I think your math is correct. Though probability has no memory, the chance of having 2 of 1 in 2000 so closely (within 4) is 1 out of million.
The question is with increased hash rate this pool gets more bad luck and is supposed to be less variance.
It will be great if we can find the issue. Otherwise we can just blame bad luck.
I remember some years ago when slush pool had a several month long streak with really bad luck, but as i remember it they found out it was some bug cousing it
.
This is maybe just bad luck but how big is the chance of getting 2 devilblocks so close to each other ? Or should I see it as every new block is a new block and what happened before doesnt matter at all ?
The first block was 1 in 2000
The second block is About 1 in 2000
But they were not back to back which would be a 1 in 4,000,000 chance.
My math skills to determine. The chance of two .9996 cdf blocks out of four are not up to the task.