Who builds this, and why you can check every number on it

SteadyShares is built and maintained by Warrick Campbell, a software engineer, working independently. There is no research desk, no analyst team and no outside investor. That is worth saying plainly because it cuts both ways: nobody here is paid to like a stock, and there is also nobody to blame but the person who wrote the code. Questions, corrections and complaints all reach the same inbox: hello@steadyshares.com.

The site takes no payment for coverage, placement or ratings, runs no advertising, and has no affiliate relationships with brokers or fund managers. If that ever changes it will be disclosed on this page before it happens, not after.

Every figure on the site traces to one of these, and nothing is filled in by hand:

• Company filings and market data. Prices, fundamentals, analyst estimates and dividend records come from public market data feeds and the companies' own reported figures.
• SEC filings. Institutional holdings come from Form 13F, insider activity from Form 4, and company identity from the SEC's own registrant file.
• STOCK Act disclosures. Congressional and executive-branch trades come from the House Clerk, the Senate eFD system and the Office of Government Ethics.
• Our own models. The rating, moat score and fair value are computed here from the inputs above. They are labelled as ours everywhere they appear, and the methodology page gives the formula, the assumptions and the known failure modes.

Data has a timestamp wherever it is shown, because an undated figure in finance is an unusable one. A 13F is a photograph up to 45 days old and says so on the page. A quote that has not refreshed is marked stale rather than passed off as live.

Four rules, applied to every page, every model output and every article:

• Ship the filed number or say we do not have it. A gap is never filled with a plausible-looking figure. Where a metric is missing it renders as nothing, never as zero, and never as an estimate dressed up as a fact.
• Every score states what it rests on. Derived scores carry a data quality figure, so a rating built on two inputs cannot pass for one built on twelve. Below a quality floor, the product refuses to state a verdict at all.
• Every claim states its own failure mode. A screen prints the criteria it used and when that approach stops working. A pick without a bear case is an advert, so the risk is stated in the same breath as the case.
• Nothing here is advice. SteadyShares is not a financial adviser and does not know your circumstances. It is a research tool: a place to start reading, never a place to stop.

Data products get things wrong. The question that matters is what happens next, so this is the commitment:

• Report an error to hello@steadyshares.com with the page and what looks wrong. You will get a reply.
• A figure that is wrong because the source is wrong gets said so, on the page, rather than quietly adjusted to look better. We publish the filed record, including its oddities.
• A figure that is wrong because our code was wrong gets fixed at the source, and the fix is noted in the changelog. Material errors that could have affected a reader's decision are called out there specifically, not buried in a list of improvements.
• Known limits are published rather than hidden. The methodology page carries a standing list of what the free data cannot tell you, and the models say where they break.

Security researchers can find the disclosure contact at /.well-known/security.txt. Reports are read by a human and acknowledged.

What the site stores about you, and what it does not, is set out in the privacy policy. Session replay is off unless you explicitly allow it, and everything typed is masked when it is on.