Assets, Data, Structure Underwriting
Underwrite the asset, validate it with data, then structure downside protection
- Difficulty
- Advanced
- Time to result
- ~months to results
- Steps
- 6
- Confidence
- 99%
The framework sequences underwriting around three components: assets, data, and structure. It begins by determining which underlying assets actually drive company or loan performance, rather than relying on a broad growth story. Next, investors ingest historical records and match statistically significant characteristics to estimate how comparable assets behaved across different economic states. This does not eliminate uncertainty, but it grounds assumptions and supports more accurate credit pricing. Finally, the transaction is structured to absorb or limit the modeled downside through collateral, seniority, covenants, financing terms, hedges, or a blend of equity and credit. Each component constrains the next: weak assets cannot be rescued by clever analysis, data must be relevant to the assets, and structure must address the risks the analysis reveals.
Origin
Extracted from Masters in Business, where David Snyderman summarized Magnetar's approach to evaluating specialty-finance opportunities such as CoreWeave and bank risk-transfer portfolios.
Core principles
- 01Start with the asset rather than the company narrative
- 02Use historical data instead of pretending to predict the future
- 03Test performance across different economic states
- 04Price the actual credit risk before selecting a hedge
- 05Use transaction structure to protect against downside
How to run it
- 1
Begin with the assets
Identify the assets producing cash flow and determine how directly they drive the company's performance. Evaluate their durability, utilization, collateral value, and repayment mechanism.
Pro tip Translate the company story into a list of observable asset-level drivers.
Watch out A strong management narrative cannot compensate for economically weak underlying assets.
- 2
Acquire relevant data
Collect granular historical information on those assets and comparable transactions. Check coverage, consistency, and selection bias before analysis.
Pro tip Build ingestion infrastructure that can handle the provider's native data rather than requiring spreadsheet-sized samples.
Watch out Large data volume does not guarantee relevant or reliable evidence.
- 3
Match significant characteristics
Identify features that historically explain performance and match the proposed assets with comparable observations. Use out-of-sample history where possible.
Pro tip Reserve part of the historical data for validation rather than fitting every observation.
Watch out Spurious characteristics can create false precision.
- 4
Model economic states
Estimate how the assets respond to changes in rates, defaults, demand, utilization, and other relevant conditions. Focus on ranges and scenarios rather than one forecast.
Pro tip Include states that have not occurred recently but remain economically plausible.
Watch out A base-case forecast alone obscures nonlinear downside.
- 5
Price the risk
Convert expected losses, uncertainty, liquidity, and capital requirements into a required return. Compare that return with alternative investments.
Pro tip Separate compensation for credit risk from compensation for complexity or illiquidity.
Watch out Mispriced tail risk can overwhelm years of attractive carry.
- 6
Protect through structure
Use collateral, covenants, priority, hedges, amortization, or mixed equity-credit financing to protect against identified downside. Retest returns after adding those protections.
Pro tip Tie each structural term to a specific modeled failure mode.
Watch out Complex structure is useful only when it changes loss behavior.
In the wild
A bank presents a portfolio containing up to ten thousand loans for a regulatory-capital transaction. The investor compares the loans' characteristics with data from hundreds of prior transactions, tests their historical behavior out of sample, prices the transferred credit risk, and then designs hedges for the remaining macro exposure.
→ The investor can accept bank credit risk while the bank retains its customer relationships.
Magnetar evaluated CoreWeave by starting with its high-performance-compute assets, using operational understanding and data to assess their behavior, and applying financing structures around them. The relationship eventually included both a Series B equity round and a large asset financing.
→ The structure supported rapid company growth while giving the investor both equity and credit exposure.
Common mistakes
Starting with the growth narrative
Exciting markets and management stories can distract investors from the assets that must ultimately produce the return.
Using data without comparability
Historical records are misleading when their characteristics do not match the proposed assets or current transaction.
Adding ornamental complexity
Structure that is not tied to an identified downside scenario adds legal complexity without meaningful protection.
Is it for you?
Best for
It is best for specialty-finance, asset-backed, regulatory-capital, and hybrid equity-credit opportunities with accessible operating data.
Not ideal for
It is not ideal for businesses whose value is mostly speculative, intangible, or unsupported by relevant historical data.
From the transcript
“It is. It's very specific, but we always start with the assets. So it's assets, it's data, and its structure.”
“We can't predict the future, so we're trying to do is use historical data to predict how an asset reacts in different states of the…”
“and finally we use structure around that to protect the downside of the investment itself.”
From the episode
David Snyderman on Specialty Finance and Data in Investing
David Snyderman