I found a suitable solution, for the long term average / standard deviation https://en.wikipedia.org/wiki/Algorithms_for_calculating_variance#Online_alg... I will make sure it is adjusting after a few weeks like initially planned. I will propose another PR shortly
test with random samples from 1-500ms ``` 1 online: total[2143000000]value[35] q[4.464600e+13] count[2143000001] std[144.337830] avg[250.493088] est[263.434906] 1 online: total[2144000000]value[85] q[4.466685e+13] count[2144000001] std[144.337875] avg[250.493088] est[205.304733] 1 online: total[2145000000]value[375] q[4.468767e+13] count[2145000001] std[144.337830] avg[250.493088] est[237.945862] 1 online: total[2146000000]value[277] q[4.470850e+13] count[2146000001] std[144.337845] avg[250.493088] est[311.532623] 1 online: total[2147000000]value[373] q[4.472931e+13] count[2147000001] std[144.337799] avg[250.493088] est[251.274414] ```