0
$\begingroup$

I have a dataset containing observations at time t and t+1 of ratings A (best) to E (worst) and Default. I want to use transition matrices to predict future ratings t+2, t+3, etc

The transition matrix of the full portfolio is monotonic (transition to a further state is less probable than a transition to a nearer state). However, some sub-portfolios are quite small and therefore not monotonic. This is only due to their size and not a pattern of behavior. All ratings are theoretically possible. Default is an absorbing state. How can I smooth the matrices? Preferably with an R package as I'm not a mathematician :-)

enter image description here

Improve this question
$\endgroup$
3
  • $\begingroup$ Please clarify your specific problem or provide additional details to highlight exactly what you need. As it's currently written, it's hard to tell exactly what you're asking. $\endgroup$
    – Community Bot
    Commented Nov 2, 2023 at 11:21
  • $\begingroup$ Can you clarify what smoothing you want to do, and relationship between subportfolios $\endgroup$
    – seanv507
    Commented Nov 2, 2023 at 11:24
  • $\begingroup$ You could use regularised multinomial regression as provided by ʼglmnetʼ package. This would smooth subportfolios to the total portfolio transition matrix $\endgroup$
    – seanv507
    Commented Nov 2, 2023 at 11:28

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.