implementation (I could write a Windows .net console command line application in C# programming language interfacing with Oracle Database 10g Express Edition database which would calculate new grades and store them into database, if free Oracle Database 10g Express Edition wouldn't be enough in real life one should buy a commercial version of the database; also, which would look to me as a more natural route, I may be able to provide help with the implementation within the current IT solution).

Kind regards,
Robert Jurjevic
Vafra

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

Re: GRADING ANOMALIES

Post by Roger de Coverly » Mon Dec 21, 2009 12:08 pm

Robert Jurjevic wrote:
So, you could correct grades after every game (not publish them but use them internally in calculation). I wonder how difficult would that be to implement (actually one would need date and possibly time when each game ended)?.
In practice the dates are not available, mostly because of leagues not recording when games took place and also because of adjournments and adjudications making the actual date of the game less well-defined. Also league results are often submitted in bulk at the end of the season.

A major weakness of the current ECF system is its inability to provide updates during the season either to accelerate the frequency of rebasing the list or even to report the story so far in a manner akin to the international "live ratings". The process of estimating new players (now extended to juniors) is unhelpful in this respect since an established player's performance against an unrated player is measured by reference to games that haven't been played it. To explain this point, if I play against a unrated player or junior I will not know the grade to use in the calculations until next August. This would thwart an attempt to publish in December an unofficial list covering the first half of the season.

User avatar
Robert Jurjevic
Posts: 207
Joined: Wed May 16, 2007 1:31 pm
Location: Surrey

Re: GRADING ANOMALIES

Post by Robert Jurjevic » Mon Dec 21, 2009 1:07 pm

Hello Roger,

Live grading (internally) in FIDE-like grading sytems...
Roger de Coverly wrote:
Robert Jurjevic wrote:So, you could correct grades after every game (not publish them but use them internally in calculation). I wonder how difficult would that be to implement (actually one would need date and possibly time when each game ended)?.
In practice the dates are not available, mostly because of leagues not recording when games took place and also because of adjournments and adjudications making the actual date of the game less well-defined.
I do not know how difficult would be to enforce that when one writes the result of the game on his or her score sheet that one also writes the date and the time when the game ended? If that info would be provided with every game result one would be able to use live grades internally in a FIDE-like grading system.
Roger de Coverly wrote:Also league results are often submitted in bulk at the end of the season.
This should not be much of a problem, it is enough that you get the game results on time for grading, then the software would sort the games in chronological order and grade using live grades internally. You would still publish the grades only once a season. (well if you play a player near the end of the season, say fast improving junior, you may not know which grade of his will be taken into account when adjusting your grade for that game, the junior probably scored well in the season and your grade would be adjusted using junior's live grade which should be higher than the grade from the grading list, the same holds when adjusting junior's grade for that game, your live grade will be taken into account when adjusting junior's grade)
Roger de Coverly wrote:A major weakness of the current ECF system is its inability to provide updates during the season either to accelerate the frequency of rebasing the list or even to report the story so far in a manner akin to the international "live ratings". The process of estimating new players (now extended to juniors) is unhelpful in this respect since an established player's performance against an unrated player is measured by reference to games that haven't been played it. To explain this point, if I play against a unrated player or junior I will not know the grade to use in the calculations until next August. This would thwart an attempt to publish in December an unofficial list covering the first half of the season.
If one would use a FIDE-like grading system a simple solution to ungraded players and juniors could be to increase 'k' factors for both ungreded players and juniors relatively to other players. (estimated grades of ungraded players would affect the grades of graded players but one would make the grades of ungraded players to change faster which should ensure that the estimated grades are corrected as fast as possible, there is also a possibility to make ungreded players' grades to change at the normal pace and when grading graded players ignore games against ungraded players) (one can assume that juniors in principle improve faster than adults and one can make junior grades to change faster than adult grades) (proper solution to "junior problem" would probably involve a method of assessing which players should be given higher 'k' factors and for how much, say you may have a fast improving adult too) (actually if you grade using live grades internally you may need not to adjust juniors' 'k' factors at all, you grade them as anybody else, the fact that you use live grades may be enough to account for their fast improvement)
:)

The FIDE-like 'matrix' model with (internal) live grades...

