Model architecture.
Every InstitutionalModels™ engagement is built to a consistent standard: separating data from logic, and logic from outputs — so that assumptions are transparent, calculations are auditable, and results hold up under scrutiny.
Separation of concerns.
Institutional models fail when data, logic, and outputs are entangled. Assumptions become invisible, calculations become impossible to audit, and small edits cascade unpredictably. We architect every engagement in three distinct layers, connected only in one direction.
Before development begins, we work with the client team to align on data sources, key assumptions, decision points, and the audiences the model is meant to serve. This upfront work is what enables the discipline of the three-layer architecture to hold up over time.
Houses all inputs and assumptions — timing, capital structure, fees, exit scenarios. Third-party sources (rent rolls, ARGUS exports, operating statements) are imported separately, keeping source data distinct from assumptions so that both can be audited independently.
Contains supporting schedules and a core cash flow tab designed for readability — inline formulas, no hidden dependencies, no cross-sheet tracing. Every calculation reconciles to a source input, and every intermediate result can be reviewed on its own.
The presentation layer draws from calculations without feeding back. This separation enables customization for specific audiences — executive summaries, investment committee memos, LP reports, board decks — without disturbing the underlying model.
Independent error-catching and parameter verification runs alongside the model — so assumptions can change without breaking reconciliation.
Documentation is built throughout development, not after — explaining both mechanics and reasoning so the model remains usable across analyst turnover.
Disciplined architecture is what enables institutional-grade credibility and long-run flexibility. A model that stays honest as it evolves.