Methodology

This page documents exactly how our two signatures work, so you never have to take our word for either one. If what you read here does not match what you see on the site, that is a bug — tell us.

The First-Principles Take

The First-Principles Take is our structured analysis format, applied to analysis pieces and to news stories where a genuine “so what” applies. It always follows the same four-part structure:

1. Claim. A single, precise, falsifiable statement of what is being examined. Not a mood (“bullish”), a claim (“X caused Y”).
2. Evidence. What we are reasoning from, itemised with its source. On-chain data and primary documents outrank secondary reporting; secondary reporting outranks anonymous commentary, which rarely appears at all. Estimates are labelled as estimates.
3. Reasoning. The steps connecting the evidence to the claim, written so a reader who knows the subject can identify exactly which step they disagree with, if any.
4. What it does not prove. The alternative explanations that remain live, the data that would change the conclusion, and the inferences the evidence cannot support. This is not a disclaimer appended at the end — it is the section we spend the most editing time on.

Every Take is written by a person, carries the tag “reasoned, not hyped,” and is explicitly not financial advice. We do not generate Take prose with a language model, and a Take is not published without a second person checking it against its own cited evidence.

The Bitcoin Signal — sources and cadence

The Signal board is assembled from public, keyless sources. Nothing on it requires an account or an API key, and every metric names its source directly.

Every value is cached server-side. If a source is briefly unavailable, the board serves the last known-good value with its original timestamp rather than a placeholder, and never fabricates a number. If no good value exists yet, the metric is omitted entirely.

What we deliberately do not show

Realised cap, HODL-wave derivatives, supply-in-profit bands and exchange net-flow estimates all depend on entity-clustering heuristics — educated guesses about which addresses share an owner. Those heuristics differ between data providers and are rarely published in full. We discuss these metrics in our analysis, where assumptions can be stated in the open, but we do not put a number for them on a live board we could not defend to the decimal place.

Forecast policy

We rarely make forward-looking statements about price. When we do, they are framed as scenarios with explicit assumptions — “if X holds, Y follows” — never as expectations, targets, or certainties. We do not publish price-prediction headlines.

Corrections policy

Substantive errors are corrected and logged publicly, with a date, at our corrections page. We do not silently edit a published claim.