Blog.

Articles on statistics, machine learning and LegalTech

When the Model Is Right but the Decision Is Wrong

A machine learning model can tell you with high accuracy which patient will die, and at the same time lie to you about the medication's effect. That's the difference between predicting and explaining, and when that distinction is blurred, statistics can become dangerous.

Jun 16, 202615 min read

EstadísticaAnálisis de datosCalidad

Beyond the Mean: Why Extremes Are the Real Early Warning Signal in Data

Traditional statistics focuses on the mean, but in practice critical events — failures, opportunities, risks — happen in the tails. This post explains why monitoring extremes is as important as knowing the average, and how to do it with statistical rigor.

May 19, 202610 min read

BayesProbabilidadInferencia

Bayesian Statistics: Reliability Based on Probability, Interpretable Inference, and Design‑Driven Updating

The Bayesian approach turns uncertainty into clear decisions: from generic priors to forecasts that update with every new data point, without losing coherence.

Apr 22, 202614 min read

EstadísticaLegaltechPredicción

Statistics and Law: The Silent Legaltech Revolution

Law has always been seen as an art of qualitative argumentation, but the emergence of statistics is changing the rules of the game. This post explains how quantitative models improve judicial prediction, legal risk management and damage assessment.

Apr 17, 202611 min read

BayesianoStanbrms

Messias: Bayesian Football Prediction with Three Models That Talk to Each Other

The key to Messias is not a single prediction model, but the combination of three specialized Bayesian models: Double Poisson for goals, Bradley-Terry-Davidson for match results, and Hierarchical Negative Binomial for corners. Here is the architecture and the statistical reasoning behind it.

Feb 28, 202612 min read

Modelos MixtosShinyR

MLM Studio: An Integrated, Visual, and Interactive Workflow for Longitudinal Analysis

Repeated measures analysis demands sequential decisions: comparing means across groups, selecting random effects structure, and evaluating covariance between intercept and slope. MLM Studio organizes that workflow with interactive graphics, interpretive cards, and model comparison tools, without writing code.

Feb 20, 202612 min read

BayesianoRShiny

La Tienda: Wholesale Price Intelligence with Bayesian Updating and Temporal Focus

The retail merchant needs to decide today how much to buy and at what price, but the key lies in what will happen in the coming days. La Tienda combines Bayesian updating, seasonal decomposition and trend detection to anticipate movements, adjust inventories and avoid cost overruns from late decisions.

Feb 11, 20269 min read

LegalTechBayesianoNLP

SpinLoyalty: Objective Measurement of Legal Risk Through Bayesian Statistics

Unfair competition in trademark registration is often assessed using subjective criteria. SpinLoyalty proposes a statistical approach that turns appreciation into measurement, calibrates discrepancies between examiners, and standardizes the notion of differentiation; the result is a quantifiable and updatable probability.

Dec 5, 20259 min read

LegalTechColombiaIA

LegalTech in Colombia: State of the Art and Foundations of a Comprehensive Course

Colombia is at a pivotal moment in the digital transformation of its legal sector. Laws such as 527/1999, Decree 806/2020, and Law 1581 form the regulatory foundation. Upon this foundation, a LegalTech ecosystem is being built that legal professionals can no longer overlook.

Mar 11, 20258 min read