Menzie has a chart

 

Log normal, which we translate as lob binomial. Each of those colors represent a binomial, and when a transaction is to be had, one of the colors will take it.

In the time series, these are binomial distributions, otherwise they would be Weiner process, I think.

But I digress.  These are growth rates, fractional ratios. They converge at 2020 by Menzie picking a sequence start, correct because of covid. This chart works when we get a black swan, but I would start the sequence about march 2020.

After the requant, the binomials have diverged, they are price discovering. All the binomials are at lower energy levels, having lost most of their coin toss bets in the covid. 

Each of those is a sector, and we presume they are independent, mostly, and are complete in the monetary economy.  hen we can use optimum portfolio theory and distribute your next bet among them with a maximum entropy five headed coin. That causes the binomial to adjust pr of heads to make a balanced binomial and the is the equivalent of an interest charge.

The basic idea of optimum portfolio is to always simulate a monte carlo machine that can reproduce the same process ex post, back to the sequence start. And that is your best guess in allocating the next coin toss. Which is to say, allocate the toss to one of the portfolios where the pr of heads is tilted in your direction, and take a marginal gain on interest charges. That balances the binomial. Ignoring a one of the portfolios leads to a point where it has a hedge and you toss a coin its direction.  There is always relative change in probability, it is adaptive.