Fundamental Data & Stock Research Platforms

Every product here will show you a revenue line for the same company, and they do not all mean the same thing by it. Three questions separate them and none is answered on a pricing page — where the numbers come from, whether restatements were folded back into prior years, and how many of those years the tier you are looking at actually unlocks. History depth is sold as a tier far more often than it is a property of the product: the same terminal can cost $24.95 or $119.95 a month depending only on how far back it will show you.

• Global coverage and deep history in a terminal — TIKR.
• The numbers in Excel or Google Sheets — Wisesheets at $60 a year.
• Segment and KPI data lifted out of the filing — Daloopa.
• Filings, transcripts and investor presentations — Quartr.
• A valuation model whose inputs you can change — Stock Unlock.
• Screening across a long history of ratios — GuruFocus.

What actually decides the choice here

Every product in this listing will show you a revenue line for the same company, and they do not all mean the same thing by it. Seven questions separate them, and none is answered on a pricing page.

Where the numbers come from. There are three routes and they fail differently. A vendor can licence a standardised feed — several of the retail terminals here run on S&P Capital IQ's standardised financials, which is why their income statements agree with each other and not with the filing. It can tag filings itself, in which case the figures match the 10-K and the line items drift from company to company. Or it can ride the free floor: the SEC's XBRL company-facts endpoint, which needs no key and costs nothing, and which is the reason a dozen cheap products exist at all. A standardiser is slow and opinionated, a self-tagger is idiosyncratic and occasionally wrong, and the free endpoint silently drops anything a company tagged with its own extension elements. Ask which one you are buying.

How far back, and whether that is the product or the tier. History depth is sold by the tier more often than it is owned by the product. TIKR is the clearest case: the same terminal gives 5 years free, 10 years on Plus at $24.95 a month, 10 years and 40 quarters on Pro at $54.95, and 30 years on Ultimate at $119.95. Nothing about the underlying data changed between those screens. Decide first how many full cycles you need to see — for most of what people value on these platforms, ten years is less than one.

As-reported or restated. A vendor that folds restatements, discontinued operations and reclassified segments back into prior years gives you a clean, comparable series that nobody ever saw. A vendor that keeps the original figures gives you a messy series that is exactly what the market was pricing. Both are legitimate; only one is honest to backtest against, and the distinction is invisible in the interface. In XBRL the default convention makes it worse, because a restated amount is the plain value and the superseded one is the specially tagged exception.

Segment and KPI data. The numbers that actually move these businesses — subscribers, bookings, same-store sales, remaining performance obligations, margin by segment — do not exist in any standardised chart of accounts, because they are different in every company. They are reported under company-specific extension tags, which the free SEC endpoint excludes by design, and a standardiser has nowhere to put them. Everything else in this category is available somewhere for $60 a year; this is the part that is genuinely hard, and it is the entire argument of the products built around it.

Non-US coverage. This collapses faster than anything else and is where these products most differ in practice. GuruFocus is unusually candid about why — its US membership is $549 a year and Europe, Asia and Canada each cost extra, because each one had to be bought. Most products are less explicit and leave you to discover the gap after you have built a watchlist. Read the exchange list, and treat "global" with the suspicion it has earned.

Whether the data leaves the product. A number you cannot export is a number you cannot check. Some products here are deliberately screenshot-only; others exist for the opposite reason — Wisesheets sells 30 years of financials across more than 50 exchanges into Excel and Google Sheets for $60 a year, or $120 for its faster tier, precisely because it sells the data rather than an interface to look at it in. If your work ends in a model of your own, put export near the top of the list, and find out whether that means an API, a CSV, a spreadsheet function or a PDF.

How much of the screen is the vendor's own model. Fair values, quality grades, composite scores, normalised owner earnings and predicted returns are outputs, not observations. They are often the most prominent figure on the page and the least attributable. The usable ones tell you the formula and let you change the inputs; the rest are a conclusion wearing the typography of a fact.

The words this market uses loosely

"Fundamental data" covers three different things — the figure as reported, the figure after a vendor mapped it onto a common template, and a figure the vendor computed. Products move between the three without marking the boundary.

"Fair value" and "intrinsic value" are never properties of a company. They are the output of a discount rate, a growth path and a terminal assumption that somebody chose. The only question worth asking is whose, and whether you can edit them.

"As-reported" should mean the original filing, untouched by later restatement. Some vendors use it to mean only that line items were not remapped, while still carrying restated values.

"KPI" and "segment data" are used both for figures lifted out of the filing and for figures a vendor derived from them. Derived numbers are fine; being unable to tell which is which is not. This is why the serious KPI products hyperlink every cell back to the page it came from.

"Coverage" is three unrelated numbers wearing one word — how many tickers, how far back, and how many line items per company. A vendor quoting 100,000 companies may carry five years and forty line items for most of them.

Three jobs sharing one audience

Fundamentals is the data job: assembling a history you can rely on. The buyer wants source, depth, line-item count, restatement policy and export, and does not care what the interface looks like. Judge these against each other on the seven questions above, and on nothing else.

Valuation is the modelling job: turning that history into a number you would act on. The product is a set of assumptions and the freedom to change them, so judge it on which models it offers, which inputs are editable, whether it shows you the sensitivity rather than a single figure, and whether it validates against enough history to be worth running. A valuation tool that will not let you move the discount rate is a rating service with extra steps.

Filings is the documents job: 10-Ks and 20-Fs, transcripts, presentations, 8-Ks, and getting to them quickly. Judge on latency after the call ends, how far the transcript archive goes back, whether non-US investor-relations material is covered at all, and whether the text is searchable across companies rather than one document at a time. Transcript history is another thing routinely sold by the tier — the same platform may hand you 90 days or two decades.

The three share an audience and almost nothing else. A KPI extraction service, a DCF workbench and a transcript archive answer different questions, and a ranking that puts them in one column is measuring nothing. Decide which job you are here for, then compare only inside it.

One filter worth applying yourself

A search for the best fundamental analysis site returns two markets mixed together, and they are judged on completely different things. Switch the vendor's opinion off and see what remains. If a filing archive, a financial history and a model you control are left, it is an instrument, and coverage, depth and export are the right questions. If the star rating or the analyst fair value was the reason to subscribe, it is a recommendation service, and the only question that matters is its track record. This catalogue lists the first kind, which is why several well-known names are absent from the listing below.

Where this category connects

If you want these numbers in your own code rather than in an interface, several of the providers here sell an API and are listed as such in market data APIs. A valuation history that a strategy will be tested against has a further requirement nothing here guarantees — point-in-time figures free of survivorship bias — which is the first thing to settle in backtesting frameworks.