Restrictions:

(none)

Definitions:

o 'n0' is some constant number of games, 'n0>0', note that if 'kij=kji=1' if 'i'th and 'j'th players played at least one game in the season and 'kij=kji=0' if 'i=j' or if 'i'th and 'j'th player did not play each other in the season, then 'n0' is a number of games after which a player's new grade would equal his or her actual performance

o 'm' is the total number of players in the system, 'm>0'

o 'n_' is game number vector where 'ni' is a number of games 'i'th player played in the season

o 'g_' is a grade vector where 'gi' is a live grade of 'i'th player, a live grade for the first game in the season for a player is a player's grade from the previous season

o 'g2_' is a new grade vector where 'g2i' is a new grade for the season of 'i'th player (player's new grade for the season is calculated using the player's game results in the season and the player's and the player's opposition live grades)

o 'Q_' 'is 'mxm' matrix where 'qij' is actual performance of 'i'th player against 'j'th player (if the players 'i' and 'j' played one game only then 'qij' is 0 if 'i'th player lost, 50 if 'i'th player drew and 100 if 'i'th player won, if the players 'i' and 'j' played more than one game than '0<=qji<=100', note that 'qij' and 'qji' are related, i.e., if 'qij=0' then 'qji=100','if 'qij=50' then 'qji=50', if 'qij=100' then 'qji=0', etc.), note that 'qij+qji=100' if 'i'th and 'j'th players played at least one game in the season, note that 'qij = qji = 0' if 'i=j' or if 'i'th and 'j'th player did not play each other in the season, 'Q_' is actual performance matrix

o 'P_' is 'mxm' matrix where 'pij=f(gi-gj)' is expected performance of 'i'th player in the game against 'j'th player, note that ECF is using 'f(dij=gi-gj)=50*(1 + (gi-gj)/50)' if '|dij=gi-gj|<=40' and 'f(dij=gi-gj)=90' if 'dij=gi-gj>40' and 'f(dij=gi-gj)=10' if 'dij=gi-gj<-40', FIDE logistic equivalent is 'f(dij=gi-gj)=100/(1 + 10^(-(gi-gj)/50))', note that if 'gi=gj' then 'f(gi-gj)=50', note that 'pij+pji=100', 'P_' is expected performance matrix

o 'K_' is 'mxm' matrix where 'kij' are 'k' factors used in calculation of new grades for the season, 'kij=kji=0' if 'i=j' or if 'i'th and 'j'th player did not play each other in the season, 'kij+kji=k', where 'k>0' is a constant, if 'i'th and 'j'th players played at least one game in the season, 'K_' is 'k' factors matrix

New grades for the season are calculated as follows (this is a generic formula covering a number of grading systems including the current ECF one, what makes the grading systems different is a choice of matrices 'K_' and 'P_')

Code: Select all

g2i = gi + Sum_j ((kij*(qij - pij))/n0)
where for FGSl (FIDE-like gradign sytem with internal live grades) 'kij=kji=1' if 'i'th and 'j'th players played at least one game in the season, 'kij=kji=0' if 'i=j' or if 'i'th and 'j'th player did not play each other in the season, 'pij=f(dij=gi-gj)' where 'f(dij=gi-gj)=100/(1 + 10^(-(gi-gj)/50))' and 'n0=30'.

Kind regards,
Last edited by Robert Jurjevic on Mon Dec 21, 2009 2:12 pm, edited 1 time in total.
Robert Jurjevic
Vafra

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

Re: GRADING ANOMALIES

Post by Roger de Coverly » Mon Dec 21, 2009 1:45 pm

Robert Jurjevic wrote:You would still publish the grades only once a season.
Sooner or later, the ECF will have to face up to the issue that publishing grades once a season ( and nearly two months after the cutoff) is looking increasingly obsolete. For example FIDE publishes every two months with plans to publish monthly and has a deadline for submission of a couple of weeks. Also results are posted to the website as received so that anyone can establish their "live" rating just by adding two numbers together.

