Oil News Index F6
Six fundamental questions, put every day to a thousand oil articles through hundreds of prompts: close to a hundred thousand answers, combined into one number. Published next to the day's crude move, with a plain account of how much of that move the press explains and how much it does not. A description of the news flow, not a forecast.
What the oil press is saying
The read of the last complete press day (UTC), the move it accompanies, and the six questions behind the number. A new final value every morning; a provisional one every six hours through the day.
Made with spinfThe engine that reads the oil press is open as an API
spinf reads a document once and scores every question you ask of it, at a fraction of the usual cost. In beta: try it on your own documents.
The attribution cards
For the last market day: the actual WTI move over 1, 5 and 20 days, the move implied by the change in the read, and the split between the six questions. Each card carries its own hit rate from the back-analysis and, once live, from the published record.
The read against the price
Level, change or momentum; WTI or Brent; any year since 2022. The correlation shown is computed on the displayed window, from the same rows as the data files.
Level: OIL-F6 against WTI (CL)
Loading the history…
The read (six fundamentals, averaged) against the log price minus its trailing 60-day median: where the press has the market versus the last two months. The read on a market day is the press published on the next trading day.
How well it explains, and how well it does not
The numbers behind the claims, out of sample where a fit is involved. They are a description of the press: the read has no lead on the price, and the residual is large by construction.
How it works
What a reader needs to interpret the numbers ships with the data (spec.json). Four steps, none of them fitted to the price.
- 01
Collect
Every English article of about two hundred selected outlets, from public feeds (GDELT) and private newsfeeds, de-duplicated. More private feeds are being added. The oil universe is the articles with at least three oil terms: roughly a thousand a day.
- 02
Read
An open-weights 12B language model reads each article once against about five hundred prompts: every question is asked through several distinct formulations, and a day yields close to a hundred thousand answers on the oil press alone. No event extraction, no price language; the scoring itself is our own and not published.
- 03
Aggregate
The answers behind each question are combined over the day, standardised on the question’s own trailing year and signed toward a higher oil price. OIL-F6 is the mean of the six.
- 04
Relate
The read is compared with the WTI and Brent moves. One rolling beta on its change gives the news-implied move at 1, 5 and 20 days; the six questions split it exactly. What is left is reported in points, not explained away.
Thousands of reads a day, on one machine
The reading runs on an inference engine we built for this job. It reads an article against hundreds of prompts at a fraction of the usual cost, so the same open-weights model runs at about twenty thousand tokens a second on two consumer graphics cards. That throughput is what makes a daily read of this scale, and the intraday updates, affordable without a data centre.
Three things people ask
Not as a signal. The index describes what the press wrote, and the press writes about yesterday. In our tests it has no lead on the price at any horizon we tried, hourly or daily, and the one-day card covers about a tenth of the move. It is designed to explain, not to predict. Alongside trading it has uses that do not depend on prediction:
- Post-fact attribution. After a move, which fundamental story the press carried, how much of the move it accounts for, and how much it does not. A large residual means the move was not in the news.
- Risk and context. The state of the narrative behind a position: disruption risk at +2 is a different world from disruption risk at 0, whatever the price is doing.
- A baseline for surprises. A read of the whole press against which a single headline, a broker note or another feed can be measured.
- A benchmark for news models. Any news-based model on crude can be compared with a free, fixed reference that reports its own hit rates.
OIL-F6 is provided for information and research purposes only. It is not investment advice, not a recommendation to buy or sell any instrument, and not a forecast of prices. Past relationships between the index and prices do not guarantee future ones. Anyone using these data does so at their own risk.
Yes. The pipeline behind OIL-F6 is not oil-specific: the same reader, engine and validation run today on seventeen sets, from refined products and natural gas to gold, US rates, six currency pairs and five equity indexes, and on sources beyond the public feeds. The picture is similar wherever the market has a press of its own: the read describes the move well, predicts nothing, and needs its own set of questions and its own tests.
The first of those to be published is the Equity News Index (EQ-S10): one read per company for twelve Nasdaq-100 names, built the same way and published next to the share price move, with the questions selected for their steadiness over five years instead of fixed from a textbook.
We build and run custom indexes of this kind for other assets and private news sources, and we can put one on a stream you care about. See what we do or get in touch.
Open data
Everything on this page is a static file: the history by year, the latest object, the spec and the back-analysis numbers. They live on the data API page, free with a key.
Free with an API key
Register once with a name, a company and an email; the key comes by email and opens every file, from the browser, a script, a spreadsheet or an agent.
Oil News Index F6 is news analysis provided for research and entertainment only. It describes what the press is saying; it has no forecasting power, is not investment advice and should not be relied upon for trading decisions. See the terms of service.