Underwriting an arbitrage deal: the five numbers that matter

A good deal survives a bad month. Underwrite for that, not for the best month on the comps. Most arbitrage deals that go wrong were not bad luck. They were underwritten on a summer average, a guessed cleaning cost and no plan for the slow season.

This guide walks through the five numbers that decide whether a unit works: revenue, occupancy, operating costs, startup capital and payback period. Each section explains how to estimate the number, and the last section runs one full example. Every figure in it is an assumption for illustration, not a market benchmark. Replace each one with your own research before you sign anything.

Revenue is nightly rate times booked nights. The mistake is taking both from the best listings in town. Build your estimate from units that look like yours, not units you wish yours were.

How to pick comparables:

• Same bedroom count and similar sleeping capacity. A two-bedroom that sleeps six is not a comp for one that sleeps four.
• Within walking distance, or at least the same neighborhood. A few blocks can change demand.
• Similar property type and amenities: parking, laundry, outdoor space, a pool if the building has one.
• Listings with a full year of history, so you can read trailing-twelve-month revenue rather than a single strong month.

Pull five to ten comps and throw out the top and bottom. The top performers often have years of reviews and a professional operator behind them. Use the middle of what is left.

Then discount. For your first 90 days you will have no reviews, so you will usually need to price below established listings to win bookings. A simple approach is to model those months at a fixed percentage below your base case (say 20%) and see whether the deal still holds. If a lower rate in the ramp-up period breaks the deal, it was too thin to begin with.

Occupancy is booked nights divided by available nights. An annual average hides the problem that matters most: cash flow is monthly and rent is due every month.

A market at 75% in summer and 40% in winter is not a 57% market for cash planning. In the low months you still pay full rent, full utilities and full software costs, with roughly half the bookings. If you only model the average, you will not see the months where the unit loses money until you are living through them.

Split the year into three seasons:

• Peak: the months your comps are nearly full and rates are highest.
• Shoulder: the in-between months.
• Low: the months when demand drops. Look at comp calendars and pricing in those months, not just annual totals.

Model each season with its own occupancy and its own nightly rate. Rates usually fall when occupancy falls, so a low month gets hit twice. If you plan to fill slow months with mid-term stays, model that as a separate scenario with lower nightly revenue and fewer turnovers, and check that your lease and local rules allow it.

Operating costs are where optimistic spreadsheets quietly cheat. List every line, even the small ones. Small lines add up to a real share of revenue.

1. Rent. Your largest fixed cost, due regardless of bookings.
2. Utilities and internet. Guests run heating and air conditioning harder than long-term tenants. Get past bills from the landlord if you can.
3. Cleaning and laundry. Price it per turnover, then estimate turnovers from your average stay length. Shorter stays mean more turnovers.
4. Platform fees. Check the current host fee on each platform you plan to use and apply it to revenue.
5. Supplies. Toiletries, paper goods, coffee, replacement linens and towels.
6. Software. Dynamic pricing, guest messaging, smart locks, noise monitoring.
7. Insurance. A short-term rental policy, priced from a real quote, not a guess.
8. Maintenance reserve. A percentage of revenue set aside for broken appliances, damaged furniture and repairs the lease puts on you.
9. Taxes and permits. Occupancy taxes, registration and permit costs. These vary by city and county, so look them up for the specific address. Our regulations checklist covers what to look for.

Two lines get forgotten most: the reserve, because nothing has broken yet, and taxes, because operators assume the platform handles all of them. Confirm what the platform collects and what you must file yourself.

Startup capital is the cash that leaves your account before the unit earns anything. Count all of it:

• Security deposit.
• First month of rent, and any last month or move-in fees the lease requires.
• Furnishing and setup: furniture, beds, kitchen kit, linens, decor, smart lock.
• Professional photography.
• Permit or registration fees, if the address requires them.
• A cash buffer of at least two months of rent.

The buffer is not optional. It is what carries you through a slow first season, a cancelled booking streak or a big repair. Some operators keep it in a separate account so it is not quietly spent on the next unit. For a fuller line-by-line budget, see how much it costs to start rental arbitrage.

The payback period is startup capital divided by monthly net cash flow. It answers the question that matters on a lease with an end date: will you earn back what you put in before the lease runs out?

Many operators aim to recover setup costs within the first year of the lease. A shorter payback gives you room for things you did not plan for. A payback longer than the lease term means you are betting on a renewal you do not control.

Use the conservative monthly net from your seasonal model, including the low months and the ramp-up discount, not the peak-month figure. Leave the cash buffer out of the calculation if you expect to keep it, but add back anything you end up spending from it.

Everything below is an assumption chosen to show the method. None of it describes a real market.

Assumptions: a two-bedroom unit, rent of $1,800 a month, utilities and internet $250, cleaning $80 per turnover, a 3% platform host fee, supplies $100, software $60, insurance $50, and a maintenance reserve of 5% of revenue. Four peak months, four shoulder months and four low months. Taxes and permit costs are left out to keep the example simple; in a real model, they go in.

Over a year that is 4 x $1,737 + 4 x $676 + 4 x -$814, or $6,396, which averages about $533 a month. Notice the low season: four months at -$814 is $3,256, almost the entire two-month buffer of $3,600.

Startup capital (assumed): deposit $1,800, first month $1,800, furnishing $8,000 and photography $300, or $11,900 at risk, plus the $3,600 buffer.

Payback: $11,900 divided by $533 is about 22 months. If you apply a 20% ramp-up discount to the first three shoulder months, first-year net falls by roughly $2,100 and payback stretches further still.

On these assumptions the deal fails a one-year payback target. That is a useful result. It tells you what has to change before you sign: lower rent, a cheaper furnishing plan, a longer lease or better low-season revenue. If none of those is realistic, walk away. You can rerun scenarios like this quickly in the rental arbitrage calculator.

• Do you have written landlord consent for short-term or mid-term rental, and do local rules and any HOA allow it at this address?
• Are your comps the same bedroom count and close by, with the top and bottom thrown out?
• Did you model the low season on its own, with its lower rate?
• Is every cost line in, including the reserve, taxes and permits?
• Does the buffer cover your worst run of months?
• Is the payback period comfortably shorter than the lease term?

Listings on ArbiBase come with consent and a rules check already documented, but run these five numbers yourself on every deal. The underwriting only protects you if you believe the inputs.

Treat underwriting as a filter, not a sales pitch to yourself. Build the seasonal model, use the conservative month, and only sign when the deal still pays back inside the lease after a bad season and a slow start.