i have tried to generalise a bit the ideas of @Geertjancap . In my derivation beta will be a function and as such will have a distribution (rather than be a constant) cc @100trillionUSD https://twitter.com/phraudsta/status/1194148418286800896

this gives us a "model of demand" which we can view as a graph over time

You can ignore the apparent drop of for the last month (which is november) as it isnt complete yet. I will undertake further examination of this to test for trends and stationarity.

adf test for stationarity rejects unit root on demand, indicating that demand (in this derivation) is stationary.

Now we estimate demand as a function of log time. this gives us a starting point of .68*log(time)-0.7

i will write an article to expand further