About Stupid Correlations

Real data. Fake causation. A fun little site.

The short version: We take real historical datasets from reputable sources and find pairs that happen to move together — even when there's absolutely no logical reason for them to. The numbers are real. The story we tell around them is a joke.

Why does this site exist?

Somewhere between a statistics textbook and a meme account, there was a gap. We wanted to fill it. If you've ever rolled your eyes at a news headline claiming "Study finds X causes Y!" without looking at the data, this site is for you.

Every chart on our collection page was generated from actual historical data — pumpkin spice sales, ghost sighting reports, Nicolas Cage films, whale catches, TikTok user counts. We compute the statistical correlation between two unrelated series, animate it into a video, and write a caption that pretends the correlation means something. It doesn't. That's the point.

Where does the data come from?

We pull from sources including:

  • Our World in Data — demographic, economic, and environmental series
  • U.S. government statistics — Census, BLS, CDC, NOAA, NHTSA, and similar
  • Industry reports — music, gaming, streaming, sports, entertainment
  • Public cultural datasets — Wikipedia, archive.org, IMDb, and well-maintained community trackers

Every `/c/:slug` correlation page links to or cites its sources at the bottom of the chart. If you find a dataset you think is misattributed or outdated, let us know — we take accuracy of the numbers seriously, even when the conclusions we draw are ridiculous.

How a correlation gets made

Every video on the site goes through the same pipeline:

  1. Pick two datasets at random from our library of ~200 cleaned, normalized time series.
  2. Compute the Pearson correlation over the overlapping years.
  3. If the correlation is high (positive or negative), it's a candidate. If it's barely correlated, it's still fun because the chart shows how weak the link really is.
  4. Generate a photorealistic background image that matches the subject matter (currently via fal.ai FLUX Pro Ultra).
  5. Render a short animated chart over the background with a witty headline, caption, and in-character expert quote.
  6. Add narration via text-to-speech and encode the whole thing as an MP4.

The companion apps for iOS and Android let you do the same thing yourself — pick categories, roll the dice, and get a shareable chart in seconds.

Is this an app, a website, or what?

Both. The mobile apps are where you create your own correlations. This website is where we publish a curated collection, host the random-video experience on the home page, and — soon — sell merch with your favourite charts printed on it. The underlying dataset and rendering pipeline are the same across both.

Who builds this?

Stupid Correlations is built and maintained by a small independent team that genuinely likes data, genuinely likes jokes, and has spent way too long arguing about whether "causation" is funnier than "causality." We're based in Europe. We're not a big company. We don't sell your data (see our privacy policy) and we don't have investors demanding that every chart lead to a purchase funnel.

Is any of this serious?

The statistics are. The narratives are not. If you've arrived here looking for methodology critique: we use Pearson's r over the overlapping years of two series, with optional smoothing on long series. We normalize for chart display but the underlying correlation is computed on the raw values. Our correlations are a starting point for conversation, not a replacement for peer-reviewed research — and we very much do not endorse any of the causal claims our captions make.

If you've arrived here wanting to laugh at how easy it is to cherry-pick data into a "story," welcome. You're in the right place.

Get in touch

Questions, dataset suggestions, licensing, or something went wrong? Visit the support page or email us directly at support@stupidcorrelations.app. We read everything and we reply to almost everything.

— Correlation is not causation. But it's hilarious.