I have been looking for an approach to model the "overall financial system" to try and predict the probability of market turning points.
I came across the following. It is an open source model created by the New York federal reserve. Its quite recent code from 2017.
https://github.com/FRBNY-DSGE/DSGE.jl
From my understanding, which may be incorrect, the model "fits" about 13 different time series (GDP, Labor, investment, interest rate etc)
using about 50 parameters and then does projections forward in time. The entities used are household, firm, bank, investor, government.
I was hoping for something with more asset classes (stock sectors, bond sectors, real-estate ) that could predict bubbles
but that appears to be far beyond the state of the art.
This code runs in a python like language called julia. It ran for a week on my laptop (making progress, not hung)
and then stopped, but something went wrong since the next step won't complete. I'm surprised it takes so long to run, there is not
that much data.
Reading the history of these models is not encouraging, each time something "new" happens, they need to revise the models...
Any suggestions on how to extend this type of approach would be welcome.