Analysis Methodology

1. Principles

• Transparency — every metric is paired with the inputs and assumptions used.
• Conservatism — when in doubt, we choose the assumption that produces the lower expected return.
• Back-testing — outputs are tested against subsequent realised data on a quarterly cadence.
• Versioning — every change to the model is versioned and dated.

2. Data sources

3. Gross rental yield

Gross yield is the simplest income metric:

Gross yield = Annual rent ÷ Purchase price

It ignores costs and is provided only as a comparability anchor against widely-quoted figures.

4. Net rental yield

We compute net yield using:

Net yield = (Annual rent − Operating costs − Vacancy allowance) ÷ All-in acquisition cost

Default assumption ranges (each adjustable per listing):

• Property management: 10–14% of collected rent
• Maintenance and repairs: 4–6% of collected rent
• Insurance: 0.10–0.20% of property value per year
• Municipal and refuse charges: actual where known, otherwise band estimate
• Vacancy allowance: 4–10% depending on long-let vs short-let and location
• All-in cost: purchase price + transfer fees / VAT + legal fees + furnishing reserve

5. IRR projection

We model a 10-year hold and compute the internal rate of return on equity that solves:

Σ Cashflow_t ÷ (1 + IRR)^t = 0, t = 0 … 10

Cashflow assumptions:

• Year 0: equity outflow = all-in acquisition cost − financing drawn
• Years 1–10: net rental cashflow, with rents grown at long-run Cyprus CPI
• Year 10: exit value = current price × (1 + district price-index growth)¹⁰, less selling costs
• Financing (if modelled): amortising loan at the rate and tenor specified per listing

IRR is shown as an annualised figure. We do not apply a discount rate; the output is the project IRR on equity.

6. Price benchmarks (€/m²)

Per-square-metre benchmarks are computed using covered internal area as the denominator where reliably available. Where listings publish only total or built-up area, we apply a documented normalisation factor and flag the listing accordingly.

7. Comparable selection

• Geographic radius scaled to neighbourhood density (typically 0.5–2 km).
• Property type matched (apartment, house, plot).
• Build year window of ±10 years where sample allows.
• Minimum 8 comparable transactions or active listings; otherwise the listing is flagged "low confidence".
• Statistical outliers (top/bottom 5% by €/m²) are winsorised, not deleted.

8. Confidence bands

Each headline metric is presented with a confidence band derived from the standard deviation of the comparable set, scaled by sample size. Narrow bands indicate a deep, homogeneous comparable pool; wide bands indicate thinner data and warrant additional caution.

9. Known limitations

• Off-plan vs resale comparability is imperfect; off-plan units may carry developer-specific premiums.
• Premium districts and unique waterfront assets often have thin samples — confidence bands widen accordingly.
• Short-let revenue depends on management quality and platform algorithms that we cannot fully observe.
• Historical price-index growth is not a forecast; we use it as a baseline expectation.
• Tax treatment is generic; individual circumstances are not modelled.

10. Back-testing procedure

Each quarter we compare prior model outputs against newly-realised transaction and rental data, compute bias and dispersion, and adjust default assumption ranges where systematic deviations appear. Material changes are noted in the model changelog.

11. Model versioning

The current model is v2026.04. Changes are versioned by year and month of release and the prior version is retained internally for reproducibility of historical reports.

12. Disclaimer

All metrics and projections on the Service are estimates produced by statistical models. They are not offers, valuations, or guarantees. Please read our Risk Disclosure and obtain independent professional advice before making any investment decision.