Methodology
Written by Matic Pogladič · Draga Slovenija / Tehster OÜ
Why the median, not the average
Every price on this site is a median: the middle value, with half the transactions below it and half above. Averages are useless on a property market because a handful of transactions move them: one penthouse in a municipality that sees fifty sales a year lifts the average by several per cent and moves the median by nothing. Property prices are right-skewed (the upper tail is long, the lower one cannot go below zero), and for that shape the median is the measure that describes a typical transaction.
The median is computed on the price per square metre rather than the total, because homes vary more by size than by quality. A price per square metre is attributed to a transaction only when the attribution is unambiguous: one whole building part, or one whole parcel. A contract that transfers an apartment, a garage and a storage room for a single sum counts as a recorded sale but never enters the €/m² calculation.
The open-market filter
Only transactions the register flags as open-market enter the statistics. Excluded are:
- sales between family members and between related parties,
- sales at public auction in enforcement, bankruptcy or compulsory settlement,
- expropriations and sales in lieu of expropriation,
- finance leases and the creation or transfer of a building right,
- for leases: letting between related parties, and non-profit or restitution tenancies.
Their prices do not reflect market conditions and would drag the median down. The transaction counts on each page leave them out too. The raw register holds more rows than there are transactions, because it keeps one row per unit (building part or parcel) and a contract covering several units is one transaction; we count each transaction once.
When a figure is printed at all
A small sample gives you a precise-looking number that means nothing. Hence the thresholds:
- 5 priced transactions in a group before a median appears in the breakdown table;
- 10 before a median takes a headline card or the page title;
- 15 in a year before that year plots a point on a trend line: the gap stays a gap, because empty space is more honest than a straight line through a number we do not have;
- 20 on each side (sales and leases) before a gross yield is computed;
- 200 in total and 30 in the last two years before a settlement or cadastral district gets a page of its own.
Where a threshold is not met, no estimate is printed in its place. What happens instead is the next section.
When there are too few transactions
The headline figures stand on the last two years. In smaller settlements and cadastral districts the number of priced transactions in that period is sometimes below 10, which is too few for a median worth trusting. Rather than print an empty card, the page widens the period: first to the last five years, then to the last ten, and finally to the whole register, taking the first period that clears the floor.
The period used is printed beside the figure, every time, so it is always clear what the number was computed from. Comparisons against the municipality or against Slovenia are then computed over that same period on both sides; where no reference value exists for it, the comparison is dropped rather than made across two different periods, which would report the calendar as if it were a premium. If not even the whole register clears the floor, the page says there is no data and states how many transactions are on record.
Two things deliberately do not widen. The page title and the search description stay on the standard two-year window, because a headline price drawn from a decade would misrepresent the page in a result list. So does the by-municipality ranking, where 212 places are put in one order and have to be measured over one period to be ranked at all. Trend charts are unaffected: they already plot each year separately, wherever a year has enough transactions to plot.
Data vintage and period
The headline sale figures cover the last two recorded years (for example 2025–2026). The window is measured on the year of recording, the same axis the trend charts and the municipality ranking use, so one period label means one thing everywhere and no comparison runs across two different axes. It is counted back from the newest year in the register, not from today, which would quietly empty the window whenever a release runs late. The date of the last transaction and the date of the last rebuild are both in the footer of every page.
Trend charts run back to 2007 for sales and 2013 for leases, the years each register opened. Values are nominal, with no inflation adjustment, so they compare directly with the prices in today's listings and contracts.
The authority publishes a new annual file and tops up the current year during the year. The last year is therefore usually still under-filled: a transaction reaches the register weeks to months after it is signed.
Leases: signing year and the mature window
The lease register is an annual snapshot of running contracts: the same contract reappears every year it remains in force. A time series built on the reporting year therefore measures how long tenancies last, not what rent costs. Every rent series here stands on the year the contract was signed. Contracts signed before 2013 appear only if they were still running then, so they are not a sample of their own year and are cut off.
Leases are filed with a long, uneven lag: a signing year keeps gaining late-filed contracts for about two more years, and those late filings run systematically dearer, so the newest one or two signing years read both thin and about 15% too high. The rent card and the rent trend therefore stand on the last two mature signing years (currently 2024–2025), where filing is essentially complete, not the newest; years that have not matured are not drawn. Each rent figure prints the period it was measured over.
A lease is also measured in €/m² per month and must fall inside a plausible band (0.1 to 500 €/m²); a value outside it is a data-entry slip and never enters a median. When one block deal (worker rooms, say) is filed as several identical leases at one address, with the same day, rent and area, it enters a median once.
Gross rental yield
The formula is simple, and gross on purpose:
yield = (median rent €/m² per month × 12) / median price €/m² × 100
Each median comes from its own register's standard window for the same area and for apartments: the rent from the mature lease window, the price from the two-year sales window. The figure deducts no rental income tax, maintenance or management cost, reserve fund, insurance, agency fee, or months when the apartment stands empty. A net yield is in practice a third to a half lower. Use it to compare places, not to forecast the return on a particular investment.
Place names
The register writes the same place in capitals in one file and in mixed case in another, and abbreviates some municipalities (“SV. TROJICA V SLOV. GORICAH”). Both spellings are folded onto one page. The Slovenian grammatical forms of municipality names are written by hand; for settlements and cadastral municipalities the locative is derived by rule, and where the ending is ambiguous the page uses the neutral “v naselju X” form rather than guessing.
Corrections
If a figure looks wrong, we will fix it: write to info@dragaslovenija.si with the page address. An error in the source register is corrected by the Surveying and Mapping Authority; an error in our calculation or wording is corrected here, usually within days, and the page gets a new last-updated date.
The data is published under CC BY 4.0. Cite it as: Draga Slovenija (dragaslovenija.si), GURS ETN data. In Slovenian: Vir: Draga Slovenija (dragaslovenija.si), podatki GURS ETN.
Interactive map
Every transaction on this page is on the interactive map: with filters by year, price, floor area and property type, with parcel boundaries and building footprints.