Portfolio valuation
Valuation methodologies
Valuation methodologies
Every ABF position needs a valuation methodology. The right approach depends on what market data exists, how bespoke the position is, and how much judgment is required. This page covers the core methodologies and when to use each.
Discounted cash flow (DCF)
DCF is the workhorse methodology for Level 3 ABF positions. You project the cash flows from the underlying assets, run them through the deal waterfall, and discount back to present value.
Building the cash flow model
A DCF valuation for a structured position requires projecting cash flows at the asset level, then applying the deal structure:
Step 1: Start with contractual cash flows
From the underlying collateral pool:
- Scheduled principal payments
- Contractual interest payments
- Any fees that flow through to the position
Step 2: Apply prepayment assumptions
Prepayment rates vary by:
- Asset type (auto loans prepay faster than consumer loans, which prepay faster than mortgages)
- Vintage (seasoned pools typically prepay faster)
- Rate environment (refinancing incentive affects behavior)
- Credit quality (higher-quality borrowers refinance more readily)
Express prepayment as CPR (conditional prepayment rate) or SMM (single monthly mortality).
Step 3: Apply default assumptions
Default rates are expressed as:
- CDR (conditional default rate) - annualized rate of new defaults
- Cumulative loss curve - total expected losses over the life of the pool
Build default assumptions from:
- Historical performance of similar pools
- Current delinquency trends in this specific pool
- Macroeconomic conditions
- Originator track record
Step 4: Apply severity assumptions
Loss given default (LGD) reflects what you lose after recovery efforts:
- Recovery rates by asset type (auto loans recover 40-60% of balance, consumer loans recover 5-15%)
- Recovery timing (how long until recoveries are realized)
- Legal and collection costs
Step 5: Run through the waterfall
Apply the deal structure:
- Payment priority (senior vs. subordinate tranches)
- Triggers and credit enhancement mechanisms
- Reserve account captures and releases
- Excess spread allocation
The output is projected cash flows to your specific position, accounting for how losses are allocated across the capital structure.
Selecting the discount rate
The discount rate has two components:
| Component | Description | How to Determine |
|---|---|---|
| Base rate | Risk-free benchmark matching the asset’s duration | SOFR swap curve, Treasury curve |
| Spread | Credit and illiquidity premium | Market comparables, new issue pricing, transaction data |
Finding the right spread:
| Anchor | How to Use | Strength |
|---|---|---|
| Comparable term ABS spreads | Find public deals with similar collateral, adjust for structural differences | Strong if good comps exist |
| New issue pricing | What spread would this position clear at if issued today | Reflects current market |
| Secondary trading levels | Any trading in similar paper | Direct market evidence |
| Implied spread from purchase price | Work backward from what you paid | Useful as starting point |
Worked example: discount rate selection
You are valuing a BBB-rated tranche of a consumer loan term ABS. Here is how to build the discount rate:
Market observation: Comparable public consumer ABS BBBs are trading at SOFR + 275 bps.
Adjustments:
| Factor | Adjustment | Rationale |
|---|---|---|
| Deal size ($150M vs. $500M+ public deals) | +25 bps | Less liquidity in smaller deals |
| Less seasoned originator | +25 bps | Execution risk, less established track record |
| Total adjustments | +50 bps | |
| Final discount spread | SOFR + 325 bps |
Documentation: Record the comparable you used, the adjustments applied, and the rationale for each adjustment.
Sensitivity analysis
For Level 3 positions, you must stress your assumptions and disclose how the mark changes. A standard sensitivity table shows value across a matrix of assumptions:
| CDR | Discount Rate -50 bps | Base Case | Discount Rate +50 bps |
|---|---|---|---|
| Base - 1% | $10.5M | $10.2M | $9.9M |
| Base Case | $10.0M | $9.7M | $9.4M |
| Base + 1% | $9.5M | $9.2M | $8.9M |
Illustrative pricing. See pricing disclaimer.
This table becomes part of your Level 3 disclosure and helps LPs understand the range of reasonable values.
Market comparables
When you can find comparable transactions, they provide a market-based anchor for your valuation. This is often more defensible than pure DCF because it reflects actual market activity.
