How We Rank Rental Markets: The Math Behind Our Lists

Our database holds 3,073 US counties. Only about 1,400 of them can appear in a ranking we publish. The other 1,700 or so have a home price but no usable rent figure, and a rental market list built on a guessed rent is a list of guesses. So more than half of the country's counties are left out on purpose, and this post explains that choice along with every other one behind the lists.

A ranking is an opinion expressed in arithmetic. You should know the inputs, the assumptions and the blind spots before you let one steer a purchase. What follows is the full method behind the best markets rankings, including a list we built and then removed because it was a duplicate in disguise.

What the rankings measure and the data behind them

Every ranking starts from the same question: if you bought the median-priced home in this county and rented it at the local median rent, what would the numbers look like?

The data

Zillow publishes rent figures only where it sees enough rental listings to form an index, which is why rent coverage is less than half of price coverage. The counties left out are mostly smaller ones with few rental listings.

The underwriting

Each rankable county gets the same modeled purchase, so the only things that vary between counties are price and rent:

• 20 percent down on a 30-year fixed loan
• One mortgage rate applied to every county, printed in the methodology note on each ranking page
• Closing costs at 3 percent of price
• Operating costs at 35 percent of rent, covering taxes, insurance, maintenance, vacancy and management

From those come a monthly mortgage payment, monthly cash flow in dollars, a cap rate and the cash needed to close. National lists show the top 50 counties and require a population of at least 50,000, which keeps a single odd sale in a tiny county from topping a national list. The price-band lists and state lists have no population floor because they are already narrow. The metro rankings use the same underwriting on metro-level medians, require a metro population of 250,000 and show the top 25.

The scores

Beyond the modeled deal, each county carries four component scores from 0 to 100.

Cash flow score comes from the rent-to-price ratio, defined as annual rent divided by home value. The tiers:

Appreciation score comes from year-over-year home price growth, and it is deliberately not a list of the fastest gainers. Steady growth of 3 to 8 percent a year scores highest. Growth above 12 percent is penalized as a sign of an overheated market. Falling prices score low but above zero.

Stability score at the county level rests on one input: population growth. Growing counties score high, shrinking ones score low, and a county with no growth figure gets a neutral 50. The scoring code has slots for unemployment and job growth, but we do not have those series at the county level, so they are not in the county score. That is a real limitation and it is covered below.

Affordability score is home price divided by median household income. A ratio of 2 or less scores 100 and the score steps down as the ratio climbs. Where income data is missing, the model substitutes $70,000.

Overall score blends the four: cash flow 30 percent, appreciation 25 percent, stability 25 percent, affordability 20 percent. On the market map, a county with no rent data still gets an overall score, reweighted to appreciation 40 percent, stability 30 percent and affordability 30 percent. Cash flow is dropped from the blend for those counties because an estimated rent would contaminate the score. The map labels scores of 75 and up Strong Buy, 60 to 74 Buy, 45 to 59 Hold and below 45 Avoid.

The refresh

Ranking pages rebuild on a 7-day cycle from the latest metrics in our database. The underlying county metrics are reloaded from Zillow's published files when we run the data refresh, and each ranking page prints the date of the metrics it used. The mortgage rate is a single assumption we update by hand, so check the rate shown on the page against a current quote before you read the cash flow figures as live.

Why dollar cash flow and cap rate produce different lists

We publish one list ranked by monthly cash flow in dollars and another ranked by cap rate. They disagree, and the disagreement is the useful part.

Here is a worked example with two invented counties, underwritten the way the rankings do it. The loan rate is an example 7 percent on a 30-year term.

Check County A: 1,200 - 639 - 420 is $141 a month. NOI is (1,200 - 420) x 12, or $9,360, and 9,360 / 120,000 is 7.8 percent. County B: 2,700 - 1,597 - 945 is $158, and 21,060 / 300,000 is 7.0 percent.

County B wins the cash flow list by $17 a month. County A wins the cap rate list by most of a point and returns more than twice as much on the cash invested. Both lists are correct. They answer different questions. The dollar list favors higher-priced markets, because a thin margin on a big number beats a fat margin on a small one. It suits an investor who wants fewer doors and has the capital to buy them. The cap rate list measures yield per dollar and ignores financing, which makes it the cleaner comparison across markets. An investor with $70,000 could buy two houses in County A, collect $282 a month and have cash left over.

The trap of duplicate rankings: one metric in different clothes

A site can publish ten "best markets" lists that are really one list with ten titles. We now check for that with algebra before a list ships. The check has caught two duplicates so far, one before launch and one after a page was already live.

