Fair value infrastructure for private credit
Answered by construction, not reconstructed after the fact.
Every quarter your auditor asks why a mark moved, how much of the change was rates rather than the market rather than this specific credit, and where each input came from. Aestima is valuation infrastructure built so those answers already exist when the question arrives — every mark produced by a verified computational engine, every input carrying its source on its face, every judgment recorded alongside the alternative it rejected.
- Test instruments
- 20Fully specified, spanning the instrument universe the platform covers.
- Expected values
- 325Every value the answer set asserts — each one computed before the engine existed.
- Mismatches tolerated
- 0Acceptance is all-or-nothing: no tolerance band, no partial pass. A single mismatch stops the system.
The problem
Most valuation processes answer after the fact.
Every quarter, a private credit fund marks its book. Every quarter, the same questions arrive from the auditor: why did this mark move, how much of the change was rates versus the market versus this specific credit, where did this input come from, and what would the number have been under the reading you did not take.
Most valuation processes answer those questions after the fact — reconstructed from spreadsheets, dependent on whoever built the model, and fragile under staff turnover and deadline pressure. Aestima is built so they are answered by construction.
Coverage
The instruments private credit funds actually hold.
Not a generic discounted-cash-flow shell with a credit label on it. Each structure below is modelled on its own mechanics.
- 01Floating-rate term loansBenchmark floors and reset schedules.
- 02PIK and cash/PIK structuresToggle mechanics and accrued balances carried through the mark.
- 03Callable notesYield-to-worst mechanics across the call schedule.
- 04Unitranche facilitiesFirst-out and last-out splits valued separately.
- 05RevolversFunded and unfunded components treated distinctly.
- 06Loans with equity kickersIncluding the day-one reconciliation of package value against price paid.
- 07OID and effective-interest accretionWith the carrying-value-to-fair-value bridge.
- 08Performing to non-accrualThe method switch and the recovery analysis that follows a migration.
At the fund level
- Whole-portfolio measurement
- Every position measured in a single run.
- Quarterly roll-forward
- Opening balance plus originations, less repayments, plus or minus change in marks.
- Back-testing
- Prior marks tested against realized exits.
The attribution
Rates. Market. Credit. Reconciled.
At origination, each position is calibrated to its actual purchase price — the discount margin that ties the model to the transaction. Each quarter after, the mark rolls forward with a sequential attribution: how much of the change came from benchmark rate movement, how much from market spread movement in the asset class, and how much is specific to the credit.
Each step is a full repricing, not a sensitivity — and the steps reconcile to the total. This is the page that answers the auditor’s first question before it is asked.
Exhibit — sequential attribution
Senior secured term loan, floating
- Period
- Q3 → Q4
- Basis
- Points of par
- Method
- Full repricing, each step
Opening
Benchmark rate
Market spread
Credit-specific
Closing
Value axis truncated to 96.00–98.00
| Step | Driver | Change | Fair value |
|---|---|---|---|
| Opening fair value | Calibrated to purchase price at origination | — | 97.42 |
| Benchmark rate | Three-month reference rate, −42 bp over the quarter | (0.38) | 97.04 |
| Market spread | Asset-class discount margin, +35 bp | (0.61) | 96.43 |
| Credit-specific | Net leverage 4.8× → 4.4×; issuer margin +110 bp | 0.24 | 96.67 |
| Closing fair value | Each step a full repricing, not a sensitivity | (0.75) | 96.67 |
Attributed (0.75) · Total change (0.75) · Unexplained 0.00
ReconciledIllustrative. Figures are synthetic and do not reflect any position, fund, or engagement.
Verification
We did not ask you to trust our implementation.
Before the engine was built, we constructed a fixed answer set: twenty fully specified test instruments, with every expected value — 325 in total — computed by multiple independent implementations, built in isolation from one another, working only from the written specification. The engine is accepted only when it reproduces all 325 values exactly. To the cent, to the basis point.
Built in isolation
Independent implementations, working only from the written specification — never from each other.
Disagreement is traced
Where implementations disagreed, the disagreement was traced to its root cause and resolved by an explicit, recorded ruling.
Frozen, then continuous
The answer set was frozen only when every implementation agreed exactly. Verification then runs continuously.
A single mismatch stops the system
Provenance
Every figure traces on its face.
Where a legal document admits more than one reading, or a convention could reasonably be chosen either way, the report records the treatment adopted and the value under the treatment rejected — so a reviewer sees not just the answer but the size of the judgment.
- Documented
- The verbatim source language and the page it came from, carried on the face of the exhibit.
- Assumed
- The stated basis for the assumption, disclosed rather than buried.
- Elected
- An analyst judgment, with its rationale — and the value under the treatment rejected.
The engagement model
You provide the documents. We handle the rest.
- 01
You provide the documents
Credit agreements, schedules, and portfolio data. We handle the rest.
- 02
First quarter establishes the baseline
Each position is calibrated to its actual purchase price — the discount margin that ties the model to the transaction.
- 03
Every quarter after is an update and a review
The attribution, roll-forward, and exhibits are produced by the platform, reviewed by MELD's analysts, and delivered on your reporting timeline.
Drafts are clearly staged for management review before anything is presented for auditor reliance. MELD Valuation has produced over 10,000 valuation reports across private credit, venture, and complex securities. Aestima is that experience, built into infrastructure.
Bring us next quarter’s book.
A thirty-minute call with the people who built it. We will walk your instrument mix and tell you plainly what the first quarter of setup would involve.