How Brixfox Computes Its Netherlands Numbers

The catalogue: recreatiewoningen, credited to their makelaar

We read every recreatiewoning for sale in the Netherlands from the listing pages of the selling agents, daily for new arrivals and monthly for the whole catalogue. Building plots and generic “objects” — caravan pitches, berths, a chalet with no bedroom count — are excluded: there is nothing to underwrite. Ordinary homes that happen to appear in a holiday search (permanent residence permitted) are excluded too, because the nights we would model for them may not be lettable at all.

Each listing is credited to the makelaar selling it, resolved from the agent's own profile, with a link to their site where they publish one. Until an agent is resolved — usually within a day of a listing appearing — the page says “listing agent”. We never present the portal we found a listing on as the seller.

A home that sells is retired on its own page's evidence: the listing status changes to verkocht and, where it is printed, the sale date is kept. Absence from a crawl only decides which pages get re-read; it never retires a listing by itself. Prices are asking prices, kosten koper unless the listing states otherwise, and the Netherlands publishes no per-sale price register we can use — so treat medians as the top of the negotiating range.

Expected ADR and occupancy: comps by distance

The nightly rate and occupancy we project come from the Airbnb listings we track around the property's own coordinates — Funda publishes exact ones — selected by real distance and matched on bedroom count, never by park or town name. A projection needs at least five comparable listings and a minimum comp-engine confidence, or it is not made.

Occupancy is derived from real availability calendars, tracked daily and diffed to detect bookings. We have tracked Dutch calendars since May 2026, night by night from July — one summer and no winter. Dutch holiday letting is sharply seasonal, so an observed summer occupancy is scaled to an annual equivalent (a factor of 0.62 until twelve months exist) and capped at 55%, with market-level annual figures capped at 72%. Nightly rates are the rates hosts ask. A structured waterfront or sea-view position on the listing carries a measured +10% rate premium; a photo-quality adjustment moves the rate within a stated band and its reasoning is shown on the page.

Where a property is matched to its own live Airbnb listing — photo-verified by two independent models — that listing's observed occupancy is shown alongside the projection.

Letting rules: read from the listing, capped in the model

Most Dutch holiday homes sit on a park, and the park contract — not national law — decides whether you may let independently, must let through the park's rental pool, or may not let at all. There is no per-property register to check against. We read the position from each listing's own words and show it as a signal — own letting, park pool, mixed, restricted, or unknown — never as a permission.

Where a listing states a cap on the nights that may be let, the projected revenue is capped at those nights: a 50-night limit against a 35% market occupancy sells 50 nights, not 128, and the page says which set the number. Stated park charges, ground rent on a leased plot, and any transfer-tax or notary exemption the listing claims replace the model's defaults for that home, with the sentence they were read from kept for the buyer to check.

Net yield: fully managed, after Box 3, on the all-in cost

The headline number is the net yield of a fully managed holiday let, after the Box 3 wealth tax, divided by the all-in investment — the asking price plus the costs a buyer of a non-primary home actually pays: 10.4% transfer tax (overdrachtsbelasting, 2026 rate), about €1,800 of notary and Kadaster fees, €1,200 of legal due diligence, and turnkey furnishing with short-let launch costs (€15,000 for a house or chalet, €9,000 for an apartment). A self-managed scenario is shown separately as upside: it drops the letting commission but prices the owner's time at zero, which is why it is an upper bound rather than the default.

Operating costs cover the letting agency or park pool at 30% of gross, channel and payment fees at 4%, cleaning and consumables net of guest recharges at 5%, the park service charge (€2,800 a year for a house, €1,800 for an apartment, unless the listing states its own), municipal property tax (OZB at about 0.11% of a WOZ value taken at 85% of market), a 0.8% annual maintenance reserve — recreation homes run higher than city flats — and insurance. Where the stated charges exceed the revenue the park permits, the page says the home loses money rather than rounding the loss to zero.

A private owner pays no income tax on the rent. Instead the home is taxed in Box 3 on a deemed return — 6.04% of the asset value at the 36% 2026 tariff, about 2.2% of value a year — which we deduct as an annual cost. On roughly a tenth of the catalogue that tax consumes the entire net return, and the page says so. Everything is stated before home-country personal tax and before financing; capital value is projected at 2.5% a year for the five- and ten-year figures.

Ownership and zoning

Dutch holiday homes are held freehold (eigendom); a chalet may stand on a leased plot, in which case the annual ground rent is read from the listing and charged. A recreatiewoning is zoned for holiday use — year-round residence is not permitted — and that is what makes it purpose-built for short stays. Guests pay the municipality's tourist tax per night, and the municipality sets any registration rules; verify both for the address before you buy.

When we show no number

A projection with too little behind it is worse than none. When fewer than five comparable listings are found, when the comp engine's own confidence is too low, or when the implied net yield exceeds 12% — the model's own sanity ceiling, above which the likelier explanation is a broken price rather than an exceptional home — the financial figures are withheld and the listing is flagged. A price that is not a sale price at all (price on request, sale by tender, a park's entry fee scraped by mistake) is refused rather than published. The other scoring pillars stand on their own evidence, so such a listing keeps a lower, real score instead of a hollow one.