Key points
- Company insiders have the best-documented signal of the three, but most of it is gone by the time a trade is public. Research on insider "clusters" found that once a typical one-day cluster is disclosed, the stock's return afterward is flat to negative. Only rarer, multi-day clusters kept an edge after disclosure.
- Copying a hedge fund's full 13F holdings mostly does not work because of the 45-day lag. Copying a patient fund's single highest-conviction position does better, with research putting that "best idea" effect at roughly 3 to 4% a year above the fund's own average.
- The two largest, most recent published studies on congressional trading since the 2012 STOCK Act found no edge at all. Stocks senators and House members bought actually underperformed a matched benchmark over the following six months.
- The pattern across all three: the real edge belongs to the person making a deliberate, unusual, information-driven trade, at the moment they make it. By the time that trade is public, a meaningful share of it is already gone.
I read a lot of these filings for fun, and the question I get asked most is some version of "so should I just buy what they buy." It is a fair question. There is a real paper trail for Congress, for hedge funds, and for company insiders, and copying any of the three is easier than ever with a phone and a few minutes. I went looking for what the actual research says happens once you try it, and the honest answer is less flattering than the "just follow the smart money" pitch you see everywhere. All three have a real signal buried in them somewhere. All three also have a lag problem that eats a bigger chunk of that signal than most people assume.
Company insiders: the best evidence, and a disappearing window
Insiders file the fastest of the three. An officer, director, or anyone owning 10% or more of a company has two business days to report a trade on a Form 4. A large body of research, going back to a 1986 study by H. Nejat Seyhun and running through a widely cited 2001 paper by Josef Lakonishok and Inmoo Lee, found that companies with heavy insider buying went on to outperform companies with heavy insider selling by several percentage points over the following year, an effect concentrated almost entirely in smaller companies.
Where it gets more complicated is timing. Lakonishok and Lee found that most of that return shows up gradually over the following six to twelve months, not in a burst right after the filing hits. And a newer look specifically at cluster buying, three or more insiders purchasing around the same time, found something worth sitting with: once a typical one-day cluster is disclosed, the stock's return afterward is flat to slightly negative. The signal was already priced in before the public ever saw the filing. The exception was rarer, multi-day clusters, insiders buying on several separate days close together, which did keep a real edge even after disclosure. A recent example on this site was Gloo Holdings (GLOO), where three insiders bought a combined $6 million of a stock offering days apart in July 2026. It looked like a textbook cluster, but the purchase had actually been arranged in writing before the deal priced, closer to a pre-committed group buy than the spontaneous kind the strongest research points to.
There is also the routine-versus-opportunistic split from a 2012 study by Lauren Cohen, Christopher Malloy, and Lukasz Pomorski. Routine trades, made on a predictable schedule an insider repeats year after year, carried no signal. Trades that broke from that pattern did, worth an estimated extra return of roughly 0.8% a month. So the useful question is never "did an insider buy." It is whether several insiders bought over more than a single day, and whether the trade looks like a break from that person's normal pattern.
Hedge funds: skip the whole portfolio, watch the best idea
A 13F only has to be filed within 45 days of quarter end, and funds routinely use most of that window. That lag is the central problem with mimicking a 13F, since a position can already be gone by the time you see it. Copying an entire multi-thousand-position portfolio from a fund that trades constantly tends not to work for exactly that reason.
What does hold up better is narrower: a fund's single highest-conviction position. Research on mutual fund and hedge fund "best ideas," the one stock a manager holds the largest concentrated bet in relative to the rest of their own portfolio, found that position going on to beat the market, and beat the rest of that same manager's holdings, by somewhere around 3 to 4% a year for mutual funds, with a similar or larger effect for concentrated hedge fund positions. That edge did not reverse or fade away even years later, which suggests it reflects real conviction rather than a trade that was about to be unwound anyway. Michael Burry's Scion Asset Management is a familiar example of the type: a small, concentrated book where a handful of positions, like his recent short against Micron (MU), tell you far more about his actual view than a diversified fund's two-thousand-line 13F ever could.
