Very unbalanced movements and the NGS
Suppose a club plays a movement (such as an incorrectly arrow-switched truncated Howell) in which there are extreme disparities in the level of competition between pairs: some pairs of pairs sit in opposite directions every round and never meet at the table (effectively making them allies – a good result for one will always benefit the results of the other), whereas other pairs of pairs sit in the same direction every round except when they meet at the table (adding an extreme amount of competition – a good result for one will always hurt the results of the other, especially in the round where they face each other, and no good result for one can ever help the result of the other). This means that if a pair is particularly lucky or unlucky in who their randomly chosen allies and enemies are, they'll be facing a very different effective strength of field from that suggested by the pairs who participate in the session.
It strikes me that the NGS should be taking into account the levels of competition between the various pairs in the actual movement played, when calculating the par score. (I think the correct formula for calculating strength of field is a weighted average between the NGSes of the other pairs in the event, where the weights are based on the levels of competition between the pairs. In particular, in the case where two pairs sit in opposite directions more often than they sit in the same direction, and never meet at the table, the weight will be negative, because having a strong "opponent" in that position will actually help rather than hurt your results.)
Based on an extremely unbalanced session I just observed, the NGS doesn't seem to be making this sort of correction at the moment (the NGS-calculated strength of opposition for a pair was similar to the strength of field for the event as a whole, despite their super-opponents being much stronger than their allies). Is there enough information in the files uploaded to the EBU to determine how many rounds two pairs spent as table opponent / indirect opponent / ally? If not, should there be? If there is, should we be using it to give more accurate NGS corrections?
(This probably wasn't a major issue prior to the rise of online play – most clubs would be unlikely to run movements that are much more unbalanced than a truncated Mitchell or truncated Howell – but BBO is hugely excerbating it, because its usual choice of movement when there are more rounds than tables is arrow-switched incorrectly and introduces this sort of extreme disparity.)
Comments
The NGS Full Guide explains how SOpp is calculated and you are correct that it doesn't go to the lengths you suggest. I suspect it would be impractical to do so, especially given the limited effect each game has on your overall grade.
As Gordon says.
I would add that I don't play on BBO so I am not at all familiar with its movements, but if the movements for small numbers of tables is as flawed as you suggest, then either
-- you could see things like swings and roundabouts, in that, over a long term, the luck of where you happened to start in the movements will even out, or else
-- you could persuade BBO or whoever to change the relevant movements to something better. After all, it's not just National Grades that are subject to the whims of an imperfect movement but also the actual ranking list for the sessions themselves. But are the movements really that bad?
Barrie Partridge - CTD for Bridge Club Live
There's a competition table for the "BBO Howell" here. (For a balanced 1-winner movement, all the numbers should be the same; for a balanced 2-winner movement, the grid should be divided into four quadrants, with two quadrants having all the numbers the same and the other two being zero.) You have competition numbers spread pretty evenly over the range from 22 to -6 when they're all supposed to be 8, which is a very long way away from being balanced.
I think people have complained about the movements to BBO already and been told that there's no plan to fix them. I can try to add my own complaint if there's a reasonable chance it will help. (Probably I should actually work out the correct movements myself and send them to BBO, to save them having to work out the movements themselves.)