About Prism Data Lab

Prism Data Lab is a site about financial data, the analysis of it, and the honest state of what language models and other AI systems can do with it. It is maintained by one person, with no fund and no product behind it; the articles are produced from the sources each one names, under that person's direction, and how this site is made has the details.

Editor

C. B. Zakarian (a pen name) is the owner and editor of Prism Data Lab. C. B. Zakarian holds a Ph.D. in Information Systems, with a focus on data science, from the University of North Carolina at Greensboro (2016), and is an assistant professor.

Since October 6, 2026, every new piece on Prism Data Lab is read and approved by C. B. Zakarian before it is published. Pages published earlier are being reviewed in weekly batches, and each reviewed page shows the date of its review.

What is published

Data analysis articles start with a question, pull a public dataset, and answer it, with the dataset and the code that produced every figure shipped alongside. If a chart or a table is on the page, the CSV and the script that made it are one click away.

LLMs in finance articles look at where language models are actually being used in financial work: reading filings, summarising research, extracting data, screening, risk narratives. They report what a paper measured or what a filing discloses, cited by arXiv identifier or SEC accession number, and they are candid about failure modes, because in finance a confident wrong number is worse than no number.

Data sources articles are practical: how to pull a series from FRED, a filing from SEC EDGAR, a table from the Bureau of Labor Statistics, how to parse it, and how not to misread it.

Methods articles cover time series, risk measures, backtesting discipline, and the statistics that keep an analysis from fooling its author.

Briefs are short, dated readings of a single release, filing, or paper.

What this site refuses to do

Nothing here is investment, tax, accounting, or legal advice, and no article recommends a position. No forecasts dressed as facts. No figures from memory: every number is cited to a series identifier, an accession number, or a table, with the date it was retrieved, because data gets revised. No exclamation marks.

How the numbers stay honest

The code is the audit trail. Every analysis article's listing runs from the raw download to the printed figure, and if a figure cannot be reproduced that way it does not appear. Corrections are welcome and acted on: contact.