The Assumptions Behind Every Model

The model is an argument, not a forecast

Every underwriting model does its arithmetic faithfully. It tells you the exact truth about the assumptions you fed it, and nothing about the property itself.

A spreadsheet hands you a clean number (18.5% cash-on-cash, a year's cash flow over the total cash invested), and the cleanliness makes it feel like a fact. It is not a fact. It is the mechanical result of the assumptions you chose. Move one of them (occupancy from 70% to 62%) and the "fact" slides several points. The arithmetic never lied; it did exactly what you told it to with the inputs you gave it.

That is the trap of a model: the precision is real, but it all comes from the inputs. Rigor does not live in the third decimal place of the output. It lives in where the inputs came from. A projection carried to the dollar looks careful, but it is only ever as good as the assumption you were least sure of when you typed it.

The assumptions that actually matter

In most STR underwriting models, only a handful of assumptions dominate the result: the nightly rate, the occupancy, and the financing terms. The remaining inputs usually have a much smaller effect on the final decision.

It helps to sort the inputs into three buckets:

• Revenue — average daily rate (ADR, the average nightly price of the nights actually booked), occupancy, and how both move across a seasonal year.
• Cost — operating expenses with their maintenance line, and the capital reserve most models forget. The capital reserve is money set aside each year to replace things that wear out.
• Capital — the down payment, the interest rate and term, and the upfront cash to furnish and equip the property before it earns a dollar.

Within those, a small number carry the outcome. A perfect estimate of your streaming-subscription line will not rescue a deal built on a fantasy occupancy. So spend your attention where it changes the answer (the rate, the occupancy, and the debt), and don't polish the trivia. (The mechanics of the revenue inputs live in Estimating STR Revenue Without Fooling Yourself; the full cost stack in What STRs Actually Cost to Run.)

Where assumptions go wrong — almost always in the same direction

Assumptions rarely fail at random. They fail optimistically, and for three recurring reasons.

Anchoring. The first number you see sets the frame, and every later adjustment quietly stays close to it. The seller's pro-forma, the comp tool's projection, the agent's "it could do nine a month": once that figure is in your head, your "conservative" estimate tends to land just a little below it rather than where the evidence actually points. Seller and broker pro-formas are marketing documents, not financial projections; treat them as the seller's opening argument, never your baseline. The fix is order of operations: build the number from your own inputs before you look at anyone else's, then compare.

Survivorship bias. The comps you can see are the listings that survived. The ones that underperformed and got delisted, and the owner who quietly sold at a loss, are not in the data set. A market-average occupancy is an average of the survivors: a figure that, by construction, excludes every listing your property might turn out to resemble in a bad year.

False precision. "$8,437 a month" hides the truth that the real range is something like $6,000 to $10,000 depending on a season you don't control. A number to the dollar disguises a guess as a measurement, and a guess you've disguised from yourself is one you'll defend instead of test.

The four moves — The Builders Finance Underwriting Method

Builders Finance handles every number in a deal the same four ways: Source it, Haircut it, Record it, Stress it — the Builders Finance Underwriting Method. They are not four ideas; they are one process, and every underwriting decision runs through it. Most of the rest of this guide is those four moves in detail; The STR Deal Underwriting Manual runs the same four moves across a whole deal.

Record it: The Assumptions Ledger

The most useful artifact in underwriting is not the model. It is the one-page ledger of the assumptions behind it.

For every core assumption, The Assumptions Ledger captures three things: the value you used, where it came from, and the haircut you applied and why. It takes ten minutes and it changes the character of the whole exercise: from a number you produced into a set of claims you can stand behind. The ledger below is filled in for the example deal these guides use, an illustration rather than a forecast.

The Assumptions Ledger does three things a bare model can't. It makes the underwrite auditable: you, a partner, or a lender can see exactly what the answer rests on. It makes it defensible: you can explain every input instead of pointing at a spreadsheet. And it makes it revisable: once you own the property, you replace each assumption with your actuals and learn precisely where your judgment was too optimistic, which is how the next underwrite gets sharper.

Stress it: break the model on purpose

Before you trust a model, break it on purpose. Move each dominant assumption a realistic amount and watch what happens to the answer.

Run occupancy down eight to ten points, pull the rate back ten to fifteen percent, and bump the interest rate up a hundred basis points (one percentage point): all three inputs that carry the outcome. If cash-on-cash stays positive and the deal still covers its debt after that, the assumptions have genuine margin of safety, and the base case is a floor you can live on. If a mild, entirely plausible move flips it from "buy" to "pass," then the base case was riding on optimism, and now you know it, before your money is in.

Stress-testing also tells you where to spend your remaining research time. If the answer barely flinches when you change the cleaning-cost line but swings hard on occupancy, occupancy is the assumption to go get right. (See Does This Deal Actually Pencil? for break-even occupancy and Estimating STR Revenue for revenue.)