Potential grading submission mix up

[MartinCarpenter]

I was just amused to check my predicted FIDE rating and have seen something rather odd. I don't actually care at all on my own behalf, but it looks like it could conceivably be affecting other people from Div3(I suspect N?).

There's definitely something gone badly wrong with my interim grade (http://ratings.fide.com/card.phtml?event=439541) - I've only played 9 games but I've apparently got 13 games counting in my grade. My opponents list is right (and 9 strong) but I've only played them all the once.

Perhaps this has been caused by something to do with the 3N/3 switch over? I've got two listings from 2014 which seem to overlap rather: one is 4NCL 2013-14 Performances for part results (ENG) from October 2014, the other is rounds 9-11 from June 2014.

It looks like the part performances entry is basically in error and causing the extra games? Could easily be some random FIDE mix up of course Mentioning it as it might potentially be something wrong with how some of them got submitted so affecting other people?

If it is just random FIDE stuff then I really don't care. In fact its actually making my projected grade a considerably more accurate reflection of my strength than my 4NCL performances so far (I've been clustering good results.).

I am informed that its a vaguely known sort of random FIDE mix up/incompatibility thing. No known solution except to accept that it makes these new grades even less reliable than otherwise

[Stewart Reuben]

Martin. I'm confused I can only find 7 games for you. Bennett, Latham. Lambourne, Robertson, Fogg, Adams, Atkinson.
Where are the other two? We might be able to get this correct before the 1 December list. And, as you say, it may be a systematic error.
2014 average field in 7 games. + 4. So 4 x 20 = 80.
Rating so far 2094.

[MartinCarpenter]

From what Andrew Bak has said on the Yorkshire Chess forums it is a systemic error quite possibly affecting basically everyone who got a new grade via Div 3 this year Richard Allis and Chris Bak definitely affected too, could well be others. Could be localised to 3N, I wouldn't know.

Some sort of odd mix up to do with not being able to accept chunks of games of less than 3 in size towards partial grades, so all the normal weekends were getting bundled together into a lump sum but then the final weekend - which is 3 matches in a row - got submitted too and those ones have ended up double counted Seemingly known about but whether there's a solution isn't.

Not clever of course but not that important with the FIDE grades at the level of Div3 being fundamentally so random as to be worthless

The two missing games are Robert Dean and John F Lambert. They might very well not have had final FIDE grades when I played them. My full opponents list gives all of them, together with the count of 13 games which should be 9. 9 games in the estimated grade, goodness knows which ones. It might well be that it should only have 7 in it right now.

[Sean Hewitt]

The problem is with FIDEs attempts to process part ratings for unrated players in the 4NCL last season. For some reason, they included some rounds from the previous season too, which meant that for some players those games counted when they should not have whilst for others they were counted twice. There's no submission mix up.

I have been working with FIDE to unravel this, but it's not trivial due to a number of knock on consequences. I will however get it sorted for all players concerned.

[MartinCarpenter]

Thanks I did suspect it was at least as likely to be at FIDE's end as anywhere else.

[John Upham]

Who is the current ECF IRO?

[Sean Hewitt]

Thanks I did suspect it was at least as likely to be at FIDE's end as anywhere else.

Martin - I think that this is now fixed. Please let me know if you spot anything else. Sean

[Mike Truran]

Sean

Many thanks. It's appreciated.

Mike

[MartinCarpenter]

Definitely appreciated and thank you for the effort! Having just checked, it looks better but I'm not sure if it is fully fixed I'm fact I'm now thoroughly confused.

I'm hoping I've caught them halfway through fixing it I'll try to remember to check back in a week or two.

The good thing is that I'm now only down as having played 11 games at least, which is closer to 9

Still one duplicated game - vs Robertson - and Chris Bak still has one duplicated too. Both our opponents have different FIDE grades in each 'version' of the game though, so maybe they can't detect it as duplicated due to that?! Richard Allis' set of games still looks a bit of mess.

The scores for my games are now flat odd. I should be +4,=1 with white (+2,=1 rated) and +2,=2 with black and they've somehow got me +3,=3 with white and +4,=1 with black. My only visibly still duplicated game is a draw with black vs Robertson, which obviously doesn't help to explain this. The game list has the colours right, so I can only imagine the website summary has colour flipped my results for some bizzare reason.
(Then one black draw and one win being double counted or something.).

[Sean Hewitt]

Definitely appreciated and thank you for the effort! Having just checked, it looks better but I'm not sure if it is fully fixed I'm fact I'm now thoroughly confused.

I'm hoping I've caught them halfway through fixing it I'll try to remember to check back in a week or two.

The good thing is that I'm now only down as having played 11 games at least, which is closer to 9

Still one duplicated game - vs Robertson - and Chris Bak still has one duplicated too. Both our opponents have different FIDE grades in each 'version' of the game though, so maybe they can't detect it as duplicated due to that?! Richard Allis' set of games still looks a bit of mess.

The scores for my games are now flat odd. I should be +4,=1 with white (+2,=1 rated) and +2,=2 with black and they've somehow got me +3,=3 with white and +4,=1 with black. My only visibly still duplicated game is a draw with black vs Robertson, which obviously doesn't help to explain this. The game list has the colours right, so I can only imagine the website summary has colour flipped my results for some bizzare reason.
(Then one black draw and one win being double counted or something.).

Martin - FIDE are in the process of sorting the structure of the games counted which I expect should be sorted this weekend. As far as I can see, you'll end up with 7 games counting towards your part rating.

[NickFaulks]

Sean

Many thanks. It's appreciated.

Mike

The Elista office has put in a lot of effort to sort this out. A word of thanks to them would not be out of place.

[Mike Truran]

Wasn't it FIDE's mistake in the first place? Or am I misunderstanding Sean's 27 November post?

[David Sedgwick]

The Elista office has put in a lot of effort to sort this out. A word of thanks to them would not be out of place.

Wasn't it FIDE's mistake in the first place? Or am I misunderstanding Sean's 27 November post?

My take on this was that the whole issue arose because the effects of the changes to the FIDE Rating Regulations in 2013 were not fully considered. In that sense it was indeed FIDE's mistake.

That left the poor b..... infantry in Elista in the firing line.

Thanks to them for trying to cope with the mess. Thanks also to Sean and anyone else at this end who helped.

[MartinCarpenter]

I'd guess so. The 4NCL was having to be special cased after all.

That just makes mix ups - which are basically inevitable when you're trying to do worldwide grading on the sort of scale they're dealing with! - even more likely. It must be a horrible job running the thing and I've got nothing but sympathy for all concerned.

[Roger de Coverly]

My take on this was that the whole issue arose because the effects of the changes to the FIDE Rating Regulations in 2013 were not fully considered. In that sense it was indeed FIDE's mistake.

The specific problem was how to cope with leagues played in instalments like the 4NCL and the interaction of players without ratings against rules requiring a minimum number of games to qualify for a rating. Previously, even with bi-monthly and monthly ratings, the 4NCL had submitted a bulk file at the end of the season. This copes with players gaining part ratings or ratings for the first time in the same way as would apply to an eleven round tournament.

FIDE then decreed that instalment leagues should submit results as they happened. This created a problem because they appeared to leave in place a rule that a minimum of three games was needed for a part rating to qualify. The question as to why it was that FIDE were attempting to prevent players gaining ratings through national leagues was raised at the time. I'm not sure it ever got a satisfactory answer. For the future I believe that the rule of three has been scrapped as more trouble than it's worth. That still leaves previous seasons to be sorted out. It appears to now be the intent that all 4NCL games against rated players should count towards a part or full rating, but implementation seems to be a little problematic.