Where research terminals get their data

A research terminal at 40 to 60 dollars a month is not, mostly, a data company. It is an interface, a screener and a licence, and the licence is the part worth ten minutes before you subscribe: it determines what the product can and cannot let you do.

This is not an exposé. The numbers are institutional grade precisely because they were licensed from the institutional vendors. But the supply chain has one consequence worth knowing: the company you pay is a licensee, and licences get renegotiated.

How it works

Four layers, and money moves down all of them.

Companies file. Statements land at a regulator as documents, in each issuer's own presentation, with restatements and reclassified line items arriving later.

Wholesale vendors normalise. A handful of firms turn those documents into a standardised dataset with consistent line items across issuers, countries and years, and attach identifiers, corporate actions and estimates. In equities research the names that recur are S&P Global Market Intelligence (the Capital IQ datasets), FactSet, LSEG and Morningstar.

One of those names is got wrong constantly. Reuters is a news agency. The market-data business people mean is LSEG — Thomson Reuters' Financial & Risk division was sold to a Blackstone-led consortium in 2018, renamed Refinitiv, then acquired by London Stock Exchange Group, which says on its own site that it completed the acquisition in 2021, that the Refinitiv brand has been fully retired, and that LSEG Data & Analytics is the direct successor to that business. Product attribution lines lag this by years.

Downstream products license a slice. This is where a research terminal sits: it buys the right to show certain datasets to its own subscribers, on certain terms, and builds the screens.

You subscribe to the fourth layer and see none of the three beneath it, unless somebody says.

Some do. Four of the products in this catalogue's fundamentals and research section say so in their own words:

• Koyfin's help centre says it has licence agreements with over a dozen vendors and does not scrape the web, and names them by dataset: global equity fundamentals, consensus estimates and valuation from S&P Capital IQ, funds from Morningstar, economic data from FRED and Trading Economics, FX from True FX, cryptocurrency from Polygon. Last updated December 2024.
• TIKR's home page describes "financial data powered by S&P Global CapitalIQ on 100,000+ stocks across 92 countries and 136 exchanges", as the page read in September 2026.
• GuruFocus puts it in the terms of use, last updated July 2026: stock quotes from QuoteMedia, Inc. (CSI), company fundamental data from Morningstar, and analyst estimates from both Refinitiv and Morningstar with priority given to Refinitiv.
• Stock Rover's support database names Morningstar as its primary source under a licensing agreement, with additional data from Zacks, Intrinio, Quandl and Yodlee, and real-time pricing from IEX.

Two of those lists carry brand names their owners have retired — Refinitiv, and Quandl, which Nasdaq acquired in 2018 and folded into Nasdaq Data Link. That is what an attribution line is: it tells you what the chain was when somebody last wrote it down, which is why you read it with its date attached.

Others do not publish a supplier at all. Several products describe their coverage in great detail — company counts, years of history, number of metrics — without naming who supplies it anywhere this page could find. That is a finding rather than an accusation: one fewer thing you can check before paying, and answerable in one email.

What a downstream licence actually restricts

A wholesale contract almost never says "you may have the data". It says what the licensee may do with it, and the distinction that matters is display versus redistribution. A display-only licence lets the product show you a number on a screen you are logged into. A redistribution licence — the expensive one — lets it hand that number to you in a form you can take away. Four things sit on the far side of that line, and they are the four complaints people have about cheap terminals:

1. Export. CSV and Excel download of the licensed fields.
2. API access. A programmatic endpoint, which is redistribution by another name.
3. History depth. Deep back-history is often a separately priced dataset upstream.
4. Bulk download. The whole universe at once, rather than one company at a time.

This is why so many research terminals will show you a number and not let you have it, and you do not have to take it on trust: Koyfin publishes the mechanism. Its help centre says price, performance, technicals, percentile ranks, model-portfolio, economic and fund data can be downloaded from tables — and that financials, estimates and valuation data for global equities are currently restricted from download by its data vendor, in tables and charts (entry last updated July 2025). The same help centre says Koyfin does not offer data via API because of restrictions from its data providers (that entry last updated October 2021).

Notice which fields those are: the three Koyfin cannot let you download are exactly the three it attributes to S&P Capital IQ. The chain is visible from outside if you read both pages together.

The restriction also runs the other way, down to you. GuruFocus's terms of use state that membership does not grant rights to redistribute or resell its data, online or offline, and that a commercial data licence must be obtained to do so. Whatever you build on a retail research subscription inherits terms set two layers up.

