Histograms (that at infinity should form a Probability Density Function - but we have have only ~60-70 points) show an interesting behavior at 0.3-0.5 and 0.6-0.8 (and specially 0.82-0.83).
Priviet!
I did something like this long time ago (
https://bitcointalk.org/index.php?topic=1306983.msg54879435#msg54879435)
BUT: it makes no sense!
Somewhere (I do not have time to search for the post) author stated that he generated random private keys and then replaced the beginning with ZEROs, just to keep incrementing ranges. So any statistical approach is useless IMHO.
Sure thing - i've also noted that on github (making some dummy simulation showing histogram for the first 60s points and last 100 point of the group), but again, currently we have no other criterion for reducing\choosing intervals.
We could try to hack the Random Generator used to generate the sequence - that should be a "Algorithmic" random generator i assume - but that's probably even harder than 32 BTC puzzle task.
(should remember that there is no strong proof that inverse problem for both Random Generator or Private Key -> Public Key -> Address have no algorithmic solution, who knows, may be some smart, well motivated person with enough free time will find\construct an algorithm of reverse "de-hashing".)