About Macro or Noise

Most financial commentary explains what should happen. "Hot inflation is bad for bonds." "A strong jobs report sends yields up." "The Fed's preferred inflation gauge moves the dollar." These stories are repeated so often they feel like facts. Macro or Noise exists to check them — to ask, for each one, whether the data actually agrees.

The name is the whole idea. When an economic release lands and a market moves, is that a real, repeatable macro effect — or just noise that happened to line up with the headline? Our job is to tell the two apart, honestly, and publish the answer either way.

What we do

We run event studies: for a given release (CPI, the jobs report, a Fed decision, and so on), we gather every occurrence, measure how a market moved around it, and compare that against a baseline of ordinary days — with a significance test on every claim. The full method is explained in plain language in what is an event study?.

Two questions drive every page:

1. Does the market move more than usual? (size / volatility)
2. Can you predict which way? (direction)

The recurring answer is instructive: several releases reliably make markets move more, but almost none give you a dependable direction. That gap — between "volatile" and "predictable" — is the most useful thing our tests reveal, and it is exactly what most commentary glosses over.

How we work

• Public data only. Every figure comes from U.S. government sources in the public domain: the Bureau of Labor Statistics (BLS), the Bureau of Economic Analysis (BEA), the Census Bureau, the Department of Labor, the Energy Information Administration (EIA), and the Federal Reserve (via FRED). We deliberately avoid proprietary data feeds and copyrighted indices, so every number we publish is one we are free to compute and share.
• A baseline, every time. We never report a move in isolation. Each result is measured against normal days, so you can see whether an effect is real or just noise.
• Significance and multiple-testing correction. Every claim carries a p-value, and results are corrected for the fact that we test many releases at once — so a pattern that only looks significant by chance is flagged, not headlined. See the full results grid.
• Honest verdicts. When the data says a popular belief doesn't hold up, we say so — clearly. A confident "no effect" is a result we are just as happy to publish as a dramatic one.

Who runs this site

Macro or Noise is built and maintained by a single independent operator — one person who writes the code, runs the statistics, and writes every page. There is no staff, no sponsor, and no financial-industry affiliation.

Rather than asking you to trust a byline, the site is built so that you can check the work yourself. Every figure is produced by the site's own code from named public-domain series, and every page lists the exact series it uses. The full method and the complete results grid are published — including the tests that found nothing.

Questions and corrections are welcome at [email protected]. Corrections in particular: two real errors have already been found and fixed this way, and both are described on the contact page rather than quietly patched.

What we are not

We are not affiliated with any government agency, and nothing here is financial or investment advice. We do not forecast markets or tell you what to buy. We describe what a defined set of historical releases actually did, with the caveats that come with any historical study — results can change with a different sample, period, or definition. Please read our Disclaimer.

Where to start

• What is an event study? — the method, in plain language.
• Which U.S. data releases move the Treasury curve? — a map of what moves and what doesn't.
• Glossary — definitions for the releases, markets, and statistics we use.

The goal isn't to be right. It's to measure.