We Checked 355 Congressional Stock Purchases. The Edge Wasn't There.
The claim that members of Congress reliably beat the market — helped along by early access to non-public information — is one of the most repeated stories in retail trading. It's also one of the few popular trading myths you can actually check instead of argue about, because the STOCK Act requires every one of these trades to be publicly disclosed.
We checked. Using disclosed trades from 63 named members of Congress between June 2025 and July 2026 — 825 trades after cleaning the data, which took longer than expected and is worth explaining before the numbers.
The data had to be cleaned first
The raw feed we archive from contained 1,528 records tagged as congressional trades. Building this piece, we found that 702 of them — 46% of the entire file — belonged to one name: "Alan Armstrong," trading tickers like WMB (Williams Companies) and a global basket of large-caps in sizes up to $25 million, with no party or state on file. That isn't a member of Congress. It's the actual, real-world CEO of Williams Companies — corporate insider trading, mislabeled at the source as a Senate disclosure.
We removed it. Every genuine sitting member of the House or Senate has a party and a home state on record; this filing had neither, so we now require both before a record counts. The collector that feeds this archive has the same filter going forward, so this can't quietly creep back in on the next weekly refresh. What follows uses only the cleaned 825.
The headline number
Across 355 congressional stock purchases with a computed return, the average return in excess of a market benchmark was −0.67%, and only 46.5% beat the benchmark at all.
That's not the dramatic underperformance a messier version of this dataset would have implied — it's close to a coin flip, tilted slightly negative. But "close to a coin flip" is also not "beats the market," which is the actual claim being tested. Over this window, buying what Congress bought did not give you an edge.
The one split that isn't close to a coin flip
The STOCK Act sets a disclosure deadline, and not every trade meets it. Splitting purchases by whether they were filed on time or late turns up the sharpest result in the data:
| Filing | n | Avg excess return | Beat market |
|---|---|---|---|
| Filed on time | 314 | +0.69% | 49.0% |
| Filed late | 41 | −11.14% | 26.8% |
Trades disclosed on schedule were, on average, a wash against the market. Trades disclosed late underperformed it by double digits, and beat it barely a quarter of the time. If privileged information were driving these trades, you'd expect the opposite pattern — the best trades hidden longest. The data runs the other way: late filing correlates with worse outcomes, not better ones. That's a correlation, not a claim about why any individual filed late, and the late group is a small sample (41 trades) — but it's the most consistent signal in the dataset.
The median disclosure took 25 days — inside the legal window, not evidence of a pattern of lateness on its own.
What still holds up
Not every member trades the same. Among the small number with enough tracked purchases to average meaningfully, one — Rep. Josh Gottheimer, 19 tracked purchases — averaged +16.71% excess return over this window, comfortably the best of anyone with a large enough sample to matter. That's a real, public, individually strong result. It's also one person, one window, and not evidence about "Congress" as a category — the entire point of this piece is that the category doesn't hold up even though specific individuals within it can look very different from the average.
The honest caveats
This is one archive, one roughly 13-month window, during a period when the broader market itself was strong and hard to beat. A different regime could produce a different result. The return and excess-return figures come from the upstream tracking dataset we archive from, not a benchmark methodology we compute ourselves, and we're stating that plainly rather than implying more precision than we have. 355 purchases is a real sample, not a large one, and slicing it further (by individual, by chamber) shrinks it fast.
What the data does support, cleanly: the specific claim that copying a congressional stock purchase is a reliable market-beating strategy did not hold over this window. That's a testable claim. We tested it. It failed — and finding a real contamination bug in the process is exactly the kind of thing that should make you trust a "we checked" claim less by default, not more, until someone actually shows their work.
Scores in this piece come from the ZonedInOut model. How the score works →
Frequently asked
Do members of Congress beat the stock market with their trades?
Does late disclosure of a congressional stock trade suggest insider trading?
How reliable is public data on congressional stock trading?
Get the Sunday Brief
One email a week, before the open: what the board did, which calls worked, which didn't, and what changed. Free. No spam, unsubscribe anytime.