How it is calculated

About 300 RSS feeds from crypto outlets and project blogs, mostly in English. Each feed is polled on its own schedule; a feed that has failed for more than a day is switched off.

We keep the headline, link and date. Article text is never reprinted; the link goes to the outlet.

A story is a set of articles from different outlets about the same news. Articles are grouped automatically by content; one outlet repeating itself never makes a story.

Stories are sorted into topics; a tile's size on the home page is the number of stories in its topic.

The short “What's happening” summary on a topic page is written every two hours by a language model from the headlines, outlet counts and tone of the topic's most-covered stories alone; it adds nothing the headlines do not say.

A language model labels each English article positive, neutral or negative, by how it frames the news for the market.

It does not guess: if an answer fails validation, the article stays unlabelled and does not count towards tone until it is rated again.

The share of positive articles among those rated in the last 24 hours, from 0 to 100.

It is compared with the usual level, the mean of the last 7 complete days. It describes how the press writes, not what to do.