Market data methodology

Sale and rent prices are asking prices from active listings, not final transaction or closing prices. Paraguay has no public transaction registry; asking prices are the observable data and we label them as such.

We use medians rather than averages to reduce the effect of outlier properties. $/m² only includes listings with valid area data.

The $/m² that headlines the market pages is published under the name "TuLugar Price Index": the median asking price per m² of apartments for sale, computed with one single definition across all four geographic scales (country, region, city, neighborhood). In the downloadable monthly series it is the median_apartment_ppsqm_usd column, with its month-over-month, quarterly and year-over-year changes.

When citing the figure, name it "TuLugar Price Index" and link the corresponding place page. Definition changes are documented on this page with dates and never restate already-published months.

The daily calculation base is the monitored active inventory: listings published on TuLugar plus active agency supply we track continuously and verify against its source. Parking spaces, listings not approved by moderation, and prices flagged as suspect by our anomaly detection (possible currency or data-entry errors) are excluded.

Deduplication: when the same unit is listed by several agencies within one identified building, exact reposts (same transaction type, price, area and bedrooms) are counted once. Distinct units in the same building count separately.

The headline "$/m²" is computed over apartments for sale ONLY. Apartment and house m² are built area (a house enters only when the source reports its built area separately from the plot); a lot’s is land alone — so adding them does not average prices, it mixes units. Houses, land and commercial are published separately, each with its unit stated.

Prices in other currencies are converted to USD at the exchange rate of the calculation day.

Methodology change (July 2, 2026): repost deduplication, the moderation filter and anomaly exclusion were introduced. Daily series before that date use the previous basis and are not recomputed.

Methodology change (July 30, 2026): the headline $/m² moved from apartments+houses to apartments ONLY, and all four geographic levels (country, region, city, neighborhood) moved onto a single shared definition. Previously different pages computed the same indicator differently and could disagree with each other. July 2026 was published from 10 new-basis days instead of 11 so the monthly series could start on the corrected definition.

Methodology change (July 21, 2026): the basis moved from listings published on TuLugar to the full monitored inventory, adding continuously tracked active agency supply — a several-times-larger sample. Earlier series keep the previous basis and are not recomputed. July 2026 is computed exclusively from new-basis days (July 21–30), and comparisons across different bases are omitted: a variation is only published when both months share the same basis.

Methodology change (September 6, 2026): counts and medians moved to unique units. The same property listed by several agencies, or received from two different sources, counts once when country, operation, type, city, currency, exact price and floor area agree. Earlier series are not recomputed; the full construction is in the general TuLugar Datos methodology.

Argentina (September 6, 2026): inventory is collected through a complete monthly census, in the last week of each month, and the month's figures correspond to that close. August 2026 was published from the first complete census, replacing the earlier version computed on a partial sample.

Currency reading (September 8, 2026): every price has one USD reading, decided from the currencies listings are quoted in within that country and from that country's plausible price range. A guaraní amount labelled as dollars is read as guaraníes; an amount with extra zeros is not corrected but left out of the price figures (it still counts in inventory). The price as published by the source is never modified.

Paraguay (September 7, 2026): the same complete monthly census, and August 2026 was published from the first one. Measured inventory went from about 23,000 properties to 172,400 unique units: that is a change of basis, not a change in the market, which is why August publishes no variation against the earlier months. Apartment price per m² barely moved between the two bases ($1,617 and $1,636), which says coverage changed and the price level did not.

Houses per built m² (September 16, 2026): house price per m² is now computed on built area rather than on the total plot. Only houses whose source reports built area enter, with built areas of 30 to 2,000 m² and prices of USD 100 to 10,000 per m²; every report states how many houses the figure rests on. In the same revision, rural estates were reclassified as land, and the median sale and rent prices on the market pages became apartment figures. August 2026 was republished on these definitions.

Each closed month is published as the mean of the month’s daily medians (more robust than an end-of-month snapshot), together with average and end-of-month inventory. A month needs at least 15 days of daily data to be published.

Published months are immutable: once closed they are never recomputed or retroactively corrected. Monthly, quarterly and annual variations are frozen at publication time.

Since July 2, 2026 the monthly series is also published broken down by property type (apartments, houses, land and commercial). The monthly reports cite per-type figures exclusively: blended all-type medians move with inventory composition — in Paraguay land is the largest category — and do not represent price changes.

We analyze public short-term-rental listing data: published rates, availability calendars and listing attributes. Market metrics are computed over entire properties only (no private or shared rooms), with nightly rates between $3 and $5,000 to exclude errors.

Occupancy is estimated from availability calendars; estimated monthly revenue derives from each listing’s observed rate and occupancy. Every market metric is the across-listing median of per-listing monthly values.

The monthly series includes listings tracked in each month even if they were later delisted — no survivorship bias. Each month needs at least 5 listings with data.

The series starts in March 2026: earlier data sits on a pre-fix pricing basis and is not comparable, so it is not published.

The "Trends (90 days)" sections use a daily series that includes all listing categories; it is labeled as such wherever it appears.

Daily statistics recompute automatically every day; monthly series publish automatically when each month closes. The data pipelines have freshness monitoring with alerts.

Site inventory (published listings) can differ from the deduplicated market inventory shown on these pages: they are different measures, each correct in its context.