The problem is that the ECF system creaks somewhat when you try to publish more frequently than annually or publish results processed in the year so far. Six monthly updates have been tried with the rapid-play list. One of the issues that your grade can change without a single game being played. Systems like the current process for new players or juniors which postpone everything to the end of the season hinder the information flow. Organisers and players would like to know up to date rankings even if they don't choose to use them for eligibilities, pairings and board orders.

User avatar
Robert Jurjevic
Posts: 207
Joined: Wed May 16, 2007 1:31 pm
Location: Surrey

Re: GRADING ANOMALIES

Post by Robert Jurjevic » Mon Dec 21, 2009 2:20 pm

Hello Roger,

FIDE-like grading system with internal live grade calculation...
Roger de Coverly wrote:
Robert Jurjevic wrote:You would still publish the grades only once a season.
Sooner or later, the ECF will have to face up to the issue that publishing grades once a season ( and nearly two months after the cutoff) is looking increasingly obsolete. For example FIDE publishes every two months with plans to publish monthly and has a deadline for submission of a couple of weeks. Also results are posted to the website as received so that anyone can establish their "live" rating just by adding two numbers together. ... The problem is that the ECF system creaks somewhat when you try to publish more frequently than annually or publish results processed in the year so far. Six monthly updates have been tried with the rapid-play list.
If ECF would use a FIDE-like grading system with internal live grade calculation (say FGSl described below), ECF would be free to publish the grades either once a year, twice a year, once every two months or even provide live grades online.

If you calculate using live grades internally the frequency of grade publication does not affect the grade calculation at all, though it may affect the things such as determination who is eligible to play in certain events, etc.

I wouldn't go so far to push grade publication more than once a year and would be delighted if ECF would consider adopting one of the FIDE-like grading system with internal live grade calculation (say FGSl described below), once that has been done, say grade publication every two months is more of an administrative issue (i.e. you collect the game results and run the program instead of once a year every two months).

I think that the biggest obstacle to adopting one of the FIDE-like grading systems with internal live grade calculation (say FGSl described below) could be a need to enforce keeping records of date and time when the games ended (though one would think, if adopting one of the FIDE-like grading systems with internal live grade calculation would solve the "junior problem", would allow a better ways to deal with ungraded players, and in general would provide as up to date grades as possible, that writing down these few extra numbers alongside the game result may be worthwhile).

The FIDE-like 'matrix' model with internal live grades...

Restrictions:

(none)

Definitions:

o 'n0' is some constant number of games, 'n0>0', note that if 'kij=kji=1' if 'i'th and 'j'th players played at least one game in the season and 'kij=kji=0' if 'i=j' or if 'i'th and 'j'th player did not play each other in the season, then 'n0' is a number of games after which a player's new grade would equal his or her actual performance

o 'm' is the total number of players in the system, 'm>1'

o 'n_' is game number vector where 'ni' is a number of games 'i'th player played in the season

o 'g_' is a grade vector where 'gi' is a live grade of 'i'th player, a live grade for the first game in the season for a player is a player's grade from the previous season

o 'g2_' is a new grade vector where 'g2i' is a new grade for the season of 'i'th player (player's new grade for the season is calculated using the player's game results in the season and the player's and the player's opposition live grades)

o 'Q_' 'is 'mxm' matrix where 'qij' is actual performance of 'i'th player against 'j'th player (if the players 'i' and 'j' played one game only then 'qij' is 0 if 'i'th player lost, 50 if 'i'th player drew and 100 if 'i'th player won, if the players 'i' and 'j' played more than one game than '0<=qji<=100', note that 'qij' and 'qji' are related, i.e., if 'qij=0' then 'qji=100','if 'qij=50' then 'qji=50', if 'qij=100' then 'qji=0', etc.), note that 'qij+qji=100' if 'i'th and 'j'th players played at least one game in the season, note that 'qij = qji = 0' if 'i=j' or if 'i'th and 'j'th player did not play each other in the season, 'Q_' is actual performance matrix

