[gnasher]

I want to be able to calculate cross-IMPs that are adjusted for the strength of opponents. That is, I want to calculate raw cross-IMPs, then adjust each pair's score using the cross-IMP scores of their opponents (and possibly do this recursively if necessary).

I'm also lazy, so I'm hoping that someone already has a spreadsheet or some code that they don't mind sharing with me. Does anyone have such a thing?

Also, is there anything wrong with this simple approach:
(1) Calculate cross-IMPs for each pair on each board.
(2) Calculate average cross-IMPs per board for each pair.
(3) For each board, add the opponent's value of (2) to (1), to give an adjusted cross-IMP figure.
(4) Repeat until the numbers stop changing.

[paulg]

I have looked at doing this and will send you a spreadsheet if I still have one in usable form.

One thing you should do when going through this process is not to double count the cross-imps of the opponents. That is, the cross-imps of the match between a pair and their opponents is included in both their final cross-imps, so you should calculate the opponent's cross-imps excluding the match with the target pair to get their strength against everyone else. Hope that makes sense!

[PhilKing]

I just did this manually for Beijing, for instance, weighting each pair's performance according to strength of opponent. Obviously that is not perfect, since various opposition may have been playing predominately weak/strong pairs, but it is much better than nothing, is quick and requires no programming.

Now you are a selector, you should be able to access the relevant document. I guess for judging pairs in the Premier League, you need something more sophisticated though.