Estimates are the most concentrated part of it

Fundamentals can, at a push, be rebuilt from filings — that is the business Daloopa is in, and why BamSEC and Quartr sell access to primary documents rather than a normalised feed. Consensus estimates cannot: they are aggregated from sell-side analysts' models under contributor agreements, and few firms hold those relationships. That is why estimates appear in every attribution line above as a named wholesale vendor's dataset.

It is also the layer where wholesale vendors have been buying the layer above them. S&P Global announced the completed acquisition of Visible Alpha on 1 May 2024 — a provider of consensus forecast estimates, KPIs and analytics from sell-side analyst models distributed through a web platform, APIs and feeds — and placed it inside S&P Global Market Intelligence's Capital IQ Pro platform. A distribution channel became part of the supplier.

The layer beneath your tool can move

Three cases in this catalogue, in each of which the product was fine and the layer underneath it moved.

• IEX Cloud shut down on 31 August 2024, three months after the announcement, with no successor product and no migration path.
• Quandl was acquired by Nasdaq in 2018 and became Nasdaq Data Link, with new commercial terms on the premium half.
• The unofficial Yahoo Finance API was never a contract at all, and the endpoints behind it were restricted and changed without notice.

Nobody affected had done anything wrong. They had a dependency they could not see and could not renegotiate.

What the terminal genuinely adds

Being a licensee is not being a middleman. The work between the feed and the screen is real:

• Coverage normalisation. Making a German industrial and a US software company comparable on the same line items, in the same currency, across twenty years.
• Restatement handling. Showing the figure as originally reported or as later restated, and doing it consistently. This is where two tools on the same feed legitimately disagree.
• The screener. A wholesale dataset is not a query interface, and turning several hundred fields into something that answers a question in a second is most of the engineering.
• The interface and the speed, which is why anyone chose one of these over a spreadsheet.
• The price. This is the real one. The wholesale vendors sell per seat, per year, at figures they do not publish and negotiate per contract. A browser terminal at 40 to 60 dollars a month is the only route by which an individual sees this data at all.

What you can do about it

Check the sourcing before you subscribe, in this order. It takes five minutes.

1. Search the help centre for "data source", "where do you get" and "provider" — where Koyfin and Stock Rover put it, and the likeliest place for a plain answer.
2. Read the terms of use with the browser's find. Search the names — Morningstar, S&P, Capital IQ, FactSet, LSEG, Refinitiv, QuoteMedia, Zacks — then the verbs: "redistribute", "resell", "export", "download", "personal use", "commercial". This is where GuruFocus puts it, and a page written by a lawyer says more than a marketing page.
3. Look at the small print under a data table or chart inside the product. Display licences usually require attribution, so the licence often names itself.
4. Note the date on whatever you find, and treat anything undated as describing the past.

Ask support one question, in one email. "Which data provider supplies your fundamentals and consensus estimates, and does my subscription allow me to export them to CSV?" The answer, or the absence of one, tells you most of what this page is about.

Five questions separate a durable product from a thin wrapper. Does it publish its sources at all? Does more than one supplier appear on that list? Can you export the fields you care about? Is there an API, and on what terms? And who owns the company — because the wholesale layer has a record of buying the layer above it, and an acquisition is when terms get rewritten.

Treat exportability as the thing that actually protects you. A subscription you cannot export from is a rental: the day it ends, or the day its upstream licence changes, you have nothing. One you can export from becomes an asset you keep. Check it on the free tier before you pay, on the fields you care about rather than the ones the marketing page shows.

Have a second source for anything you rely on. Not a second subscription — a second route. Pull a few figures for three companies from two places and see whether they agree; where they do not, the difference is almost always a restatement convention or a fiscal-period mapping, and knowing which is worth more than the figure. OpenBB puts one Python interface over many providers, and the market data APIs section covers the layer below the terminals, where export and API access are the product rather than the exception. If the work is really about documents, Quartr and BamSEC get you to the filings and transcripts themselves, which no licence sits on top of.

Then buy the interface, not the idea of proprietary data. If two terminals name the same wholesale vendor for the fields you care about, the numbers are not what you are choosing between. The screener, the speed, the export rights and the price are.

The wholesale layer itself — who owns it, and where its pricing power comes from — is mapped in who owns the market data industry.