Every figure on a Brixfox ABC property page has a stored, auditable basis. This page explains that basis once, in full — what feeds each number, which assumptions we apply, and the cases where we deliberately show no number at all. The ABC islands are our smallest market and this page says so where it matters.
The ABC catalogue is homes advertised by estate agencies on Aruba, Bonaire and Curaçao. There is no single portal on these islands the way Idealista serves Spain or Funda serves the Netherlands, so we crawl each agency's own site — around thirty of them — once a week.
What does not enter the browsable set: building plots and land, commercial premises and offices, and homes the agency's own page marks sold, under contract, under offer or sale pending. Those are judged on the listing's own type, title and status words, not on a guess — a house on a big plot is a house, and a building with an apartment above the shop is a home. Multi-unit buildings, guest houses and apartment complexes stay in, because they are exactly the stock a short-let buyer prices.
Every night we re-read the longest-unchecked listings on their own pages, because no ABC agency takes a sold home's page down: several keep it up at the same web address with a small "sold" label, and one of them simply adds the word to the URL. A home whose own page says it is sold or gone is retired. A status badge on a card NEXT to the listing — in a "similar homes" rail, or in a search filter — is never treated as evidence about the listing you are reading; we only read what a page says about itself.
Agencies here quote in US dollars, euros, Antillean guilders (ANG) and the new Caribbean guilder (XCG) — about half the catalogue is not in dollars. Every price is stored in the currency the agency published and converted for display at the PEGGED rates, not a floating market rate: the Aruban florin and the Caribbean guilder are both pegged to the dollar, and the euro leg uses one reference rate held in code. So a dollar figure on an ABC page is a conversion of a real quoted price, and the same listing reads the same on its card, its map pin and its detail page. Those are asking prices, not closed sales, so treat any median as the top of the negotiating range.
A home's projected nightly rate and occupancy come from the real Airbnb listings around it. We track live calendars across all three islands and read each one daily, so an area's occupancy is what those calendars actually did, not a survey. A home's comps are the tracked listings within a short radius of its pin, in a matching bedroom band, weighted by distance and by how complete their calendar data is.
The nightly rate is an asking rate, and we say so everywhere. Airbnb does not publish what a night actually sold for. Occupancy is measured — it comes from watching a calendar change day by day — but the rate beside it is the price the host advertised, discounted towards a booked rate. Anyone quoting you a "realised ADR" from Airbnb data is quoting an asking price too.
Where a home's pin actually is matters, and it varies. Some ABC agencies publish an exact coordinate; several publish one pin for a whole resort, so every unit in it shares a position; and a few geocode every listing they hold to a single point. We classify each home's pin accordingly and treat a resort or fallback pin as what it is — a neighbourhood, not an address. The seasonal curve we annualise by is measured for these islands specifically: they sit outside the main hurricane belt, so their season is genuinely flatter than the rest of the Caribbean, and using a Caribbean-wide curve here would be wrong.
The net yield on an ABC page is the fully-managed, after-tax return: projected annual revenue less a full operating stack — professional management, booking channel fees, cleaning and consumables, utilities, maintenance and insurance — and less tax, over the acquisition cost. It is the number an owner who does not fly in to change the sheets would keep.
Where you also see a gross figure it is labelled gross and it is over the ASKING price, because that is the number property portals quote and it lets you compare. The two are never printed under one label. This was a real defect on these pages: the hero once showed one basis while every card beneath it showed another, and a listing whose card read 1.5% sat under a headline saying 4.8%. Every ABC figure now comes through one function.
The headline on the hub and the areas pages is the median of the top-decile homes by BrixfoxScore, with the cut-off printed beside it ("score ≥ N") and the whole-catalogue median underneath. It is the top of the market by our own ranking, stated as such — not the average home, and not a cherry-picked listing.
A projection that implies an implausible gross return on the asking price is withheld rather than published. Above a fixed ceiling the likeliest explanation is not an extraordinary deal but a broken input — a monthly rent scraped as a sale price, a mis-parsed figure, a bedroom count that reads 25 because the page said "2 bedrooms 5 bathrooms". Those rows publish no yield.
A home that publishes no yield publishes no revenue and no ROI either. That sounds obvious and it is the rule most easily broken: a later stage of the pipeline can recompute a figure the earlier one refused, and a blank yield beside a confident "30% ROI" is worse than showing nothing. A withheld row says so in words rather than printing a zero, because a zero looks like a measurement.
Aruba, Bonaire and Curaçao are three separate jurisdictions with three separate tax and permit regimes, and none of them publishes a short-term-rental register a third party can look a listing up in. So no ABC listing on this site is described as licensed, permitted or approved for short-term letting, and none ever will be on the evidence available. Where Portugal has the AL register and Spain has six regional ones, these islands have nothing we can check, and inventing a badge would be a false legal claim rather than a missing feature.
Bonaire decides short-stay letting by the zoning of each plot. Bonaire’s spatial plan (the ROB) allows short-stay letting, meaning stays of up to three months, only in designated zones. Whether a plot is in one differs by neighbourhood and by plot, and in residential zones it is not allowed. The island government has enforced that since May 2024. In March 2024 it wrote to residents of eleven neighbourhoods: Sabal Palm, La Palma, Suikerpalm, Belnem, Bona Bista, Bolivia Ariba, Nawati, Santa Barbara, Hato, Sabadeco and Lagun Hill. Where a listing’s pin can stand for its plot (the agency’s own map pin or an exact street address, with no other home for sale on it), Brixfox reads the zone at that pin from the island’s published ROB map and says on the property page what it shows. Where it cannot, the page says the zone is not known. A pin is not a surveyed plot boundary, so have the zone checked in the ROB, or by the notary, before you buy. Every Bonaire yield assumes the home may be let short-term. Sources: the spatial plan (ROB), and the island government's pages on letting a home and enforcement.
What the numbers do show is measured activity: the live Airbnb calendars we track are evidence that short-term letting demonstrably happens in an area, which is a different statement from it being permitted for a particular property. Confirm the rules for the specific island, district and property with a local lawyer before you underwrite anything.
We would rather you read this than discover it. The ABC crawl is weekly, not nightly, so a price can be up to a week old. Our Airbnb calendar history for these islands reaches back to 2026 only, so no full year — and no complete winter high season — has been measured yet; every page prints the window its figures were observed in. Some agencies publish approximate pins and a few publish one pin per resort, which widens the comp set for those homes.
And one listing can appear twice: a few agencies advertise the same home, and one agency serves the same listing under English and Dutch web addresses, so a small share of the catalogue is a duplicate of another card. We collapse what we can identify and are working on the rest.