Knowing how to underwrite a multifamily acquisition is not knowing how to build a proforma. The proforma is the easy part — a spreadsheet will happily compound a 3% rent bump to a number that wins you the deal and loses you money. Underwriting is the discipline of starting with what the property actually does today, proving every adjustment you make to it, and refusing to let the exit carry the return. Do it in the right order and the deal tells you the truth early.
What follows is the sequence a practitioner runs, in the order they run it: anchor on in-place performance, normalize the income, mark the gap to market, model the value-add honestly, then size the debt and the exit. Each step constrains the next. Skip one and the model still calculates — it just calculates the wrong deal.
Start with the T-12 and rent roll — together
The deal begins with two documents read against each other: the trailing-12 income statement (the T-12) and the current rent roll. The T-12 is what the property collected; the rent roll is what it is contracted to collect going forward. They never agree, and the disagreement is where the underwriting lives. This is the same discipline that governs any net-lease underwrite — verify the income before you value it — but multifamily has dozens of leases turning constantly, so the reconciliation is the work.
- Gross potential rent (GPR) — the rent roll's total if every unit were leased at its current asking rent. This is your ceiling, not your income.
- In-place rent — what the occupied units are actually contracted at. The gap between this and GPR is loss-to-lease plus vacancy.
- Economic vs. physical occupancy — physical counts bodies; economic counts dollars collected. A property 95% physically occupied but 88% economically occupied is hiding concessions, delinquency, or non-revenue units.
- Other income — RUBS reimbursements, parking, pet rent, late fees, laundry. Recurring and defensible, or a one-time true-up dressed up as run-rate? Separate them.
- Trailing-3 annualized — pull the last three months and annualize. If T-3 runs well above the T-12, the seller has been pushing rents into the sale. Underwrite the trend, not the headline.
Normalize the income before you trust it
The T-12 as delivered is a marketing document. Normalizing it means rebuilding it as the year a competent operator would actually run, expense line by expense line. This is the step juniors rush and principals slow down.
- Re-expense to ownership reality — taxes almost always reassess on sale; underwrite the post-close millage against your purchase price, not the seller's stale assessment. Insurance is rising fast in most markets — quote it, don't trend it.
- Management fee, always — even if the seller self-manages for free, load a market fee (typically 3–4% of effective gross). The next buyer will, and so will your lender.
- Replacement reserves — a real per-unit annual reserve, not zero. Agency debt will require it regardless.
- Payroll and contract services — confirm headcount matches the unit count and the actual staffing model. Deferred maintenance shows up as a suspiciously low repairs line.
- Strip the one-timers — a roof replacement, a legal settlement, a one-month insurance refund. These don't belong in a stabilized year. So do non-recurring income spikes.
Mark loss-to-lease and model the value-add
Now the upside. Loss-to-lease is the spread between in-place rent and verified market rent — and the word that matters is verified. Market rent comes from leased comps you can defend: same submarket, same vintage, same unit mix, adjusted for condition. A broker's pro forma rent is a hypothesis, not a comp. If the comps are renovated units and yours are classic, you are pricing a renovation you have not paid for yet.
That renovation is the value-add model, and it has three honest inputs:
- The capital budget — interior scope per unit, exterior and common-area work, and a contingency that survives contact with a real contractor. Tie it to a unit-by-unit schedule, not a blended average.
- The renovation premium — the rent lift a renovated unit actually achieves over a classic unit, proven by your own comps. Haircut it. A premium that looks like $250 in the comps underwrites at $200.
- The pace — how many units you can turn per month given occupancy, turnover, and crew capacity. The schedule drives when the income arrives, and timing drives the return more than the premium does.
Stage the units across a realistic timeline and let occupancy dip during heavy turn quarters. A model that renovates 200 units while holding 95% occupancy is describing a property that does not exist.
Size the debt, then solve the exit
Debt is not an input you choose — it is an output the deal allows. Lenders size to the lesser of three constraints, and you should solve all three before you believe a loan amount:
- DSCR — net operating income divided by debt service, typically held at 1.25x or better on stabilized income.
- Debt yield — NOI over loan amount, the lender's downside test that ignores rate and amortization. Agency floors often sit around 6–8%.
- LTV / LTC — the appraised-value or cost ceiling, whichever binds.
Whichever constraint produces the smallest loan is your loan. Then the exit. Resist the instinct to compress the exit cap below your going-in cap; the defensible move is to exit at or above where you entered, then test what a 50–75 basis point expansion does to your equity multiple. If the deal only works on cap compression, you are underwriting the market, not the property.
Most multifamily models don't break in the math — they break in the assumptions feeding it: a renovation premium taken at full comp, a tax line left at the seller's basis, and an exit cap quietly tighter than the entry. Each is invisible in the IRR and fatal to the equity.
Where principal review earns its keep
A junior builds a model that calculates. A principal asks where it lies. The catches are almost never arithmetic — they are judgment: the T-3 running hot because the seller dumped concessions before listing, the "other income" that was a one-time RUBS true-up, the renovation premium pulled from comps in a better school district, the insurance line trended at 3% in a market repricing at 30%. None of these throw an error. All of them turn a 16% IRR into an 11% one.
That is the discipline behind our underwriting work: the in-place numbers get reconciled to source, every adjustment is sourced and defensible, and the same schema runs on every deal so the assumptions are visible rather than buried. The output flows straight into the investment memo the decision actually gets made from, and the inputs are the kind we standardize in the underwriting toolkit. Run the sequence in order, prove each step before you build on it, and the deal will tell you what it is long before the closing table does.
Need this on a live deal? Capistrano produces underwriting, lease abstracts, investment memos, and capital-raise materials — AI-leveraged, principal-reviewed.