Using public ABS spreads as benchmarks
Step 1: Identify comparable deals
Look for public term ABS deals with:
- Similar asset class (auto, consumer, mortgage, equipment)
- Comparable credit quality (rating or credit metrics)
- Similar vintage and seasoning
- Structural similarities
Step 2: Pull trading data
Sources:
- TRACE (for corporate bonds and some ABS)
- Bloomberg runs
- Dealer runs and BWICs
- Pricing service data
Step 3: Adjust for structural differences
| Factor | Typical Adjustment | Direction |
|---|---|---|
| Credit enhancement | 5-15 bps per percentage point | More CE = tighter spread |
| WAL | 2-5 bps per year of WAL | Longer WAL = wider spread |
| Issuer/shelf premium | 10-30 bps | First-time issuer = wider |
| Deal size | 10-25 bps for sub-$200M | Smaller = wider |
| Performance variance | 5-20 bps | Worse performance = wider |
Worked example: comparable adjustment
You own a AA tranche of a $175M auto loan ABS from a fintech originator. Benchmark:
- AA auto ABS from captive finance companies: SOFR + 85 bps
- Deal size: $500M+
- Issuer: Established shelf
Adjustments:
| Factor | Adjustment |
|---|---|
| Size ($175M vs. $500M+) | +15 bps |
| Issuer (fintech vs. captive) | +20 bps |
| Performance (slightly higher losses) | +10 bps |
| Total adjustments | +45 bps |
| Comparable spread | SOFR + 130 bps |
Now run this spread through your DCF model to arrive at the mark.
When comparables work and when they don’t
| Situation | Suitability |
|---|---|
| Rated tranche of term ABS with active shelf | Good - comparables readily available |
| First-loss residual of private warehouse | Poor - no comparable exists |
| Mezzanine tranche of bespoke CLO | Moderate - CLO market comps exist but require significant adjustment |
| Participation in loan pool | Poor - structure is unique |
Dealer quotes
Dealer quotes can provide market-based inputs, but their quality varies substantially. Understanding what kind of quote you have determines how much weight to give it.
Types of quotes
| Quote Type | Description | Use in Valuation |
|---|---|---|
| Executable bid | Dealer will buy at this price if you offer | Strong Level 2 input |
| Indicative bid | Dealer’s estimate of where they would bid | Weaker Level 2 input |
| Color | Dealer’s opinion of fair value | Level 3 input at best |
| BWIC results | Actual executed trade from bid list | Strong Level 2 input |
Interpreting bid-ask spreads
The bid-ask spread tells you about market liquidity and dealer confidence:
| Spread | What It Signals |
|---|---|
| 25-50 bps | Liquid, well-understood structure |
| 50-100 bps | Moderate liquidity, some uncertainty |
| 100-200 bps | Illiquid, dealer taking significant risk |
| 200+ bps | Distressed or highly esoteric |
Illustrative pricing. See pricing disclaimer.
A wide bid-ask spread is information. If a dealer bids 95 and offers 98, they are signaling that the market for this paper is thin. Your mark should reflect that reality, typically at or closer to the bid side for conservative reporting.
Quote staleness
A dealer quote from three months ago is not a current valuation input. If market conditions have changed since the quote was provided, it will misstate value.
| Quote Age | Treatment |
|---|---|
| 0-30 days | Use directly, subject to material changes check |
| 30-60 days | Use with spread adjustment for market moves |
| 60-90 days | Directional only; re-solicit if possible |
| 90+ days | Do not rely on; need fresh market data |
If you cannot get a fresh quote, document why and explain how you adjusted the stale quote for current market conditions.
Third-party pricing services
Pricing services (Bloomberg BVAL, Markit, ICE, Refinitiv) aggregate market data and provide automated valuations.
What they offer
| Feature | Benefit |
|---|---|
| Automated daily or monthly marks | Operational efficiency |
| Methodology documentation | Audit support |
| Independent source | Governance benefits |
| Coverage of liquid positions | Reduces internal modeling burden |
Limitations to understand
| Limitation | Implication |
|---|---|
| Same model, different branding | May be running DCF similar to yours |
| Thin coverage of esoteric ABF | Your positions may not be covered |
| Methodology may miss deal specifics | Structural features may not be captured |
| ”Evaluated prices” are often model-based | Not necessarily market-based |
When to override a pricing service mark
If your internal analysis suggests the pricing service mark is materially wrong, you can override it. Valid reasons to override:
- The service is missing a trigger breach that affects value
- They are ignoring a modification to the waterfall
- They are using stale comparables that no longer reflect current spreads
- They are not accounting for deal-specific structural features
Documentation when overriding:
- State the pricing service mark
- Explain why it is incorrect
- Document your alternative methodology
- Show the adjustment
Do not blindly accept a third-party price that you know is incorrect just because it is from an “independent” source.
Choosing the right methodology
Different positions call for different approaches. Here is a framework:
| Position Type | Primary Methodology | Supporting Methods |
|---|---|---|
| Rated term ABS tranche | Market comparables | DCF for validation, dealer quotes |
| Warehouse residual | DCF | Market color for discount rate |
| First-loss piece | DCF with stress | Historical loss analysis |
| Senior tranche with active trading | Dealer quotes/pricing service | Comparables for validation |
| Participation in loan pool | DCF | None - fully model-based |
Hybrid approach
For many ABF positions, the most defensible methodology is hybrid:
- Anchor to market: Find the closest observable benchmark
- Adjust with model: Use DCF to adjust for structural differences
- Document the bridge: Show how you moved from the market anchor to your final mark
This approach combines market discipline with the flexibility to account for bespoke features.