Grading Lists: Twice a Year?

[Adam Raoof]

(I will count votes for 'more!' as an implicit vote for 'twice')

I would like to propose that, long-term, we make a decision to have grading lists twice a year. What do you think?

Players new to the game currently have to wait up to 18 months for a grade. This is not a good service on behalf of players. I can't think of any objections to more frequent grading lists that cannot be overcome with some patience and understanding of the structure of English chess, and a little creativity.

We currently have a rapid grading list twice a year. Organisers of tournaments cope well with that, making sure that their entry forms include clarification about the list they will be using for eligibility. It appears to be common sense that if a player has NO grade, then they should have to use the new grade when published even if just before an event.

All countries publish lists more frequently than we do. The Netherlands, for instance, publishes four times a year. They have a similar league structure to us, and they make it a rule that the grade extant at the beginning of the season is the one used throughout. As above, I would add that it appears to be common sense that if a player has NO grade at the start of a season, then they should have to use the new grade when published.

I am not suggesting we rush this through in any way, as I realise there are fine details that have to be dealt with.

You are not allowed to vote 'once' without citing an objection

[Roger de Coverly]

I would like to propose that, long-term, we make a decision to have grading lists twice a year. What do you think?

You should sketch how you would do it. If you change to an Elo style calculation method, then no problems. If you stick with the Clarke style are you

(a) proposing that 30 games in a half year is sufficient for a grade recalculated ignoring all previous grades (with games from previous periods added where needed to make 30)
or
(b) overlap so that the "Xmas" grade contains the second half of "last season" and the first half of this and the "end season" grade is the same as the moment
or
(c) calculate an interim "Xmas" grade which just contains the first half of the season (with games from previous years added in if needed) and an "end season" grade which is the same as the current one.

Be warned that the "throw the old grade away" principle for juniors is working against whatever you do in more frequent publication.

The previous (rejected) proposal was for (b). Both (a) and (c) make more sense to my mind because of the performance measuring nature of the Clarke system.

[Sean Hewitt]

In principle 2 or 4 lists per year is preferable.

However, Roger is right about the obstacles. Leagues can state that lists at the start of the season are good for the whole season, or adjust at each publication as they see fit. Similarly congresses can determine (as we do already for FIDE rated events) which list will be used. The January rapidplay list has a number of discrepancies because of the way it is done, involving overlapping periods, though they are of little importance because most players are not bothered about RP grades. Standard grades are far more important and a working method needs to be found.

[Brendan O'Gorman]

Shouldn't this topic be in the 'Grading Debate' section?

[Joey Stewart]

It is quite a nice idea, although there would need to be some sort of a minimum number of games to get a half yearly rating, otherwise the figures could end up skewed one way or another if somebody had a bad start or an especially good run of form.

[Carl Hibbard]

Be warned that the "throw the old grade away" principle for juniors is working against whatever you do in more frequent publication.

I voted no...

You cannot publish more than once until confidence is restored in the actual calculation method, to be brutal even I have my doubts and I have some involvement

[Mick Norris]

Adam

Happy with the idea that long-term we publish twice

Short-term, we need to know the grades are right, and I'm not sure they are, particularly for juniors

You'll need to explain the Netherlands system quite carefully if you want to get this voted in - there will be lots of objections from leagues and congress, and you don't want graders giving up because they feel you are putting them under more pressure

[Roger de Coverly]

Players new to the game currently have to wait up to 18 months for a grade.

You could actually solve this problem directly. New rule - a player gets a "established" grade as soon as they've played 9 players also with published (or possibly "established") grades. Congresses mostly report results with little delay after the event. You run a quarterly sweep-up and publish the new "established" grades quarterly on the grading website. The "established" grades then count the same as published ones. It would be up to leagues and county match controllers whether they submitted results often enough to be swept up in this manner.

You would of course be scrapping much of the existing "end of season" estimation process with a knock-on effect on junior grades.