o 'P_' is 'mxm' matrix where 'pij=f(gi-gj)' is expected performance of 'i'th player in the game against 'j'th player, note that ECF is using 'f(dij=gi-gj)=50*(1 + (gi-gj)/50)' if '|dij=gi-gj|<=40' and 'f(dij=gi-gj)=90' if 'dij=gi-gj>40' and 'f(dij=gi-gj)=10' if 'dij=gi-gj<-40', FIDE logistic equivalent is 'f(dij=gi-gj)=100/(1 + 10^(-(gi-gj)/50))', note that if 'gi=gj' then 'f(gi-gj)=50', note that 'pij+pji=100', 'P_' is expected performance matrix

o 'K_' is 'mxm' matrix where 'kij' are 'k' factors used in calculation of new grades for the season, 'kij=kji=0' if 'i=j' or if 'i'th and 'j'th player did not play each other in the season, 'kij+kji=k', where 'k>0' is a constant, if 'i'th and 'j'th players played at least one game in the season, 'K_' is 'k' factors matrix

New grades for the season are calculated as follows (this is a generic formula covering a number of grading systems, what makes the grading systems different is a choice of matrices 'K_' and 'P_')

Code: Select all

g2i = gi + Sum_j=1,m ((kij*(qij - pij))/n0)
where for FGSl (FIDE-like Grading System with internal live grades) 'kij=kji=1', 'kij+kji=k=2', if 'i'th and 'j'th players played at least one game in the season, 'kij=kji=0' if 'i=j' or if 'i'th and 'j'th player did not play each other in the season, 'pij=f(dij=gi-gj)' where 'f(dij=gi-gj)=100/(1 + 10^(-(gi-gj)/50))' and 'n0=30'.

(note that in 'g2i = gi + Sum_j=1,m ((kij*(qij - pij))/n0)' there are 'ni' elements in the sum for which 'kij/=0')

Kind regards,
Last edited by Robert Jurjevic on Mon Dec 21, 2009 3:32 pm, edited 11 times in total.
Robert Jurjevic
Vafra

Alex Holowczak
Posts: 9085
Joined: Sat May 30, 2009 5:18 pm
Location: Oldbury, Worcestershire
Contact:

Re: GRADING ANOMALIES

Post by Alex Holowczak » Mon Dec 21, 2009 2:23 pm

Roger de Coverly wrote: Sooner or later, the ECF will have to face up to the issue that publishing grades once a season is looking increasingly obsolete.
I maintain that by using a computer-based system, it really shouldn't take longer than a week to update it with all the results (from organisers sending them off to graders putting them into the computer). If it takes longer than that, then the technology isn't being used properly, in my opinion.

User avatar
Robert Jurjevic
Posts: 207
Joined: Wed May 16, 2007 1:31 pm
Location: Surrey

Re: GRADING ANOMALIES

Post by Robert Jurjevic » Mon Dec 21, 2009 3:08 pm

Alex Holowczak wrote:
Roger de Coverly wrote: Sooner or later, the ECF will have to face up to the issue that publishing grades once a season is looking increasingly obsolete.
I maintain that by using a computer-based system, it really shouldn't take longer than a week to update it with all the results (from organisers sending them off to graders putting them into the computer). If it takes longer than that, then the technology isn't being used properly, in my opinion.
I would guess that the most advanced system would be one where the game results would be submitted online (by the people responsible and authorized to do so) with live grades being updated constantly. Nevertheless, developing such a software system may not be simple nor cheap, also one should add the cost of the hardware and the services (hosting Internet server, etc.). The system would still be as up to date as much as the submission of the game results is up to date.

