Does Congress Sell Before the Drop?

Read down any pair of columns and the sale and purchase figures sit within a point or two of each other on every row. If members were reliably selling ahead of bad news, the “fell 20% or more” row would separate; it does not. At 90 days the sold stocks averaged +4.06% and the bought stocks +4.11%, with the index at +4.64% and +5.28% respectively. Both groups trailed the market slightly over the longer window, which is the same pattern our member-by-member alpha ranking finds from the other direction.

The biggest drops after a sale

Averages hide the cases that get written about, so here are the largest 30-day falls that followed a disclosed sale, by member, stock and trade date. Each is a public filing and a matter of record. What the list mostly shows is when the sales happened.

The top of the list is First Republic Bank in March 2023. 4 sales were made between Mar 13, 2023 and Mar 16, 2023, and the stock fell another 61.7% over the following month before the bank was seized and sold. Those sales look prescient on a 30-day forward window. In context they were made in the middle of a collapse that was on every front page: First Republic had already lost most of its value in the week after Silicon Valley Bank failed, and members were selling into that fall alongside everyone else, not ahead of it. The same ticker also tops the list of drops after a purchase, from the same week, which says more about that week than about anyone’s foresight.

The rest of the table is single trades in the smallest reporting band, often one lot in a much larger managed account, and several were disclosed only after the 30-day window had already closed. Nothing here looks like a pattern of selling before news. It looks like what a large number of sales across a large number of volatile stocks produces by arithmetic: some of them will be followed by a big fall, at about the rate the purchases are.

The same list for purchases

For symmetry, the largest 30-day falls that followed a disclosed purchase. If the sales list reads as members dodging a drop, this one reads as members walking into one, and it is built from the same population with the same method.

What was left after disclosure

Suppose someone only followed the sales that turned out to be right. Take the 197 sales that were followed by a fall of 20% or more within 30 days, the ones a “Congress knew” story would be built from. The median one was disclosed 22 days after the trade, and 30.5% were disclosed only after the 30-day window had closed. Measured from the day each sale became public to the end of its window, the average remaining move was -7.6%, against an average full-window fall of -27.6%. Most of the drop had already happened before anyone outside the member’s household could have known about the sale. The purchase side reads the same way: 259 purchases were followed by a 20% fall, disclosed a median 28 days later, with an average remaining move of -5.7% after disclosure.

This is the same structural point as What Stocks Do Before Congress Discloses, seen from the sell side. The STOCK Act’s 45-day window means a disclosure describes the past, and the sales that would have mattered most are the ones that became public latest.

Methodology and honest caveats

For each disclosed purchase or sale with a listed ticker and a trade date on or after January 1, 2023, we take the nearest trading-day close on or after the trade date and the nearest close on or after the date 30 (and separately 90) calendar days later, and compute the percent change, with SPY over the identical window for comparison. Windows that end after our latest price bar (Sep 14, 2026) are excluded, as are trades outside our price coverage; of 9,410 dated sales and 13,121 dated purchases with a ticker in that span, 8,975 and 12,624 respectively are priced with a closed 30-day window. The “after disclosure” figure uses the nearest close on or after the disclosure date, capped at the window end. Disclosed sales include partial sales, sales inside managed accounts, and sales by a spouse or dependent child; the filings do not say which, and we do not guess. Transaction-weighted figures are dominated by the most active filers, so a single very active household moves the averages; the median row is the check on that. Figures are computed live and update as new filings arrive, so the numbers above may differ from the ones quoted the day this was published.