If you had copied Congress

The test is simple and a little unfair. Take a member of Congress with at least three disclosed stock purchases. On the day each purchase becomes public, buy the stock. Equal weight, up to fifteen positions, hold to today. Compare the result with putting the same dollars into the S&P 500 on the same days.

Unfair because you are buying late by definition, and because a member who discloses 400 trades gets scored on fifteen. Simple because it is what a person reading the disclosures could actually have done.

Members

The fifteen most active buyers in our data, ranked by points against the index as of September 5.

• Cleo Fields, +13.0. Fifteen positions from 27 disclosed buys, up 19.7 percent against 6.7 for the index over the same windows.
• Tim Moore, +7.2. Up 20.3 percent against 13.1.
• John Boozman, +2.9. Fourteen positions from 94 buys, up 3.4 against 0.4.
• Julie Johnson, +1.4. Fifteen of 38.
• David Taylor, +1.0. Fifteen of 78.
• Alan Armstrong, −0.2. Fifteen positions from 320 disclosed buys, essentially tied with the index.
• Jared Moskowitz, −0.3.
• Julia Letlow, −0.8. Up 14.9 percent, but the index did 15.7 over the same days.
• Gilbert Cisneros, −2.5. The most active buyer in the set, 420 disclosed purchases, scored on fifteen.
• Dan Newhouse, −3.5.
• April McClain Delaney, −4.3.
• Josh Gottheimer, −4.8. Down 4.0 percent against a flat index.
• Maria Salazar, −5.3.
• Tommy Tuberville, −5.4. Down 5.5 percent. Recall from our disclosure-lag piece that most of his trades reached the public more than a year late, so the copy dates are far from the trade dates.
• Markwayne Mullin, −6.1. Up 6.2 percent, but the index did 12.3.

Five ahead, ten behind. The top two are well ahead and the bottom six are well behind. The middle is noise.

Funds

The same method applied to 13F filings: buy each fund's new positions when the quarterly filing lands.

• Scion Asset Management (Michael Burry), +8.1. Seven positions, up 23.5 percent against 15.4.
• Bridgewater Associates, +1.1.
• CPP Investments, −0.3. Ontario Teachers', −1.5. CDPQ, −3.1. The pensions land near the index, which is roughly what a pension is for.
• Duquesne (Stanley Druckenmiller), −6.1. Appaloosa (David Tepper), −8.4. PSP Investments, −9.8.
• Renaissance Technologies, −11.6. Third Point (Dan Loeb), −15.0. Tiger Global, −17.0.

Berkshire Hathaway opened no new positions in the period, so there is nothing to copy. Baupost, Greenlight and Pershing Square opened one or two, too few to score.

Two of eleven funds beat the index. The famous names did worse than the pension plans.

Why copying loses

Three reasons, and they compound.

The lag. A congressional buy reaches you 30 to 45 days late when the member is on time and years late when they are not. A 13F is filed up to 45 days after quarter end and describes positions that may be gone.

The cap. Fifteen positions from 420 buys is a sample, and the sample is the most recent fifteen, not the best fifteen.

The bar. The S&P 500 is up 10 percent this year. A copied portfolio that goes up 6 has lost, and most of the members above who "made money" lost to the index.

[ADAM: your read]

The leaderboard recomputes daily at /backtest, and each member's page shows every position with its entry date, which is the disclosure date, not the trade date. That one design choice is the difference between a backtest and a fantasy.

Opinion and personal record, not investment advice.