Where evidence crosses over.

Umbral is an independent, open-source data lab. We build data products, causal analysis, and monitoring tools for the public interest — everything reproducible, everything documented.

umbral (Spanish) = threshold: the point where evidence becomes significant

What we are building

Every project publishes its code under the MIT license and its cleaned data under CC BY 4.0. Public data goes back to the public, improved.

In development

desaparecidosmx

A pipeline that turns Mexico's RNPDNO registry of disappeared persons into clean, monthly, per-state CSVs disaggregated by category, sex, and municipality — every raw response cached and reconciled against the official dashboard's totals. A Streamlit dashboard runs on the processed data.

Source: RNPDNO · code MIT · data CC BY 4.0

In development

observatorio-delictivo-mx

An open, reproducible pipeline for crime incidence, victimization, and perception of safety in Mexico. Immutable vintage-tagged snapshots, rates per 100,000 with CONAPO denominators, and SESNSP–INEGI homicide reconciliation.

Sources: SESNSP · INEGI · CONAPO · code MIT · data CC BY 4.0

Every published figure can be rebuilt from its source

Every dataset we publish ships with full provenance: origin URL, download date, license, and known caveats. Every number traces to a scripted transform — never a manual edit. If a figure cannot be reproduced, it is not published.

Honesty and interpretability standards

  1. 01
    The title is the claim

    Every chart is titled with its finding, and the finding must be defensible from the data shown. No headline the evidence does not support.

  2. 02
    Deliberate causal language

    "Associated with" and "correlates" for descriptive work. "Effect" and "reduces" only with an explicit identification strategy, its key assumption stated next to the estimate.

  3. 03
    Uncertainty is always visible

    Intervals on projections, confidence intervals on estimates, n and denominator on rates. A point estimate without an interval is incomplete.

  4. 04
    Honest axes, visible gaps

    The y-axis starts at zero for bars; any truncation is annotated. Missing or underreported data is disclosed — never silently dropped or imputed.

  5. 05
    Human dignity

    Sensitive topics are handled with dignity: people are counted, never turned into spectacle.

Who we are

Umbral is built by two people. Everything we publish carries our names and can be audited in the code.

Jay Ballesteros

Jay Ballesteros

Co-founder

Lorem ipsum dolor sit amet, consectetur adipiscing elit. Sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco.

Erick Ballesteros

Co-founder

Lorem ipsum dolor sit amet, consectetur adipiscing elit. Sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco.