A Staged Approach to Pricing Non-Agency MBS and ABS

Summary

The document addresses how to begin modeling non-agency mortgage-backed and asset-backed securities and comparing modeled prices with vendor marks. Its proposed starting point is deliberately simple: assume basic conditional prepayment and default rates, then construct cash flows for straightforward pass-through securities. This builds a baseline cash-flow and pricing model before adding more detailed product behavior.

After prices align with vendor data, the answer suggests introducing more realistic prepayment assumptions and handling more complex security structures. It offers a modeling sequence rather than software recommendations, implementation instructions, or a worked numerical example. The response emphasizes that cash-flow modeling becomes complicated quickly, so the proposed simplifications are an initial calibration step, not a complete valuation framework. Vendor agreement is suggested as a checkpoint, but no validation results or limitations of vendor prices are discussed.

Key ideas

• Begin with simple prepayment and default assumptions for non-agency MBS or ABS.
• Build cash flows for basic pass-through securities as an initial modeling case.
• Compare modeled prices with vendor data before adding complexity.
• Add more detailed prepayment behavior and security structures incrementally.
• The answer gives a workflow but no tool recommendations or worked example.

Tags

# Non agency (R/C)MBS & ABS pricing available tools & instructions


# Non agency (R/C)MBS & ABS pricing available tools & instructions












What tools and instructions can I use to model non agency MBS/ABS products, I would like to be able to price those securities, and compare my result against data vendors (markit, Reuters, etc). Ideally in python but any other options are welcome. Also I can use a Bloomberg terminal, but I don't know how to model cashflows there.

Are there any useful tools with instructions?

## Answer by Ayush Gupta (score 1)

https://quant.stackexchange.com/a/83637

It's not really straightforward. You might want to begin by assuming very simplistic CPR/CDR assumptions, and then come up with cashflows for simple securities (think passthroughs).

Once your prices tie back to the data vendors, you can start making it complicated (think how you might model prepays etc.), and on the cashflow side - how can you actually modify the security structures. It gets pretty complicated very quickly.

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.