Run term debt models

Use a Markov 4 tuple?  We get a time counter?

No, just segment your term length, and select an X= number of terms, generally four terms is about as good as you can do. IUn real time print out the target, -iLog(i) for all the terms. You will make it so with interest swaps per term segment, like temperature.

Run a time counter matched to the terms and get a flat index to accumulate probable interest costs over the term. All this stuff, all Bayes, just counting the various ways to match segmented sets of things..

Most pits are not accurate enough for term debt. Not a problem, make local short term debt with local collateral. P{its have an N problem, there is lots of room for creativity.

Why does my term model work?

it only has three segments, and the their counts. The segments with the -iLog(i) posted will always show the smallest rate due to the most trades.  Short term deposits and loans will always see that as the lowest price. The value chain should self organize among two other queues in short order.