Actually even more advanced system would be (this is just a theoretical 'joke') a system where all of the graded games would be played with some sort of DGT chess sets and boards with ability to automatically send (to the grading server) players' IDs (say by reading bar-coded ECF member cards), game results and the date and the time when the games ended (the grading server would then update the affected grades as soon as the game results would be received).
:)
Robert Jurjevic
Vafra

Alex Holowczak
Posts: 9085
Joined: Sat May 30, 2009 5:18 pm
Location: Oldbury, Worcestershire
Contact:

Re: GRADING ANOMALIES

Post by Alex Holowczak » Mon Dec 21, 2009 3:39 pm

Robert Jurjevic wrote: I would guess that the most advanced system would be one where the game results would be submitted online (by the people responsible and authorized to do so) with live grades being updated constantly. Nevertheless, developing such a software system may not be simple nor cheap, also one should add the cost of the hardware and the services (hosting Internet server, etc.). The system would still be as up to date as much as the submission of the game results is up to date.
I really don't see why that should be that difficult, or expensive. If it'll be cheaper and faster in the long run, then it'll help to pay for itself. Most online chess websites can do that sort of thing with Elo already. You just need two different things, 1) different grading algorithm, 2) a manual way of inputting game results. That can't be that hard. Carl provides the hosting for free, I believe. If results were put in on the database, the results could be shown, and people could check their own results.

J T Melsom
Posts: 1294
Joined: Wed Aug 12, 2009 11:12 pm

Re: GRADING ANOMALIES

Post by J T Melsom » Mon Dec 21, 2009 3:58 pm

I am all in favour of modernisation in terms of results reporting and processing, but as a league controller (not grader) , I find a number of clubs still lag behind the times as far as IT are concerned - even the move to e-mail reporting of results on a pro-forma has proved harder than I would have expected.

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

Re: GRADING ANOMALIES

Post by Roger de Coverly » Mon Dec 21, 2009 4:04 pm

Alex Holowczak wrote:I really don't see why that should be that difficult, or expensive. If it'll be cheaper and faster in the long run, then it'll help to pay for itself. Most online chess websites can do that sort of thing with Elo already.
For practical reasons an over-the-board rating system has to be batched based - that is you store up a set of results, for a month, two months, quarter, half year or year and process them as a batch to produce a new rating list. You have a clearly defined process - old rating list, results, new rating list. This is both more reliable and practical than trying to update after every game. I'm well aware that on-line servers do this without apparent difficultly but a moment's reflection tells you this is only in the special case of the subset of games played exclusively on that server.

Also if you are running a batch based system, you want to lock the calculation process on the begin period rating. That way you don't run into trouble when you get an out of sequence result being reported.

Whilst it would be possible to work an annual rating list on a continuous update approach, I think it's only useful as an end season review check as people will want to be able to anticipate the results it will generate. In other words, players would like to know "if I win this I gain x, if I draw y and if I lose z". You can do this from the wall chart for all rated players in the FIDE system and for all adult rated players in the ECF system.

User avatar
John Upham
Posts: 7179
Joined: Wed Apr 04, 2007 10:29 am
Location: Cove, Hampshire, England.
Contact:

Re: GRADING ANOMALIES

Post by John Upham » Mon Dec 21, 2009 8:32 pm

J T Melsom wrote: even the move to e-mail reporting of results on a pro-forma has proved harder than I would have expected.
What about Captains entering the match card directly following the match? Do you think this is possible?

John
British Chess News : britishchessnews.com
Twitter: @BritishChess
Facebook: facebook.com/groups/britishchess :D

Ian Thompson
Posts: 3551
Joined: Wed Jul 02, 2008 4:31 pm
Location: Awbridge, Hampshire

Re: GRADING ANOMALIES

Post by Ian Thompson » Mon Dec 21, 2009 9:10 pm

Alex Holowczak wrote:I maintain that by using a computer-based system, it really shouldn't take longer than a week to update it with all the results (from organisers sending them off to graders putting them into the computer). If it takes longer than that, then the technology isn't being used properly, in my opinion.
The time consuming part of doing grading isn't creating the results file for submission to the ECF. The time consuming part is correctly identifying players, e.g. new players who last played a few years and still have an ECF grading code, which should continue to be used; players who really are new and have been playing in more then one league in the current season, so they need to be graded under one grading code; players with misspelt names creating doubt about who they really are.

Post Reply