Stock trades by members of Congress

Would copying Congress have paid?

Suppose you had copied every House member’s disclosed holdings: starting from the holdings in each annual disclosure report once it was published, then buying and selling as each trade report became public, sized by the amounts reported (call options counted as the stock until they expire). Positions are sized in dollars, so members with large portfolios carry most of the weight. $10,000 put in on January 5, 2015 would have grown to $57,644 by November 2, 2026, against $45,514 in the S&P 500. That is +16.0% a year against +13.7%, with much the same ups and downs as the market (beta 1.10). Trade by trade the picture is weaker: six months after the report, the average purchase was −0.5% against the S&P 500, the typical (median) one −1.6%, and 46% beat the index.

The member’s timing or the follower’s?

Each purchase measured against the S&P 500 over the same dates, from the close on the trade date (what the member got) and from the close on the first trading day after the report (what you could have got). Only windows that have finished are counted. Reports took a median of 28 days to arrive.

Sales: six months after a sale was reported, the stock had moved −0.1% against the S&P 500 on average (a negative figure means selling was the right call); 55% of sold stocks went on to trail the index. Weighting each trade by the size band filed, purchases were +1.8% after six months.

By party and chamber

Member by member

Members with at least 20 purchases that have six months of prices after the report, ranked by the average purchase against the S&P 500 six months after it was reported. 133 other members traded too little to judge; their figures are on their own pages. Click a column heading to sort.

How much to read into this. Of the 104 members ranked, 29 have trade-by-trade results that look statistically significant (a t-statistic of 2 or more either way: 7 ahead of the market, 22 behind), where luck alone would produce about 5.2 (5%). That overstates it: one member’s purchases bunch into the same weeks and often the same stocks, so they are not independent bets. The Evidence column counts one average per month of reports instead; on that basis 13 of the 81 members with enough months clear the bar (4 ahead, 9 behind) against about 5.3 by chance. With so many members tested, a few will always look skilful or unlucky.

Limits. House members' electronic reports only: Senate reports could not be retrieved and scanned paper reports are not read, so some members' records, including several active traders, are missing. 34,380 stock trades in 1,666 companies we cover, reported from January 3, 2015, priced to November 2, 2026: about 12 years. The trade-by-trade figures use share trades only; the copy portfolio also counts 695 option trades as the underlying stock. Share exchanges are left out, as are trades in companies we do not cover; only companies we cover today are included, which flatters returns slightly (survivorship). Returns include dividends but not trading costs or taxes. Values are filed as ranges, so each trade counts equally in the trade-by-trade figures. The copy portfolio sizes each position by the middle of the band filed, so a few large holdings dominate; a partial sale takes off its band but never more than half the position; put options and gifts to charity are left out; amended reports replace the trades they correct. The trade-by-trade figures instead measure each purchase over fixed periods. These figures describe the stocks’ moves after each trade; they say nothing about why a member traded and do not suggest any member used non-public information.

Does a Congress long-short strategy beat the market?

A published design (following Quiver Quantitative's description) buys the stocks members reported buying and bets against those they reported selling, sized by the amounts reported, 130% long and 30% short, rebalanced weekly. We tested it on House reports, keeping a trade for 30, 90 or 365 days after it was reported. Choosing on 2015–2019, the 90-day version looked best (Sharpe ratio 0.92 against 0.78 for the S&P 500). From April 2020 it returned +17.9% a year with a Sharpe ratio of 0.63, against +19.0% and 0.93 for simply holding the S&P 500, so it did not beat the market once risk is allowed for. Its beta of 1.26 shows most of its return is market exposure; with equal money long and short, the bought-versus-sold gap had a Sharpe ratio of 0.00.

Shares only, entered the trading day after each report became public, before costs. House electronic reports only (the published version also uses Senate trades, which we cannot obtain). Sharpe ratio is return above the Treasury bill rate per unit of swing; the market-neutral version holds equal money in bought and sold stocks.