We've already shown that copying what Congress buys is a coin flip. But buying is the easy, optimistic half of the story. The darker accusation — the one that actually fits the "they know something" narrative — is about selling: a member sees trouble coming in a briefing, quietly exits before the public catches on, and dodges the drop. If that's real, it's testable. The stocks Congress sells should underperform after they sell.
So we took every disclosed sale we could match to a priced US stock — 4,771 of them back to 2014 — and measured each stock's return over the next twelve months against the S&P 500 over the same window. A rolling forward window means a rising market doesn't flatter or spoil the result; we're asking one thing only: after a member sold, did the stock do worse than just holding the index?
The typical sold stock lags — a little
Two numbers tell the whole story, and they point in opposite directions:
- The median sold stock returned 4.2 points less than the S&P over the next year, and 56% of sales were followed by underperformance. That's a real tilt — but it's a coin flip weighted to 56/44, not a crystal ball.
- The average sold stock returned *+2.9 points more** than the S&P. Read that again: on average, the stocks Congress dumped went on to beat* the market.
Both are true because the distribution is lopsided. Members avoid the median dud a bit more often than chance — but they also sell plenty of stocks that keep ripping, and those big winners drag the average positive. Selling saved them from the typical loser and cost them the occasional monster. Net of both, it's close to a wash.
Whether selling "looked smart" was decided by the market, not a secret
Read the chart year by year and the "informed exit" story falls apart. The hit-rate sits a few points above a coin flip in most years and below it in others. And the average swings with the weather: in the roaring years — 2020 (+9.8 points) and 2025 (+7.5) — the stocks they sold kept soaring, making every exit look premature. In the softer years, the same behavior looked prescient.
That's the tell. If the sells were driven by private information about specific companies, they'd work in bull and bear markets alike. Instead, whether a year's selling looks brilliant or foolish tracks which way the whole market went — which is exactly what you'd expect from ordinary, uninformed selling: rebalancing, raising cash, paying a tax bill, funding a house.
The sell side isn't sharper than the buy side
It's a hair different, not a different verdict. On the buy side, only 45% of congressional purchases beat the market — worse than a coin flip. On the sell side, 56% of sales preceded underperformance — a touch better than a coin flip. You could squint and call that a faint, real asymmetry: members are marginally more "right" when they get out than when they get in. But 56% is not what informed trading looks like, full exits (56% hit-rate) were no sharper than partial trims, and the positive average means a "sell when they sell" rule would have walked you out of stocks that, on balance, went on to beat the index.
The honest version
The viral version — Congress dumps before the crash — is mostly wrong. There's a faint loss-dodging tilt in the median trade, real enough to see in the data and far too weak to trade on, and it's swamped by all the winners they exit early. Combined with what we found on the buy side and inside the committee-oversight claim, the picture is consistent: congressional trades — in and out — are a slightly above-average portfolio with an ordinary amount of noise, not a stream of secrets. Watch the disclosures on /congress as an accountability tool, not a trading signal.
Methodology: every disclosed congressional sale (full or partial) from 2014 to mid-2025 that we could match to a US stock with price history — 4,771 trades; sales of funds, bonds, options and unpriceable assets are excluded. Each stock's return is measured over the 12 months following the transaction date against the S&P 500 over the identical window, equal-weighted, grouped by year. Years shown have 15+ priced sales; 2021–2022 are omitted for sparse coverage. Measuring from the transaction date tests whether the member's own timing beat the market; a copier, who only sees the trade up to 45 days later, would do no better. Not investment advice.