Well it is actually just that run of 19 blocks + 1 orphan that killed the stats.
(and the fact that the orphan was so out of the pools control it was ridiculous - caused by 3 separate external events at the same time)
The 20 starting from last block in November 630.819% and finishing with an orphan + 178.228%, and finally that worst record 744.977%
If we just pretend those 20 blocks never happened it all looks great
Those 20 are now just outside the last 50 blocks, so all the stats for 50 blocks or less look ok again and the very long term stats are good enough to cover that run of bad luck.
I am curios to learn whether the following had any relation to late November-December bad luck?
https://bitcointalksearch.org/topic/m.16605935
and the discussion between Matt and kano that follows
this change seems to correlate precisely in time with pool's performance in that period.
is this just a coincidence?
I am cool with current performance, but would you care to comment, please?