Fixed Income Structured

This breaks down structured fixed income products like MBS, ABS, and CLOs with a focus on prepayment modeling and tranching mechanics. It covers PSA prepayment speeds, the CPR to SMM conversion, weighted average life calculations, and how CMO tranches redistribute cash flows. The skill walks through waterfall structures and subordination levels with concrete examples, like calculating that a CLO's AAA tranche with 35% subordination can absorb significant losses before taking principal hits. The key insight here is the treatment of negative convexity in MBS, explaining why you can't use standard duration metrics when cash flows change with rates. Useful when you need to model prepayment risk or understand how securitization redistributes credit and interest rate exposure across tranches.

skills.sh installs: 701
GitHub stars: 224
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add joellewis/finance_skills --skill fixed-income-structured --agent claude-code

Installs into .claude/skills of the current project.

Files

Fixed Income — Structured Products

Core Concepts

MBS Pass-Throughs

A pool of mortgages whose cash flows (principal, interest, prepayments) are passed through to investors on a pro-rata basis. Agency MBS (Ginnie Mae, Fannie Mae, Freddie Mac) carry a government or GSE guarantee against credit losses, isolating prepayment risk as the primary concern. Non-agency MBS lack this guarantee and carry both credit and prepayment risk.

Prepayment Risk

Borrowers can refinance when rates drop, returning principal early. This creates negative convexity — when rates fall, MBS prices rise less than comparable Treasuries because prepayments accelerate and shorten the bond's effective life. Prepayment risk has two faces:

Contraction risk: Rates fall, prepayments accelerate, duration shortens. Investors receive principal back when reinvestment rates are lower.

Extension risk: Rates rise, prepayments slow, duration extends. Investors are locked into below-market coupons for longer than expected.