Methodology
How we work
We read the day’s crypto news and write what could follow from it — the case for it going well, the case for it going badly, and which one we find more convincing. That is the whole product. Here is how it is made, and what we deliberately refuse to do.
1. We follow five projects, chosen by the market
Once a day we take the five largest projects by 24-hour trading volume — that list is data, not taste, so the beat follows where the money actually is rather than where our habits are. Alongside them we cover what moves everything regardless of ticker: regulation, ETFs, the Fed, and security failures. Everything else is somebody else's beat. Five subjects covered properly beats ninety-five covered thinly.
2. We start from the original reporting
Every analysis begins with primary reporting from established outlets, and we link back to it. We do not republish articles — we read the source, then write our own piece. A claim that cannot be traced to a source, to market data, or to a comparable moment in the past does not make it into the text.
3. We say what could follow — both ways
Each piece answers the same question: what could this change? We write the case for it going well first, then the case for it going badly, and we say which of the two we find more convincing and why. Both cases have to be real ones. A bad case written only so a good case looks balanced is padding, and we cut it.
4. We say what would change our mind
A view worth reading names the thing that would break it — a flow that reverses, a ruling that lands the other way, a level that does not hold. Naming it is what separates a view from an opinion, and it is what lets you check us without waiting for anyone's scoreboard.
5. AI writes it, with automated safeguards
The analysis is written by AI and labelled as such on every page. Before anything is published it passes automated checks: the facts are corroborated against independent reporting, the text is tested for originality against its own source, and any unverified negative claim about a named person or company blocks publication outright. We do not publish accusations. Anything that fails is set aside, not softened.
6. We correct things
Any piece can be corrected or withdrawn at any time, and the archive is not rewritten quietly — a correction says what changed. If we got something factually wrong, tell us and we will fix it.
What we don’t do
This half matters more than the first. Most crypto sites are defined by what they promise; we would rather be defined by what we have stopped promising, including things this site itself used to publish.
We do not forecast prices
No targets, no “BTC to $X by Y”. We tested a signal engine of our own against four and a half years of history across sixty-two coins, and it never beat the simplest possible constant call, at any horizon. So we deleted it rather than dress it up. Anyone who tells you where the price will be is selling something, including a past version of this site.
We do not publish an accuracy score
We used to. A hit rate looks like accountability and is mostly arithmetic about how often a market goes up — ours read 46% while the strategy of saying nothing at all scored 59%. A number that flatters you for being vague is not a measure of anything, so it is gone. What replaces it is the fourth step above: we tell you what would change our mind, which you can check yourself.
We do not give advice
Nothing here is a recommendation to buy or sell anything. We are a publisher, not an adviser. We describe what could happen and what it would depend on; what you do about it is entirely your decision.
We do not chase volume
There is a daily cap on how much we publish, and no quota that has to be filled. If a day has nothing worth analysing, the right number of analyses that day is fewer.