Should you model your fund forward or check the one you have

Two different instruments

A construction model projects forward. You give it assumptions, entry valuations, graduation rates, dilution, exit multiples, and it tells you what the fund produces if those assumptions hold. Change an assumption and the output moves. Tactyc is the best-known instrument of this kind, and Carta's fund forecasting does a version of the same work.

A construction check compares the declared plan’s parts: fund size, company count, ownership, stage, scope and reserves. Those targets contain assumptions, but the diagnostic does not project a future return. It asks where the plan fits together and where it needs a closer look.

These are not competing answers to one question. They are answers to two questions that happen to take the same inputs.

What each is good for

A model is the right instrument when the question is conditional. What happens to this fund if graduation rates halve. How much does a slower exit environment cost. What does the return profile look like if we write twenty-five cheques instead of forty. Those questions require a projection, and a check cannot produce one.

A check is the right instrument when the question is about the configuration itself. Whether the cheque this fund can write is a cheque that buys the ownership it targets at the stage it enters. Whether the team can serve the board seats the portfolio implies. Those questions have answers that do not depend on any forecast, and running them through a model would mean introducing assumptions the question did not need.

There is a timing difference too. A model is most valuable while a fund is being designed, when assumptions are still being chosen. A check keeps working after the fund has closed, because a declared configuration is still a declared configuration in year four, and it can still be checked against what the firm has become.

Why they are complementary rather than substitutes

The clearest way to see it: a model can be internally consistent and still describe a fund that cannot be executed. Assumptions can be individually reasonable and jointly impossible, and a projection will run happily on them, because a model's job is to compute the consequence of assumptions rather than to audit them.

A plan can fit the diagnostic’s assumptions and still produce disappointing returns. Company selection, entry terms and later events remain separate questions. The score does not establish an investment case.

Using both approaches can make a review more useful: test the design, then examine how outcomes change under different assumptions. Neither replaces the team’s judgment.

Where the Colibrí Architecture model sits

Architecture’s core scores diagnose fund and firm design; they do not project fund returns. A companion Scenario Engine supports follow-on decisions for existing positions with model-generated allocation, timing and rationale. That decision support is separate from a full fund-return projection and leaves the investment decision with the GP.

It also reads two things a construction model generally does not, because they are not fund-level questions. Firm Design Congruence reads the structural choices around the fund, including how the investment committee votes and whether the team can carry its own board commitments. Cross-Fund Concentration reads across every fund the firm runs at once. The congruence checks use the configuration recorded for each fund, while concentration brings investments across funds into a firm-level view.

Sources

• Tactyc Content CenterTactycThe construction-modelling category this page distinguishes itself from, in its own words.
• Portfolio construction: How to develop a fund strategyCartaA representative statement of what a construction plan contains.