Transparent econometric models, verified data, confidence index visible on every advanced analysis.
The Brazilian real estate market makes decisions based on averages, comparables, and intuition. These three pillars share one limitation: they don't isolate causes. Lycaon Data built its platform on econometrics, a discipline whose function is precisely this: extract, from a mass of data, the causal impact of each variable on an outcome.
Every statistical estimate comes with a margin of uncertainty. Most tools hide it. We make it public: each advanced analysis displays a confidence index from 0 to 100, based on the model's goodness of fit (R²) and sample size. You know, before making a decision, how much you can rely on it.
An econometric model doesn't replace the developer's professional judgment. It informs it. Lycaon Data acknowledges and documents the limits of its models: zones with insufficient samples are flagged, periods of volatility are identified, and unobserved factors are recognized. This transparency about what we know, and what we don't, is the condition for your informed decision.
Our data goes through a rigorous process of verification, deduplication and outlier treatment before entering the econometric models.
Result: for every analysis, you know exactly where the figure comes from, how it was calculated, and the level of confidence you can assign to it.
Principle: A property's price is decomposed based on its characteristics (area, typology, features, location). The model isolates the pure impact of each characteristic, controlling for all others.
Principle: A property's price partially depends on its neighbors' prices. The model quantifies this dependence with a coefficient between 0 (zero neighborhood effect) and 0.25 (maximum neighborhood effect).
Principle: From dozens of raw indicators, the model extracts 6 synthetic dimensions (health, safety, wealth, education, employment diversity, housing) that summarize the quality of life of a zone.
Each advanced analysis (Project Optimization module) comes with a confidence index from 0 to 100.
Robust results, applicable with full confidence in your acquisition or product definition decisions.
Reliable trend, but cross-check with additional sources before locking in a critical decision.
If the model's goodness of fit is insufficient, we prefer to show nothing rather than an unreliable analysis.
Every advanced analysis displays the confidence index and sample size. No exceptions.
The limits of our models are published in the methodology: low-sample zones, periods of volatility, unobserved factors.
Listings are updated monthly. Models are re-estimated every quarter to capture market evolution.
Our team responds to any methodological question sent by email. Without delay, without filters.