[Adam Raoof]

You'll need to explain the Netherlands system quite carefully if you want to get this voted in - there will be lots of objections from leagues and congress, and you don't want graders giving up because they feel you are putting them under more pressure

Theoretically this might not mean any more work for local and regional graders, as they would continue to submit results in a timely fashion. The only difference might be that leagues and congresses would have to consider their rules to take account of the new list. There would be more work at a national level, but the grading team could do two lists a year if it was desirable.

[Mick Norris]

If the league grader feels that they have to submit results during, rather than just at the end of, the season, they might not be happy - if you make it clear that they don't, then that's fine

If you find some widely accepted on-line system for results submission, even better

[Gavin Strachan]

FIDE publish more than once a year, so not sure why the ECF can't? Yes it would mean more pressure for graders to get the results in or someone to contact event organisers for this, but inherently it is a better. From a team perspective, you may have someone having a right crap season or always playing really tough oppo and this might make the end of the season more enjoyable. Conversely, you could get players with grades rapidly improving (commonly known as Juniors) and the grade is keeping up with this improvement. Twice a year would be necessary and people enjoy the movement of grades, causes a lot of excitement.

[Adam Raoof]

If the league grader feels that they have to submit results during, rather than just at the end of, the season, they might not be happy - if you make it clear that they don't, then that's fine. If you find some widely accepted on-line system for results submission, even better

That's a fair point. We would certainly be consulting with league graders over the possible implementation.

Technically it is now possible for league organisers and team captains to input all their results directly thanks to software from Steve Law (Herts) and from John Upham (Surrey) - http://www.borderleague.org.uk/main.php. There are other examples around the country! If anyone knows other examples I would be grateful to hear of them.

[Brian Valentine]

I'm certainly coming from the same angle as others, Sean and Roger in particular.

When changes to the grading rules becomes a topic it usually comes from one particular angle. Here the objective is to get players onto the list early, like mom and apple pie, a good thing. However one needs to look at the side affects by looking at the system in the round.

The main objectives of a grading system , in summary:
1. Getting everyone's ranking as near as possible in the right order
2. Determining thresholds for titles or restricting entries for competitions
3. Giving a player a guide to whether they are improving or declining in strength.

Assuming we are wedded to the embellished Clarke system, and looking at the alternatives Roger sets out

You should sketch how you would do it. If you change to an Elo style calculation method, then no problems. If you stick with the Clarke style are you

(a) proposing that 30 games in a half year is sufficient for a grade recalculated ignoring all previous grades (with games from previous periods added where needed to make 30)
or
(b) overlap so that the "Xmas" grade contains the second half of "last season" and the first half of this and the "end season" grade is the same as the moment
or
(c) calculate an interim "Xmas" grade which just contains the first half of the season (with games from previous years added in if needed) and an "end season" grade which is the same as the current one.

Method a will increase volatility for those who play (well) over 30 games a season potentially affecting objective 1. How those playing 25-35 games a year are treated may also add pitfalls. Methods b and c create a drag on changes in performance (and therefore rankings) compared with a) and makes the junior algorithm even more dubious. It is not at all clear that the existing backfill (for less than 30 games) is statistically efficient and now that dates seem to be available for games graded a more robust approach is likely to be possible.

Any adjustments to the methodolgy may be inflationary/deflationary and fall foul of objective 3. The system was debased by creating the "jump" (for most mortals) in 2009. Any changes should be part of a package to improve the system as a whole to minimise future tinkering failing objective 3

[Kevin Thurlow]

As leagues tend to use the "start of season" list throughout the season, if you do an extra list, you need to maintain access to the "start of season" one, not just delete it from the system... I think when the RP list is published online, the old one disappears?

[Adam Raoof]

As leagues tend to use the "start of season" list throughout the season, if you do an extra list, you need to maintain access to the "start of season" one, not just delete it from the system... I think when the RP list is published online, the old one disappears?

No you can definitely look up the last published grade as well as the new 'current' one.

I don't know anything about the grading system from a mathematical point of view, but I do know that if I asked most players what the ECF means to them, it is the grading system. I suspect that if I asked players what one thing the ECF could do to improve their service, it might be to publish the grading list more often.

What we need to do is establish that as a desirable objective, then make the system work towards that end.