Nice graph, but the probability distribution of finding a block with normalized hashrate (e.g. taking changes in pool size into account) is a poisson distribution (or gamma for that matter) with a low lambda and not gaussian over 1 (if it were, the long gaps would correctly be odd). Your graph simply shows the lower end of the distribution with some outlyers (don't know if this is the English word) like the >700% blocks. Nothing weird here, sample size is simply too low to draw conclusions based on the available data about the tail of the distribution.

A decent mathematician can probably calculate the arrival time of blocks to a certain degree of accuracy but its complicated and takes some effort or detailed knowledge of the systems involved. Check out waiting times, Erlang and those kind of things concerning arrival estimation if you want to know more about it.