[Inefficient plans for correlated columns] has been a pain point for quite a while. While we've had several discussions in the area, it always seems to just kinda trail off and eventually vanish every time it comes up.
[...]
It's a very hard problem. There's no way you can keep statistics about all possible correlations since the number of possibilities is O(N^2) with the number of columns.
Actually far worse: N!/(N - K)!K! summed over K=1...N, assuming the order of columns in the correlation is unimportant (otherwise it is N factorial) - based on my hazy recollection of the relevant maths...
Right ... I was only thinking of combinations for two columns.