A large share of the deals we eliminate look fine in the base case. They fail when the model is stressed, and they fail in predictable ways that a buyer working alone frequently does not think to test.
Test one: revenue at 75%
The single most useful test. Model the property at 75% of projected revenue and confirm it still services its debt from operations without drawing on the reserve.
This is not a doomsday scenario. New listings routinely underperform in year one while reviews accumulate, comparable sets shift as supply arrives, and projections built from good data are still projections.
If a property only works at 100% of projection, it has no margin for the ordinary variance of the business, and the ordinary variance is substantial.
Test two: a disrupted peak season
Three lost peak weeks. A named storm, a wildfire, a road closure, a regional event cancellation.
This is not exotic. Gulf Coast and Florida properties face hurricane season every year. California and Colorado mountain properties face wildfire and closures. Ski properties face poor snow years.
The test matters more in markets where revenue is concentrated. A Smokies cabin losing three weeks in July loses far more than a permitted Nashville rental losing three weeks in any month, because the Nashville property's revenue is spread across the year.
Test three: insurance repricing
Model the insurance renewal at 40% above the quoted figure. On the Gulf Coast, in Florida generally and in wildfire-exposed areas of the West, this is not hypothetical.
Coastal wind and flood coverage has repriced sharply across the Gulf. Several California carriers have reduced or withdrawn coverage in high fire risk areas.
Insurance is also the expense most commonly carried into a model at a guessed figure rather than a quote. Quote it for the specific address before the offer, then stress the quote.
Test four: property tax reassessment
Model property tax at the reassessed value based on your purchase price, not at the seller's current assessment.
This catches many buyers, particularly in markets where the seller has held the property for a long time and the assessment lags the market substantially. The reassessment can increase the tax line by a large multiple.
In Colorado there is an additional layer: the state has debated reclassifying short-term rentals toward a commercial property tax rate, which would raise operating costs on every affected property. That belongs in any Colorado model as a scenario.
Test five: the long-term rental floor
For city properties and for any property in a jurisdiction with meaningful regulatory risk, model what the property is worth as a conventional twelve-month rental if the short-term regime tightens.
The deal has to be acceptable at that floor. If it is not, the purchase is a policy bet with a mortgage attached rather than a real estate investment.
In Nashville the floor is genuinely strong, supported by healthcare employment and sustained population growth. In a tourism-dependent city with weak long-term rents it may not be, and that difference should change what you are willing to pay.
What passing looks like
- Revenue at 75%: debt service covered from operations.
- Three lost peak weeks: the year still works, with the reserve absorbing the shortfall.
- Insurance 40% higher: cash flow reduced but still positive.
- Reassessed property tax: already in the base model, not a surprise.
- Long-term rental floor: an acceptable outcome rather than a disaster.
A property clearing all five is rare, which is why we eliminate about 98% of the roughly one thousand deals we screen each week. Most of the elimination at this stage involves properties the buyer already likes, which is the hardest kind to reject.
The test that is not on the list
Deliberately absent: a test assuming you can refinance in two years. If the plan depends on refinancing, the plan has a dependency rather than a strategy.
Rate environments do not cooperate on schedule. A property that only works after a refinance is a property that only works under a condition outside your control.
The same applies to a plan that depends on appreciation. Appreciation is a reasonable expectation over a long hold and a poor foundation for a purchase that does not otherwise work.
Running them properly
Each test should be a separate column in the model rather than a mental adjustment. Base case, revenue at 75%, disrupted peak, insurance repriced, and where relevant the long-term rental floor, side by side.
That structure makes the failure visible. A property that clears the base case and fails at 75% shows it in a cell rather than in a feeling, and it makes the conversation with a client concrete rather than cautionary.
It also makes the tests reusable. Once built, the same model applies to the next property with different inputs, which is a substantial part of why reviewing volume is possible at all.
The final discipline is to run the tests before falling for the property rather than after. A stress test run on a property you have already decided to buy tends to find reasons the scenario is unlikely.
Keep reading
Frequently asked questions
What stress tests should I run on a short-term rental purchase?
Revenue at 75% of projection, three lost peak weeks from a storm or closure, insurance renewal 40% higher, property tax reassessed at your purchase price rather than the seller's assessment, and for city properties the long-term rental floor.
Why model property tax at the purchase price?
Because reassessment on sale can increase the tax line by a large multiple, particularly where the seller has held the property a long time and the assessment lags the market. Using the seller's current figure understates the cost substantially.
Should I plan on refinancing to make a deal work?
No. A property that only works after a refinance depends on a rate environment outside your control. If the plan requires refinancing, the plan has a dependency rather than a strategy.