Rent-to-price and cap rate. With operating costs fixed at 35 percent of rent, cap rate equals the rent-to-price ratio times 0.65. Multiplying every county's ratio by the same constant cannot change the order. A "best rent-to-price markets" page would be the cap rate page with a different headline, so only one of the two is published.

Cap rate and cash-on-cash return. This one is less obvious. When the down payment, closing costs and loan all scale with price, the price terms cancel, and cash-on-cash collapses to:

cash-on-cash = cap rate / 0.23 - a constant

At the example 7 percent loan the constant is 27.8 percentage points. Test it on County A: 7.8 / 0.23 is 33.9, minus 27.8 is 6.1 percent, matching the table. County B: 7.02 / 0.23 is 30.5, minus 27.8 is 2.7 percent. Dividing by a fixed number and subtracting a fixed number never reorders a list. We did publish a cash-on-cash ranking for a short time. When we compared its output against the cap rate ranking, it showed the same markets in the same order, so we removed it and pointed its address at the cap rate list.

Dollar cash flow survives the test because it subtracts an absolute mortgage payment instead of dividing by price, so it does reorder the counties. A cash-on-cash ranking that carries its own information needs an input that does not scale with price, such as each county's actual property tax rate. We do not have that in the model yet, so the list does not exist yet.

The same logic explains the rest of the lineup. The under $200,000 list filters by price and then ranks by cap rate. The cheapest markets with positive cash flow list sorts by price, lowest first, and keeps only counties where modeled cash flow is above zero. The appreciation and best overall lists sort on their scores. Because scores land on a limited set of values, ties are common. Ties are broken by population, larger first, so the order does not reshuffle between page rebuilds.

The thin-data rule: why some states have no page

A state gets its own ranking page only if at least five of its counties have both price and rent data. Below that, the page would be a top three of three, which tells you nothing about where to invest and exists only to catch a search. We omit it instead.

The same rule removes 1,714 counties from every list. In our data a county with no rent has a cash flow score of zero. That zero is a placeholder for missing data, and a sort that ignored the difference would call half of rural America the worst rental markets in the country. Every ranking query filters to counties with a real price and a real rent before it sorts.

The cost of this rule is coverage. If you invest in a small county that Zillow does not index for rent, you will not find it here. You can still see its price, appreciation and population figures on the map, scored without a cash flow component.

What rankings cannot tell you about a specific deal

These are modeled estimates for comparing markets. Here is where they diverge from a real purchase.

• The median home is not the typical rental. ZHVI tracks the typical home value across all housing in a county. ZORI tracks asking rents on units listed for rent. The house at the median price and the unit at the median rent are usually different properties, often in different neighborhoods. The ratio between them is a market-level signal and says little about any one building.
• Operating costs are a flat 35 percent of rent everywhere. Property tax and insurance vary enormously between counties, and the model does not see that. A high-tax county is flattered by the rankings and a low-tax county is penalized.
• One mortgage rate for everyone. Your rate depends on your credit, loan type and down payment, and investor loans price above owner-occupied loans.
• A county is a big place. Block-level differences in rent, crime and tenant demand are larger than the differences between many counties on the list.
• Stability is population growth alone. Employer concentration, job growth and unemployment are not in the county score.
• Appreciation looks backward one year. A single year of price change says nothing reliable about the next five.
• Asking rents are not collected rents. ZORI measures listings, so concessions and vacancy are invisible.
• Metro population is derived. Our metro records carry no population figure, so we add up the populations of each metro's counties. The totals track official figures closely enough to apply the 250,000 floor, but they are approximations.
• Nothing here knows about condition, landlord law, insurance availability or the specific street.

From ranked list to underwritten offer

Use the lists to decide where to look, then stop using them.

1. Pick the list that matches your goal. Use cap rate if you are capital constrained, dollar cash flow if you want fewer and larger properties, and best overall if you want balance.
2. Shortlist three to five markets and open each on the market map. Check whether the score comes from one strong component or several, and look at the neighboring counties.
3. Replace every assumption with a real number for a real listing: the actual tax bill after reassessment, an insurance quote, rent comps for that ZIP and a lender quote.
4. Rerun the deal. Expect the result to differ from the county median, since your property is not the median.

The ranking math is ordinary rental underwriting: a mortgage payment, operating costs, cash flow and cap rate. The calculator runs that same arithmetic, so the engine that ranks about 1,400 counties on medians will underwrite your specific deal once you swap in the address, the quote and the tax bill. The list narrows the field, and the offer should rest on the property's own numbers.

If you want the reasoning behind the inputs rather than the mechanics, how to choose a rental market walks through the four factors in order, cash flow vs appreciation shows what the trade-off costs over ten years, and the 1% rule in 2026 explains why we screen on rent to price rather than on that rule. To run a specific property instead of a county median, use the single family calculator.