1. The day
A day runs in UTC, from 00:00 to 24:00, as on exchanges and crypto aggregators (except the first test day: October 29 was counted in Bangkok time, UTC+7; from October 30 everything is UTC). Prices come from KuCoin and Gate in 15-minute steps; the exchange is shown for each day. The report comes out in the morning; the +24 h windows for evening stories are completed the next morning, and the page is updated then.
2. Own move
For most of the day coins follow bitcoin. To see what belongs to the coin itself, we subtract the market’s move.
Beta – how strongly the coin usually follows BTC. It is calculated from daily log returns over shared days, up to 90. With fewer than 30 shared days, beta is not calculated. Bitcoin itself has no own move: it is the baseline.
3. Reaction to a story
For each story the bot takes the time of first publication and looks at the coin’s own move 15 minutes, 1 hour, 4 hours and 24 hours later. The result is in percentage points (pp).
The same news is often published by several outlets. The bot merges these articles into one story, and the number of outlets shows how loud the topic is.
4. Spikes and drops
A spike (a sharp rise) and a drop (a sharp fall) are an hour in which the coin’s own move crossed the threshold. The check runs on a 15-minute grid.
The usual range is the standard deviation of the hourly own move over the 7 days before the day starts. Of moves that overlap within an hour, the strongest one is kept. Stories published in the 3 hours before it and the 3 hours after it are linked to it: news almost always comes out after the move.
A spike or drop with the market is marked separately: the coin moved past the threshold, but together with bitcoin, and its own move is less than half the threshold.
5. Sources and tiers
Official – the websites, blogs, repositories and accounts of the projects themselves. Major media – specialist outlets such as CoinDesk, The Block, Cointelegraph, Decrypt. X – posts from accounts on our registry. Mass feeds – the Google News and Bing News aggregators. We do not copy texts: only a summary and a link to the original. We do not publish the text of X posts, only the link.
6. What the AI writes
The day summary and story summaries are written by Claude AI from this data. Before publication, code checks every number and ticker against the calculation, and a second pass checks the claims. If a check fails, the report comes out without the AI text, with numbers only.
7. What the numbers do not prove
A match in time is not a proven cause. Dozens of factors affect the price at once, and the market may already have priced the news in. The site gives no advice and does not say what to buy or sell.
8. One database
Everything the bot collects goes into one structured database. Nothing is lost or overwritten: every story, every price and every sharp move stays in the history with its exact time.
What is stored for each story:
| Field | What it is |
|---|---|
| time | the minute of first publication and the time the bot received the story |
| source and tier | who wrote it: the project’s official channel, major media, X, a mass feed |
| story | all reprints of the same news merged together, and the number of outlets |
| event type | hack, listing, regulation, ETF, macro, unlock, partnership and others |
| price reaction | the coin’s move net of the market 15 minutes, 1, 4 and 24 hours later |
| spike or drop | whether there was a sharp move after the story, and which one |
9. What such a database gives
A single story says little. Thousands of measured stories show patterns you cannot see in an ordinary feed:
- Which sources are worth reading. Every outlet and channel builds up statistics: how often the price actually moves after its stories.
- Which events matter. You can see after which types of news – hacks, ETFs, regulators – the market reacts more strongly, and which go unnoticed.
- What happens before a sharp move. Every spike and drop is linked to the stories published before it, so the “before the move” picture builds up automatically.
- How coins behave relative to the market. Beta, own moves and reactions accumulate by day, month and year.
On this database we build research: which combinations of sources, events and timing most often come before sharp price moves. These are statistics of the past, not a forecast and not advice: decisions to buy or sell are yours.
10. Sources and corrections
The bot keeps the registry of sources with their tiers; on the site it will appear as a table by coin. Correction log: date, page, what was there, what it became. No corrections yet.
How does news affect crypto prices: what we can say and what we cannot
This methodology answers the question "how does news affect crypto prices" exactly as far as the data allows. We can say that after a story the coin's price moved by so many percentage points without the market. We cannot say that it was this story that moved it.
There are several reasons. Dozens of factors act on the price at the same time. Large players could have learned the news before it was published. The move could have come from futures or from one large seller. That is why in all the site's texts we write "coincided with the news", not "because of the news".
Why measure, then
Because over a long distance the matches add up to a picture. If after official news of one type the price moves noticeably in dozens of cases, and after retellings on social media it does not, this is useful knowledge, even if each single case proves nothing. The database has been growing since October 29, 2026, and every week makes it more reliable – a measured crypto news impact on price, not a guess.
I run this service as an experiment on myself: I am not a trader but a marketer and developer, and I am checking whether data helps make decisions better than gut feeling. The methodology is open so that anyone can check the numbers themselves. Found a mistake – message me on Telegram.
Why does the price on the site differ from my exchange?
We take KuCoin or Gate prices in 15-minute steps and count the day in UTC. On another exchange and in another time zone the numbers will be slightly different.
How do I report a mistake?
The “Message the author” button under each report opens the author’s Telegram. Corrections are published in the log below with a date.
How does news affect crypto prices?
The bot measures how the coin's price changed 15 minutes, 1 hour, 4 hours and 24 hours after the story, minus bitcoin's move. A price move almost always has several reasons.
Why is bitcoin taken out of a coin's move?
Most of the time coins move after bitcoin. Taking it out shows what belongs to the coin itself.
Who counts the numbers – AI or code?
Code. The AI only writes the recap text, and code checks every number in it against the calculation.
Can I check the calculation myself?
Yes: the formulas and thresholds are open on this page, and the prices are public candles of the KuCoin and Gate exchanges.