What is alternative data? A plain-English guide with free, live examples
A short definition, real examples you can check today, and an honest look at whether it works.
Alternative data, in one sentence
Alternative data is any information about a company that doesn't come from the company itself: not its earnings reports, SEC filings, press releases or analyst notes. Big funds pay for things like credit-card panels and satellite photos of parking lots. Plenty of it is free and public, though, like TSA airport checkpoint counts, FDA adverse-event reports and Federal Reserve production numbers, and you can look at live examples on this site.
Traditional data is what a company says about itself. Alternative data is what the outside world shows about it.
Why it's called "alternative"
Investors have always used "traditional" data: financial statements, quarterly and annual filings, company guidance, earnings calls and broker research notes. All of it comes from the company or the people who cover it, and most of it arrives a few times a year.
Alternative data is everything else that hints at how a business is doing: how many people are flying, how many boxes are being made, how often a drug shows up in FDA reports. It's "alternative" because it comes from outside the usual channels, and it often updates daily or monthly instead of quarterly.
Common examples (paid and free)
| What big funds pay for | Free public versions |
|---|---|
| Credit and debit card spending panels | U.S. Census Bureau retail and business data |
| Email receipt panels | FRED (St. Louis Fed) and Federal Reserve production indexes |
| Satellite and drone images of parking lots and crops | EIA energy data, BTS transportation stats |
| Phone location data (foot traffic) | TSA airport checkpoint counts |
| App usage and downloads | Wikimedia page views, PyPI downloads, Hacker News |
| Web scraping at scale | SEC EDGAR filings, openFDA, NHTSA, BLS, Greenhouse job boards, GDELT news index |
This is a sample, not a full list. Every free source this site uses is listed under Method & sources.
Three free examples you can see live
Each one below is a real signal on this site. The numbers are from the site's data as of the date shown. Click through for today's reading.
1. Airport lines and airlines (Delta)
The data: the TSA publishes how many people pass through U.S. airport security each day. Why it might matter: more people in line usually means fuller planes, and the TSA posts daily while airlines report quarterly.
What it showed: 2.23M travelers/day on a 7-day average, down 2.6% vs the same week last year (data through Nov 2, 2026). Tested: no predictive power found
See the live TSA signal for Delta →
2. Box production and Amazon
The data: the Federal Reserve's monthly production index for paperboard mills, the raw material behind cardboard boxes. Why it might matter: more boxes being made can mean more goods moving through warehouses and e-commerce.
What it showed: an index of 88.1 (last 3 months, average), down 1.8% vs the same period last year (data through Jun 2026). That's inside the ±3% range the site treats as normal noise. Tested: no predictive power found
See the live paperboard signal for Amazon → It's also walked through end to end on How a signal is found.
3. FDA reports and a drug maker (Novo Nordisk)
The data: monthly FDA adverse-event reports that mention semaglutide, the drug in Ozempic and Wegovy, from openFDA. Why it might matter: report counts tend to grow with the number of people taking a drug, though a spike can also flag a real safety issue.
What it showed: 1,607 reports/month (last 3 months, average), down 59% vs the same period last year (data through Sep 2026; the FDA re-releases this data quarterly). Not tested yet
See the live signals for Novo Nordisk →
Does alternative data actually work?
Mostly not, in our tests. This site tracks 111 signals. 88 had enough history to backtest, and across 264 tests none showed predictive power after a correction for running that many tests. Seven deeper per-stock tests all failed. The one result that held up: companies that filed late SEC reports lagged similar-sized stocks by about 4% over the next 3 months (2012–2026), with big caveats.
The lesson: alternative data is a clue, not a crystal ball. Public data is often already priced in by the time you see it. Full details are on Do alt-data signals actually work?, the scoreboard and the track record.
Is alternative data legal?
Reading public government data is normal and legal. The legal questions are about how some paid data is collected and whether it contains material nonpublic information (MNPI).
In an April 26, 2022 risk alert, the SEC's Division of Examinations said alternative data does not necessarily contain MNPI, but it found investment advisers without proper written policies for checking their alternative-data vendors. (SEC risk alert, PDF)
The Other Data only uses official APIs or public downloads: no logins, no paywalls, and it respects robots.txt (see How it works). This isn't legal advice.
How to start looking at alternative data yourself
- Pick a company.
- Ask what real-world activity drives its sales. Flights for an airline, boxes for an online store, prescriptions for a drug maker.
- Find a free public number that tracks that activity, then compare it with its own past. This site uses the last 28 days vs the 28 before, or year over year.
New to signal cards? Start with How to read this site, or search a ticker on the homepage.
FAQ
What is alternative data in simple terms?
Information about a company that doesn't come from the company or its analysts, like airport traffic counts or FDA side-effect reports.
What are examples of alternative data?
Paid examples include card transaction panels, satellite images, app usage, web traffic and social sentiment. Free ones include TSA checkpoint counts, FRED economic data, openFDA reports and SEC filings.
Is alternative data only for hedge funds?
The expensive kind mostly is. Free public data is open to anyone.
Is alternative data the same as insider information?
No. Public data is public. The SEC's concern is data that contains material nonpublic information (MNPI).
Can alternative data predict stock prices?
Rarely, in our tests. 88 backtested signals showed no predictive power after a correction for running 264 tests. Only late SEC filings held up, at 3 months. Details are on our 'Do alt-data signals work?' page.
Numbers from signal-status.json and the DAL, AMZN and NVO ticker files, as of Nov 4, 2026. Education only. Not financial advice.