Valuation and Carry: Where Definitions Diverge

Almost no dispute over valuation or carried interest starts with a wrong calculation. It starts with a place the agreement is silent. Under one and the same set of partnership documents, “does unrealised gain count toward distributable proceeds”, “does the hurdle accrue simply or compounded”, and “how are catch-up and hurdle satisfied at the same time” each admit several defensible readings. A calculation engine merely executes whichever reading was chosen. The later the disagreement surfaces, the more historical periods must be restated and the more expensive the negotiation becomes. The only useful engineering act is to enumerate those silences before implementation, have finance, legal and the investment committee decide each one on the spot, and store the decision as a versioned parameter.

Four default assumptions behind valuation divergence

Default assumption	How it diverges	Recommended definition
Valuation date	Statement date, closing date or disclosure date	Fix one reference date shared by all metrics; store capture time separately
Source hierarchy	Free switching between last transaction price, comparables, net assets and cost	Rank by observability, and record which tier was used in each valuation entry
Discounts and premia	Whether liquidity, control or minority adjustments apply, and by how much	Explicit parameters with a cited basis (appraisal report or agreement clause); never implicit in the result
Fair-value tier	Inconsistent tier labels; one deal sits in different tiers in two frameworks	Adopt the disclosure tier definition and map it one-to-one to the supervisory reporting definition

The third line is the costly one. Once a discount is buried inside a model output, it cannot be recomputed or explained to an auditor. Make it a field instead: adjustment type, source document, effective interval, approver — bound to the valuation record.

Nine silences in carried-interest terms

• Hurdle accrual: simple or compounded, and whether it pro-rates by actual days of capital at risk.
• Calculation base: contributed capital, committed capital, or the net after management fees and reserves.
• Catch-up: whether the manager catches up in one step to the full sharing ratio, and in what order relative to the hurdle.
• Unrealised gains: included in distributable proceeds or not, and if included, what holdback or clawback applies.
• Crystallisation: annual, per realisation, per distribution event, or at liquidation.
• Loss netting: whether one deal’s loss offsets gains elsewhere, and whether it carries across years.
• Management fees: whether they enter the hurdle base, and any carry-forward or offset arrangement.
• Currency translation: which date’s rate applies to multi-currency contributions and realisations, and whether translation differences sit in return or in profit and loss.
• Distributable amount and deductions: how tax, reserve and outstanding expenses are netted before allocation.

Every calculation must lead back to a clause and a document

We hold to three implementation constraints. First, definitions are expressed only as versioned parameters: each set has a start and end date, and a historical period is recomputed with the version effective at that time — today’s parameters never rewrite the past. Second, any allocation result drills down to its inputs: contribution and repayment records, the clause relied on, intermediate steps, and the supporting document behind each adjustment. Third, changes pass through approval: editing a hurdle parameter records the approver, the source document and the effective date, and the impact is explained to investors at the next reconciliation.

These three constraints turn “a calculation” into “a defensible calculation”. The difference shows at the reconciliation table. When an investor challenges a period’s distribution, you locate in minutes which definitional choice produced it, rather than answering “this is how we have always done it”.

Contested item	Common practice	Our definition	Failure signal
Unrealised gains	Verbal convention, decided period by period	Fixed as a clause parameter; changing it reruns affected periods	Included in one period, excluded in the next; the NAV curve explains nothing
Historical restatement	Recompute past with current parameters	Period snapshots retained; recompute with the version then in force	Unable to reconstruct the basis when audited
Multi-currency	One closing rate for everything	Translation date as a parameter; differences reported separately	FX movement read as investment performance
Manual adjustments	Typed straight into the result sheet	Adjustments as coded entries with cause and evidence	Adjustment volume grows each year with no traceable origin

Hand this list to finance and legal before build

• Walk the nine silences one by one, recording your chosen definition and where the clause sits in the documents.
• List unresolved items with a named decision owner and a deadline. Do not let an implementer “set a sensible default”.
• Define the valuation source hierarchy and the conditions each tier may be used under, including what happens when a recent transaction price goes stale.
• Create a parameter version table for adjustments, hurdle, catch-up and translation dates, each with effective date and approver.
• Before reconciling with investors, run one full period end to end and keep the intermediate steps as a response pack.
• Agree an annual definition review, and whether a change restates prior disclosures.

When we advise against automating the split

Three cases. The terms are still under negotiation or admit multiple readings: automation then delivers one unauthorised answer and disguises a commercial disagreement as a technical conclusion — far more expensive to unwind than to settle in the documents. The fund has no contributions or distribution events yet: an algorithm without historical samples cannot be accepted, only demoed. And the organisation has no role that can relate the clause to the parameter: handing results to an unreviewable calculator carries more risk than a spreadsheet with two-person review.

Reliability in valuation and carried interest does not come from formula sophistication. It comes from every period remaining explainable. Converting contractual silences into versioned parameters, with an approver and an effective date, is the only way to make the numbers hold. Once that is done, the software is the easiest part. Definition work is described in data governance and metrics; the closed post-investment loop for market managers appears in market PE post-investment, and multi-currency, inter-generational disclosure views in family office and wealth management.