How we calculate the data

Asking prices, not sale prices. Everything comes from the for-sale listings published on Casas Oasis (1,516 listings analyzed). They are asking prices: what the seller asks for in the listing, not the price the home finally sold for. Recomputed daily. Nothing is hand-written.

What counts as recent. Homes for sale today, plus those that left the market in the last 6 months. "Recent" is measured by the real date the listing was last on sale (the day it was withdrawn, sold or expired; or, if that is not recorded, the day it was published), never by the last time the listing was edited.

What we include. Only homes priced in USD at ≥ $3,000, with bedrooms and bathrooms stated (without them the record isn't reliable). Land and commercial units are excluded (they are not homes).

What stays out. Reference market data we do not publish (listings you cannot see on the site) never enters these prices, and neither do repeated listings: a home that moderation identified as advertised twice counts once.

Outlandish prices. We compare each price with what is usual in its province and leave out the ones that are extremely far from it, using a robust measure that the extremes themselves cannot drag along (technically: a modified z-score above 3.5 on the logarithm of the price). That is how a typo such as a USD 80 million listing stops distorting the maximum or the average. Ordinary expensive or cheap homes stay in.

Median price. The headline figure of every area is the median (the middle value: half ask less, half ask more), not the mean.

Standardized price. To compare areas we standardize by size: we take the median for each size (1, 2, 3, 4+ bedrooms) and blend them weighted by the country's typical mix of homes (1 bed 8% · 2 bed 33% · 3 bed 32% · 4 bed 16% · 5 bed 6% · 6+ bed 5%). That compares like with like: a province of mostly large houses no longer looks pricier than Havana just because of its mix. It's the same approach international indices use (e.g. Zillow's ZHVI). Because it blends every size, it does not match the area's median: use it to compare areas with each other, not as the price of a home. Each size needs ≥ 5 listings to enter it, and the sizes with data must cover ≥ 60% of the mix.

Minimum sample. An area needs at least 10 recent listings to publish its median, and each breakdown (by bedrooms or by type) needs its own 10; below 30 the figure is flagged as indicative. With fewer than 10, the area's page publishes no median: it says there are few recent listings, shows the ones there are, the last known median (from the most recent 6-month period in which it did reach 10, with its dates, and only from 2021 on, because earlier prices were in CUC) and the median of the area that contains it, and it is kept out of search engines until there is data again.

Listings on Casas Oasis are priced in US dollars (USD), Cuban pesos (CUP) or euros (EUR), and each listing shows its price in the currency the seller chose.

The price statistics use USD listings only. We do not convert CUP or EUR prices into dollars: that would mean choosing an exchange rate, and the figure would depend more on that rate than on the homes. That is why those listings are left out of the medians.

The CUC (convertible peso) was retired with the 2021 currency unification. Earlier prices were in that currency and are not compared with today's: an area's “last known median” is only rebuilt from data from 2021 onwards.

The prices of every area and the market study are recomputed once a day, overnight, from the inventory at that moment.

Homes for sale that day count, plus those that left the market in the last 6 months. It is a rolling window: a price more than 6 months old stops counting.

The selling guide compares the views and contacts listings receive depending on how many photos they carry. It is measured on active listings that have been live for at least 45 days (one from yesterday has not had time to be seen) and is recomputed every day. A “contact” is someone who tapped the listing's phone number or WhatsApp.