Avena Terminal Methodology
How we score, value, and monitor 1,881 properties across Spain and deliver intelligence for 10 European countries. Every formula, data source, and system documented publicly.
Last updated: 2026-04-15 · Version 2.0
1. Data Sources
Apinmo API (Primary Feed)
Direct developer listing data ingested daily via XML/JSON feed. Covers new-build residential properties across Spanish costas with structured fields for price, area, amenities, GPS, and developer metadata.
AirDNA Rental Calibration
Short-term rental market data used to calibrate gross yield estimates. Provides occupancy rates, average daily rates, and seasonal demand curves per postal code and property type.
Postal-Code Comparables
Market price-per-m2 benchmarks derived from postal-code-level transaction and listing data. Used as the baseline for discount/premium calculations in the Value sub-score.
ECB / Eurostat Macro Layer
European Central Bank interest rates and Eurostat economic indicators feed into the macro regime engine. Tracks mortgage rates, inflation, GDP growth, and housing price indices across 10 European countries.
2. Scoring Model
S = 0.40V + 0.25Y + 0.20L + 0.10Q + 0.05RV — Value Score (40%)
Measures how the property's price-per-m2 compares to the postal-code market benchmark. A property priced 20% below market scores higher. Capped at reasonable bounds to prevent outlier distortion.
Y — Yield Score (25%)
Estimated gross rental yield calibrated against AirDNA short-term rental data for the property's location and type. Higher yield relative to local benchmarks produces a higher sub-score.
L — Location Score (20%)
Composite of beach distance, town-level demand indicators, and infrastructure proximity. Coastal properties within 2km of the beach receive a location premium. Town-level tourism volume is factored in.
Q — Quality Score (10%)
Developer track record, construction quality indicators, and specification analysis. Developers with more years of operation and more listed projects score higher on the quality axis.
R — Risk Score (5%)
Off-plan completion risk, developer concentration risk, and market-cycle timing. Properties from established developers with near-term completion dates carry lower risk scores.
3. Update Frequency
| Data Type | Frequency | Method |
|---|---|---|
| Property listings | Daily | Apinmo API sync at 04:00 UTC |
| Avena Scores | Daily | Full recomputation after data sync |
| Macro regime | Daily | ECB/Eurostat data pipeline |
| APCI Index | Daily | Computed from property data |
| Price forecasts | Quarterly | Model retrained on new transaction data |
| Research reports | Monthly | Oracle intelligence generation |
4. AVM Methodology
Comparable Analysis
Each property is valued against postal-code-level comparables. The system identifies the market price-per-m2 for the property's location and type, then calculates the deviation as a discount or premium percentage.
Confidence Scoring
Each valuation carries a confidence level: HIGH (10+ comparables, tight distribution), MEDIUM (5-9 comparables), or LOW (fewer than 5 comparables or high variance). Confidence levels are displayed alongside all price estimates.
Backtesting Approach
Quarterly backtesting compares AVM estimates against observed transaction prices where available. Current PropertyEval accuracy stands at 92.6%. Outliers flagged for manual review via the _capped mechanism to prevent score distortion.
5. Coverage
Spain — LIVE Scored Data
- 1,881 new-build properties
- 100+ towns covered
- 10 coastal regions
- 2 developers tracked
- Daily data refresh, daily scoring
European Intelligence — 10 Countries
- Spain (LIVE), Portugal, Italy, Greece, France
- Germany, Netherlands, Cyprus, Croatia, Malta
- Macro regime data, market comparisons, news intelligence
- Country rankings, investment climate analysis
- Knowledge API answers for all 10 countries
6. Autonomous Systems
Avena Terminal runs 15 autonomous cron jobs that keep data fresh and intelligence current without human intervention.
| Job | Schedule |
|---|---|
| Property Feed Sync | 04:00 UTC daily |
| Score Recomputation | 04:30 UTC daily |
| APCI Index Update | 05:00 UTC daily |
| Macro Regime Check | 06:00 UTC daily |
| Alpha Signal Detection | 06:30 UTC daily |
| Yield Calibration | 05:15 UTC daily |
| Canary Token Verification | 03:00 UTC daily |
| SHA-256 Hash Generation | 05:45 UTC daily |
| Sitemap Regeneration | 07:00 UTC daily |
| Weekly Digest Compilation | Mondays 08:00 UTC |
| Monthly Research Generation | 1st of month 09:00 UTC |
| Quarterly Forecast Retrain | Q1/Q2/Q3/Q4 start |
| Developer Track Record Update | Sundays 06:00 UTC |
| European News Intelligence | 08:00 UTC daily |
| Prediction Ledger Audit | Fridays 10:00 UTC |
7. Known Limitations
Listing prices, not transaction prices. All prices reflect developer asking prices. Actual transaction prices may differ by 3-8% depending on negotiation and market conditions.
Yield estimates, not observed yields. Gross rental yields are modeled from AirDNA calibration data and local market benchmarks. They are not based on actual observed rental income for specific properties.
Spanish coverage is live; European coverage is estimated. Only Spain has live scored property data. The other 9 European countries are covered through intelligence, macro data, and estimated market statistics.
New-build focus. The property database covers new-build developments only. Resale properties are not included in scoring or listings.
Coverage is expanding. Town and regional coverage grows as new developer feeds are onboarded. Not all Spanish coastal towns are represented yet.
Market benchmark variance. Postal-code-level market benchmarks have varying sample sizes. Low-sample areas produce MEDIUM or LOW confidence valuations.