SafeBuy · International research
Sources, scores and limitations
A useful comparison lets you inspect its assumptions. Here is what goes into this research, how the finder calculates a score and where the evidence stops.
What this dataset represents
Version 2026-09-initial · last editorial update 2026-09-07. Coverage: Georgia, Spain, Portugal, Greece, France, Germany. We start with a limited set that can be researched responsibly. Coverage is not a claim to include every attractive property market.
Each observation stores its value, unit, original currency where relevant, geographic scope, reference period, source publication date when known, review date, confidence and methodology. Country profiles separate sourced rules from SafeBuy’s editorial interpretation. Unknown data remains unknown.
Government, statistical, tax and immigration sources are preferred for rules and official series. Original market research can be used for asking-price and rental samples. These are different measurements: national transaction indices measure change, listing benchmarks measure advertised prices, and neither proves the value of a specific property.
Transparent scoring
Configuration 1.0. For each selected dimension, a published value is scaled to 0–100 using fixed anchors. Values outside the anchors are clipped. For a higher-is-better dimension: 100 × (value − minimum) ÷ (maximum − minimum). A lower-is-better dimension reverses that result.
| Dimension | Minimum | Maximum | Preferred direction | Default weight |
|---|---|---|---|---|
| Lower purchase price | 500 EUR/m² | 8000 EUR/m² | Lower | 2 |
| Gross rental yield | 0% | 12% | Higher | 3 |
| Historical price growth | -10% | 20% | Higher | 0 |
| Lower total buying costs | 0% | 20% | Lower | 0 |
The final score is Σ(normalised score × user weight) ÷ Σ(available user weights). A score is published only when evidence covers at least 60% of selected weights. Missing, low-confidence and stale values contribute neither a zero nor a weight to that calculation. The coverage percentage and unavailable dimensions remain visible: partial scores can differ because they use different evidence.
Results are ordered by unrounded-to-integer scores, then country slug for ties. No country receives a commercial bonus. Historical price growth is not a return forecast. Default growth and buying-cost weights are zero because they require a deliberate choice and, in some countries, fuller evidence.
EU-only filters country membership. Financing, residence interests and rental strategy add relevant checks, without inventing a legal-risk, liquidity or bank-eligibility score. Nationality is not collected to imply a legal eligibility decision. Tax sensitivity, land restrictions and currency exposure must be assessed for the actual investor and property.
Yields, costs and budgets
Gross rental yield is annualised rent divided by a purchase-price benchmark, multiplied by 100. Read the original methodology: the units and locations sampled vary. It excludes vacancy, repairs, management, service charges, insurance, tax and debt payments. We do not invent net yields.
Georgia’s price and gross-yield observations are read from the existing SafeBuy Georgia Property Price Index, with its permanent monthly citation. The adapter uses Tbilisi and identifies it as a city sample, not a nationwide average. It never copies a price into an independent international table that can drift from the index.
Purchase-tax rules are not automatically total acquisition costs. Total-cost scores require a sourced all-in estimate; otherwise the facts remain in the country profile. The budget illustration divides your budget by the cited price benchmark. It excludes taxes and purchase costs, does not reserve cash, and does not assert that a property of that size is available. Budget changes the floor-area illustration; it does not mechanically change country scores.
Currency and freshness
Original currencies are preserved. EUR is the comparison basis for price scoring. GBP and USD budgets use the dated FX snapshot: European Central Bank reference exchange rates, 2026-09-04. Rates are currency units per EUR. Conversion is value ÷ source-currency rate × destination-currency rate. These are research reference rates, not live transaction quotes.
Cross-currency conversion stops after 14 days without a rate update; same-currency values remain usable. A missing currency rate stays unavailable. No default 1:1 conversion is used.
Every metric has an explicit age limit based on the older of its observation date and review date. At 80% of that limit it is marked “review due soon”; beyond the limit it is excluded from scores and data-led metric rankings. Historical figures remain visible in profiles, clearly marked stale. Rechecking an old observation cannot make it new. Rules are flagged using their review dates and require human review.
Editorial process and commercial relationship
Authorised SafeBuy editors maintain the source-linked dataset through the existing admin system. Publication validates references, fields, dates and scoring configuration. Updates retain a revision history. New country or comparison pages need useful evidence and a distinct research purpose; we do not publish every country pair or arbitrary budget variant.
SafeBuy’s commercial services focus on Georgia. Listings, valuation, registry checks and a call with David Lee are offered when you want to explore that market. No phone number is required to view the research. You can choose whether your callback request includes your shortlist and budget. Read how SafeBuy is paid.
To request a correction, use the contact form and include the country, observation and primary source. French editorial content can be reviewed separately using language-independent country IDs and metric fields; no unreviewed translated pages are automatically indexed.