What paid research said today, on the record. What the options market shows, computed nightly.

Quant Data reads 14 paid research publishers every day and records what they published as an attributed ledger: who said what about which ticker, the numbers they put on the record, and what they said would prove them wrong. Next to it, options positioning Quant Data computes itself for 25 individual US stocks and 18 ETFs, dated to the settlement it came from.

It records and measures. It does not recommend a trade.

Free. Subscribing means you agree to receive the daily notes at this address. Unsubscribe any time; it deletes the address.

Every note is published in full on this site; the email is the same text, in English or Chinese. Read today's notes →

Read daily
836 pieces from 14 publishers since 2026-04-28
Computed nightly
Max pain, put/call, gamma flip, open interest for 43 tickers
Measured once, published forever
11 standing studies, failures included

Latest notes · September 11, 2026

Eight notes, seven publishers: Oracle's OCI beat and Nvidia's ballooning off-balance-sheet guarantees headline a day of AI-infrastructure stress-testing

On 2026-09-11, seven publishers tracked here produced eight research pieces. SemiAnalysis stress-tested Nvidia's rapidly expanding off-balance-sheet guarantees to AI-infrastructure buyers; TMT Breakout wrapped Oracle's OCI beat against Adobe's AI-monetization doubts and a run of sell-side rating changes; AlphaPick reviewed its Alpha Picks portfolio against sticky CPI and an Oracle-driven AI rebound; Capital Wars argued bond yields reflect growth, not credit fear, inside a 'Treasury QE' framework; Damnang interviewed a chip engineer on the AI-memory supercycle; James Bulltard logged a day of options trades; and Cassandra Unchained, Michael Burry's tracker, posted a portfolio swap plus a bonded-wine essay on shorting the dollar. Every statement below is attributed to its publisher and unverified; Quant Data holds no view.

8 pieces · 7 publishers · 15 directional statements on record · 12 falsification points · Read the full note

Where should I start?

What does the site keep?

Three records, each growing daily. All free, all with their dates attached.

  • /notes Daily · attributed ledger

    What did paid research publishers say today?

    Titles, directional statements with the originating voice, numbers as stated, and the observations each publisher said would prove it wrong. Reproduced as stated, never verified by Quant Data, never a recommendation.

    9 days published so far.

    Open
  • /sources 14 publishers · counted

    Who is being read, and what do they tend to say?

    One page per publisher: pieces published, tickers covered, the mix of claim types, and how often they publish. Audit results appear only for sources graded under the finance-influencer audit's naming policy.

    836 pieces recorded since 2026-04-28.

    Open
  • /stocks 43 tickers · nightly

    What does the options market show for one stock, and who wrote about it?

    Max pain, put/call open interest, estimated gamma flip and whole-chain open interest, computed by Quant Data from end-of-day open interest, with the research record for the same ticker underneath.

    Settled 2026-09-17. 9 of 43 tickers had a research piece in the last 14 days.

    Open

Standing research

Questions the site measured once and keeps measuring. Each is published with its method and its own numbers, and the results that failed sit in the same tables as the ones that held.

What has actually been measured?

Every event rate is pre-registered with its sample size, and the failed claims are published next to the confirmed ones. The day-type model is scored on held-out ES years. Options outputs state whether they are arithmetic or estimates.

No hold rate has been measured for max pain, gamma flip, or the heavy strikes. They are positioning snapshots, not support, resistance, or targets. The research ledger is a record of what publishers said; their numbers are theirs, not ours.

Read methodology

Range breakouts failing within 10 bars
62.6%ES 2010–2026, pre-registered · folklore says 80%
Per-event estimate, zero-shot on NQ
AUC 0.646terciles 54.2% vs 79.1% · calibration error 5.1%
Day-type model, full session
66%Top-1 of five classes · 37% majority baseline
After the first 90 minutes
53.5%Top-1 · top-2 coverage 68.6%
cib_long on ES
13,867Pre-registered events · measured rate 51.8%

For developers and AI agents: the same data as an API

The positioning data and the measured models behind these pages are also served as JSON and over MCP. A free key covers all four routes; there is no paid plan. See the API reference, the MCP server and how access works.

Free email key

10 successful calls per UTC day, shared across all four routes. No account, password, card, or login.

We send the key only to this inbox. It never appears in the URL or form response.

An AI agent can also request it through POST /v1/access/free-key. Read the privacy policy.

Nothing to buy

The free key, the daily notes and every page on this site are free. No paid plan is offered here; keys bought earlier keep working under the terms shown at purchase.

The daily notes are the only email: one per publishing day, the same text as the web version. Subscribe at the top of this page or on the newsletter page.

How access works

Common questions

What is Quant Data?

A free content site with a data layer. Each day it reads 14 paid research publishers and records what they published as a dated, attributed ledger: titles, directional statements with the originating voice, the numbers they stated, and the observations they said would prove them wrong. Alongside, it computes options positioning nightly for 43 US stocks and ETFs and publishes 11 standing studies with their methods and failures.

Where does the research content come from?

From paid research publishers accessed through the GeoScope aggregator under a licence that permits redistribution. Quant Data reproduces statements and figures as stated, attributed to the publisher, and does not verify them or take a view on them. Titles and content belong to the publishers.

Is a "bullish" entry a recommendation?

No. It records that a named source expressed a direction on a named ticker on a dated day. The record exists so that statements can later be checked against what prices did. Quant Data does not recommend buying or selling, set targets, or personalise anything.

How often does it update?

The research notes are written each day the publishers publish. Options positioning is rebuilt nightly from end-of-day open interest and every page prints its settlement date; it is never an intraday reading.

What does the newsletter contain, and what does it cost?

The same daily notes that appear on this site, delivered by email in English or Chinese. It is free; you confirm the address first, and unsubscribing deletes it. The full text of every note stays on the site.

Is there still an API?

Yes. The positioning data and the measured models behind these pages are available as a JSON API and an MCP server for people and AI agents. A free key gives 10 successful calls per UTC day; no paid plan is offered on the site. Keys bought earlier keep working under the terms shown at purchase.

Is this investment advice?

No. Everything here is descriptive: attributed records of what others said, historical measurements, and options open-interest arithmetic. Nothing is a signal, a target, or a recommendation.