Causa
Subscribe
Closes30 Oct
S&P 5007,580−1.82%Nasdaq25,437−2.86%US 10Y4.98%−7 bp29 OctDollar102.04−0.15%29 OctBrent futures$117.64+2.4%WTI futures$102.00−0.76%29 OctGold futures$4,031−0.5%29 OctLiveBitcoin$77,381−0.59%Ether$2,243−0.27%SolanaOfficial closes, each dated, from the record the Scorecard grades on: exchange data via Yahoo Finance, US yields from FRED. Crypto live from CoinGecko.
Publishing principles

How we decide what caused a move

The rules every Cause File follows, written so that you can check them.

One move, one file

We start with moves that are large relative to the asset's own recent behaviour, not just large in percent. A 2% day in bitcoin is ordinary. A 2% day in the S&P 500 is not. Each file explains one move.

The chain

We write the explanation as a chain of steps from an event to a price. Each step has to be supported on its own. If we cannot support a step, we say so in the grade.

Evidence grades

A, strong. The cause is reported by at least two independent sources, the timing fits, and the direction and size of the move match the mechanism.

B, plausible. A strong case with a clear mechanism, but other causes overlap on the same day, or the support rests mainly on one source.

C, contested. The best explanation available, with live alternatives that we name in the file. We publish C grades because hiding uncertainty is worse than admitting it.

Sources

Every paragraph is numbered to its sources. We prefer primary sources such as official statements, company results and exchange data, then established reporting, then market data providers. We link to every source and never cite one we have not read.

Falsifiers

Every file states the condition that would prove it wrong and a deadline. Most conditions are prices, written precisely enough for a script to test. Some are events, such as an official production figure, and an editor resolves those with the evidence linked.

The Scorecard

After each session, closing prices are added to the record and every open condition is tested. Intraday prices never count. Calls are never edited after publication. Results are public on the Scorecard.

The Proof Ledger

A promise not to edit calls is only worth something if you can check it. So every call is sealed when it is published: its exact words, its test and its deadline get a SHA-256 fingerprint, and each fingerprint includes the one before it, forming a chain. Change one character of any call, or remove one, and every fingerprint after it breaks. The newest link is submitted to the public OpenTimestamps calendars, which write it into the Bitcoin blockchain, so the date of the record can be proven without trusting us. Verify the ledger in your browser.

Market data

Oil is front-month futures: Brent and WTI. Stocks are the official index closes for the S&P 500 and the Nasdaq Composite. The 10-year yield is the US Treasury constant maturity series published by the St. Louis Fed (FRED). Bitcoin and ether come from CoinGecko. Each metric keeps one primary feed, and a backup feed fills a day only when the primary misses it.

These prices come from third parties and remain subject to their providers' terms. They are not covered by our licences, and we do not sell them. Every chart and the market tape name their source.

Futures rolls

Oil futures expire every month. When the contract a call was written on expires, our feed moves to the next month, which can trade several dollars away. The call keeps its published words, and its line moves by the gap between the two contracts' settlements on the expiry day, so the call stays exactly as far from the market as it was when we wrote it. Each move is shown on the call itself.

  • Brent, 30 October 2026. The December 2026 contract expired at $117.64; January 2027 settled at $113.43. Open lines moved down by $4.21. Settlement source

Gold calls are written against the spot price, while our automatic feed records COMEX gold futures, which usually sit a few dollars above spot. So the machine never rules a gold call. When the feed says a gold line has been crossed, or a gold deadline has passed, the call is marked for desk review and the desk checks the spot close before the ruling is posted.

Open data

Every edition, Cause File and call is published as JSON, described by a versioned schema at /data/schema.json. Identifiers never change: a call keeps its F-number for life, and a file that moves keeps a permanent redirect from its old address. Corrections are added as dated records; nothing published is rewritten.

Corrections

If we get a fact wrong, we fix it, date the change and log it on the corrections page. A wrong explanation is not a correction. It is a triggered call, and it stays on the record.

On 27 October 2026 we edited the wording of Editions 001 to 003 into our house voice. Facts, figures, sources, grades and every falsifier are unchanged.

Independence

Causa takes no payment to cover or avoid any company, coin or exchange. Partner links are labelled, and they never affect what a Cause File says.

Questions

What is a Cause File?

A short, sourced explanation of one market move. It traces the chain from an event to a price, numbers every source, grades our confidence from A to C, and ends with a falsifier: the condition that would prove the explanation wrong, with a deadline.

What is a falsifier?

A measurable condition, with a deadline, that would show an explanation was wrong. For example: gold closes below $3,900 on a day the Nasdaq closes above 26,000, by 13 November. Most falsifiers are tested automatically against closing prices.

Does Causa tell me what to buy?

No. We explain why prices moved and publish the evidence. We do not make recommendations, and nothing on the site is investment advice.

Who writes Causa?

Causa is an independent markets desk. We combine AI tools with editorial judgment to gather data, draft and verify each Cause File. Every figure is checked against its source. Every test is set before the market can answer it. Every edition is reviewed and approved by an editor before publication, and the desk answers publicly for every correction.

Why publish a way to be wrong?

An explanation that cannot fail cannot be trusted. A falsifier turns each explanation into a prediction that the market can test, and the Scorecard keeps the result in public.

Why do intraday prices not count?

Prices can spike and reverse within a session. Resolving calls on closing prices only stops a brief move from deciding a test.

What happens when a call is triggered?

It is marked as triggered on the Scorecard with the date and the closing values that met the condition. The original file stays online, unchanged, with a note pointing to the result.

Can I use your data?

Yes, for research, teaching and personal projects. Edition and Scorecard data are published as JSON under CC BY-NC 4.0, so credit Causa and link to the source page. Articles, including every Cause File, are under CC BY 4.0. A commercial data licence will be part of Causa Research when it opens; reserve a place on the Research page to ask about one now.

Will Causa charge?

The Brief, every Cause File and the Scorecard stay free. A paid tier, Causa Research, will add alerts, search across the whole record, data exports and team seats. The waitlist is open.

How should I cite Causa?

Credit the desk and link to the Cause File, for example: Causa Desk. "Gold breaks $4,000, and our explanation is failing its test." Causa, 23 October 2026, followed by the link. Every Cause File has a citation ready to copy.