Multifamily spreadsheets share a load-bearing simplification: the average unit. Take the T12, divide by unit count, grow it by a rent growth assumption, and the model runs. It runs quickly, it looks clean, and it prices the property as if every unit were the same unit, which is precisely the assumption the business plan usually contradicts.
Consider what averaging erases. Unit mix, first: studios, one-beds, and two-beds carry different rents, different demand, and different reposition economics, and a plan that renovates two-beds first is invisible to a model that only knows the blended rent. Loss to lease, second: the spread between in-place rents and market is the value story on most acquisitions, and it is not one number, it is a distribution across the rent roll, with some units five percent under market and some twenty. The pace at which that spread closes, lease by lease as each one turns or renews, is the actual shape of the deal's income growth. An average grows smoothly. A rent roll does not.
Third, and least modeled: the renewal-versus-turn decision, which every unit makes every year. A renewal at a modest increase costs almost nothing. A turn captures the full mark to market and pays for it in vacancy days, make-ready cost, and leasing effort, plus renovation capex if the plan calls for it. The economics of the whole reposition live inside that per-unit decision repeated hundreds of times, and a blanket turnover percentage flattens it into noise. Whether the plan pencils frequently comes down to sequencing: how many units per month the team can actually renovate, and what carrying half-done inventory costs while they do.
None of this is beyond Excel in principle. All of it is beyond what most operators will maintain in Excel in practice, because unit-level modeling at two hundred rows with per-unit states, timing, and capex is exactly the kind of spreadsheet that becomes unmaintainable, key-person dependent, and quietly wrong. The compromise everyone reaches, the average unit, is not a modeling choice. It is a surrender to the medium.
A simulator changes the terms of that surrender the same way it does in retail: the unit-level structure is native, so the depth costs nothing extra. In PropCalc™, multifamily runs on the same deterministic engine as every other asset class, with the mix, the loss to lease, and the reposition plan carried through a full capital stack and waterfall, recalculating live as assumptions move. The average unit is available as a summary. It is just no longer the model.
There is a live multifamily deal on this site, fictional numbers, fully inspectable, linked below. The instructive experiment takes one minute: change the reposition pace and watch what it does to the equity multiple. That sensitivity is the deal, and it is the exact thing the average unit cannot show you.