Just done a few spot checks on TE current match. Agree it's wrong (again). The real figures calculated from web archive don't show a big difference but they don't match. Makes you wonder who is coding that doesn't check basic arithmatic before releasing code changes.
It seems that Chess960 timeout rates given in endpoints no longer agree with those on webpages! Not sure about Standard.
We had this once before & not long ago but we were assured it had been fixed. Looks like it's broken again?