About & method

Who writes this

Catlytic is written and built by Yu Han, a business economics student at the University of California, Irvine, who spends more time inside 10-Ks than is strictly reasonable.

How this started

It began the way it usually does — trying to work out whether a company I already owned was expensive. Every source I checked gave a different answer, and not one of them showed how it got there. Screeners disagreed with each other. Articles quoted figures with no indication of where they came from. So I started opening the filings myself.

Doing that by hand does not scale past a handful of companies, so I started writing code to do it faster: pull the document from EDGAR, find the figure, check it against the text, compute the valuation. That code is what this site runs on, and it is still being improved every week.

Building it taught me something I did not expect. The standardised data most tools are built on is wrong more often than anyone admits. In one case it reported a company's total borrowings as $6M when the debt footnote in the same 10-K said $29,244M — a factor of 4,874, and no error message anywhere. Once you have seen that, you stop taking any number on faith, including your own.

What you can expect here

Every figure links to the filing it came from. Not to a database, not to another article — to the document.
The calculation is shown, not asserted. If you disagree with the method, you can see exactly where.
Corrections stay published. When something here turns out to be wrong, it is fixed and the change is recorded on the corrections log, with a date. Nothing is quietly deleted.
Failed tests are published too. The valuation model is checked against historical outcomes, and the screening rules that looked promising and then did not hold up are written down rather than dropped.

That last one matters more than it sounds. It is easy to look right if you only publish the experiments that worked.

What this site is not

I am a student, not a licensed adviser, and nothing here is a recommendation to buy or sell anything. What is on offer is the arithmetic and the source documents — not a verdict, and certainly not a price target.

In practice I think that is the more useful arrangement. A research shop asks you to trust its judgement. This site asks you to check its work, which is a thing you can actually do.

Contact: [email protected]. Corrections and challenges to any figure published here are genuinely welcome, and they get answered.

Where the data comes from

Every financial figure on this site originates in a document filed with the U.S. Securities and Exchange Commission — 10-K, 10-Q, 8-K, or proxy statement — retrieved from EDGAR. Market prices come from public market data feeds and are labelled with the date they were taken.

Standardised data (XBRL) is used to locate figures, but it is not trusted on its own. It has been wrong here in ways that would change a conclusion. In one case a company's total borrowings were tagged as $6M when the debt footnote in the same 10-K said $29,244M. Wherever a number matters to the argument, it is read out of the document text and reconciled against the tagged value.

What gets published

Every article has to pass the same test before it goes up:

Could a reader get the same answer by asking an AI assistant? If yes, it isn't published.
Does every figure trace to a document that is linked on the page?
Is the calculation shown, so a reader can disagree with the method rather than the result?
Is the date the data was taken stated on the page?
Is there a section on what would make the analysis wrong?

How the writing is produced

Data collection, reconciliation, table generation and first drafts are automated. A person reads every article in full, checks the figures against the source documents, and approves publication. Nothing is published automatically.

This matters because it is the difference between a research pipeline and a content farm. Volume is capped deliberately — a small number of articles that hold up, rather than a large number that don't.

Corrections

When something here is wrong, the article is fixed and a dated correction is added to the bottom of it. Nothing is quietly deleted. Every correction ever issued is listed on the corrections log.

What this site will not do

It will not tell you what to buy or sell. It is not investment advice.
It will not publish a price target dressed up as a fact.
It will not present a figure it has not seen in a filing.
It will not accept payment to cover a company. Affiliate links, where used, are disclosed at the top of the article.

Limits you should know about

Coverage is U.S.-listed companies. Filings from other jurisdictions are not treated as equivalent.
Valuation output is a model result, not a forecast. Models embed assumptions, and the assumptions are published with the result.
Backtests measure what a rule would have done, not what it will do.
Figures can be restated by the issuer after publication. The "data as of" date on each page tells you what was known at the time.

Disclaimer

Catlytic publishes research and data for informational purposes only. It is not a registered investment adviser, broker-dealer, or financial planner, and nothing on this site is a recommendation or solicitation to buy or sell any security. You are responsible for your own investment decisions. Catlytic is not affiliated with or endorsed by the U.S. Securities and Exchange Commission.