What they said, and what it moves
Meantly reads what leaders, central banks, regulators and companies say, as it is published, and turns each statement into a number a trader can use: an expected move for the markets it touches, at horizons from four hours to a month. Every call is checked against the real price, and the score is public.
What they said. What it moves.
What Meantly does
Every 15 minutes Meantly collects official transcripts, central-bank and ministry releases, company newsrooms and SEC filings, and direct quotes in the business and general press, from about 370 speakers: heads of state and government, central banks and their governors, finance and energy ministers, regulators, OPEC, the IMF and IEA, rating agencies, and 40 of the world's largest listed companies. Reports citing unnamed sources about any of them are collected too, kept separate and marked unconfirmed.
How a reading is made
- Collect. Transcripts, releases, filings and attributed quotes in more than eight languages. A quote is kept only when it is clearly tied to one named speaker, and an automatic check confirms the attribution. Each item links to the original.
- Read. An AI model reads the intent: who the message is aimed at, what it is trying to achieve, the pressure it puts on markets and the policy lever it leans on.
- Gate. Three checks a trader applies by hand. Agency: can this speaker act on it, or is it a position? Novelty: is this new, or a repeat of a known line? Catalyst: does the statement point to a dated event?
- Score. For each market the statement concretely touches, an expected move range at five horizons from 4 hours to a month, a confidence from 0 to 100, and the mechanism in one sentence. Horizons run in each market's trading hours, not clock time.
- Check. At each horizon the call is compared with the real price: hit, partial or miss. Hit rates by speaker, market, horizon and speaker type are published, and the system learns from its own record.
Why Meantly
Plenty of services summarise the news. Some score central-bank language for banks. None show, for every speaker who can move a market, what the words should do to the price, and then show whether they did.
| If you use… | You get | What Meantly adds |
|---|---|---|
| Central-bank sentiment scores | A dovish-to-hawkish number for a few central banks, sold to institutions. | Every central bank plus ministers, regulators, heads of government, OPEC and companies, scored as an expected move per market, with a public hit rate. |
| News-analytics feeds for quants | Entity tags and sentiment per article, on enterprise contracts. | Agency, novelty and mechanism per statement, netted per market, readable by a person, at the price of a streaming subscription. |
| AI news summaries | What happened, and a line on "market implications". | A number with a horizon, and a scoreboard that says whether the last hundred such numbers were right. |
| Forecast scoreboards | Calibration for questions people chose to forecast. | Calibration for speakers: which central bankers move the 10-year, which politicians are noise. |
| Fact-checkers | Whether a claim is true, often days later. | What the words are designed to do to a price, within minutes. Use both: they answer different questions. |
What you only get here
- The three gates on every statement. Can they act, is it new, when does it matter. Most of a trader's time goes on exactly those questions.
- Numbers, not adjectives. An expected range per market, at five horizons, in the market's own trading time.
- The whole day's pressure, netted. The Sectors page shows who is pushing each of 69 markets, which way, and how hard.
- Prediction versus market. Every resolved call, listed. Hit rate by speaker, market, horizon and speaker type.
- Leaks handled honestly. A report citing unnamed sources is read as an unconfirmed claim about the subject, never as the subject's own words.
- Global by design. Sources in many languages, a site in eight, and markets from the S&P 500 to USD/EGP.
- Searchable history. Every read and every call since launch.
What Meantly is not
- It is not advice. Every number is a speculative scenario to think with; we are not responsible for decisions made on it.
- It is not a forecast to follow blindly. The hit rate is the point: it tells you where the read is worth something and where it is noise.
- It is not a fact-checker. It reads intent and pressure, not whether a claim is true.
Reads are generated automatically by AI and can misread context, and attribution can be wrong. If you spot a mistake, tell us and we'll correct or remove it. See our content standards.
Free and Pro
The live feed, with every read, the three gates and the market-impact grid, is free. Pro unlocks the Sectors page (net pressure per market, who moved it), the prediction-versus-market record, full archive search, alerts and exports.
Who's behind it
Meantly is an independent project. It has no political affiliation and takes no money from parties, campaigns, governments or issuers.