The practical version of this is picking a patient, concentrated investor and watching what they hold the most of, not trying to replicate their entire book. A position a fund has held for years does not become meaningless because you are seeing it six weeks late. A position inside a fund that turns its whole portfolio over every few months usually does.
Congress: the popular story is stronger than the current data
I have written before about the older research showing senators and House members beating the market before the STOCK Act required disclosure within 45 days in 2012. That finding is actually more contested than the popular version of it suggests. One later re-examination of the same pre-2012 period found no informational edge at all once it controlled for other factors.
What is not contested is the post-2012 picture, and it is not close. Two of the more recent, larger published studies on congressional trading since the STOCK Act, covering the Senate through 2020 and the House through the end of that year, both found the same thing: stocks members of Congress bought went on to underperform a matched benchmark over the following months, by small but real margins. There was no evidence of an edge tied to committee assignments or leadership positions either. Whatever advantage members of Congress once had, the best current research says it is gone, not just harder to see. That tracks with how the real copycat product performed. NANC, the exchange-traded fund built to mechanically follow disclosed Democratic trades, returned about 18.5% in 2025 against 17.9% for the S&P 500, a gap too small to call an edge once you factor in the risk of an undiversified, stock-picked portfolio. Its Republican-following counterpart, KRUZ, exists for the same reason and tells the same basic story: tracking Congress gets you something close to the index, not a shortcut past it.
Even setting that aside, a retail investor copying a disclosed trade today is working with the same structural issue as copying a 13F: the STOCK Act allows up to 45 days to disclose, and disclosures only report a dollar range rather than an exact trade, so you are acting on a rounded, weeks-old data point even in the best case.
The pattern underneath all three
Put these next to each other and the same shape keeps showing up. The edge sits with the person making the trade, at the moment they make it, because of something they know or believe that the market has not priced in yet. Disclosure exists to let the public see that trade, not to let the public capture the same return by copying it later. Insiders lose most of their edge within two business days. Hedge funds lose a large share of theirs within 45. Whatever Congress had appears to be gone entirely once you require a paper trail at all.
The exceptions in each case share something too. A multi-day insider cluster, a fund's single highest-conviction position, a genuinely unusual trade that breaks someone's own pattern, all of these are cases where the signal is strong or slow-moving enough to survive being seen late. A single Form 4, a fund's two-thousandth-largest position, or a routine trade timed to a schedule are not.
FAQ
Does copying insider buying actually work?
The research behind it is real, but a single insider's purchase is a weak signal on its own. What has held up better across studies is buying that breaks from an insider's normal pattern, and clusters of three or more insiders buying over several separate days rather than all in one day. Isolated, one-day clusters tended to show flat or negative returns once they were publicly disclosed.
Why doesn't copying a hedge fund's 13F filing work well?
Funds have up to 45 days after quarter end to file, and many use most of that window, so a position can already be closed by the time it becomes public. Research suggests the exception is a fund's single largest, most concentrated position held by a patient, low-turnover investor, which tends to keep working even with the delay.
Do members of Congress still beat the stock market?
The best recent published research says no. Two of the larger studies covering the years since the 2012 STOCK Act found that stocks members of Congress bought actually underperformed a matched benchmark afterward, with no measurable edge tied to committee seats or leadership roles.
Which is the most reliable of the three to follow?
None of them work well as a mechanical "buy what they bought" strategy. The versions of each that hold up in research all share a theme: a trade that is unusual for that person or fund, sustained rather than a single event, and concentrated rather than diversified.
Sources
- Lakonishok, J. and Lee, I., "Are Insider Trades Informative?", Review of Financial Studies 14 (2001)
- Seyhun, H.N., "Insiders' Profits, Costs of Trading, and Market Efficiency," Journal of Financial Economics 16 (1986)
- Cohen, L., Malloy, C., and Pomorski, L., "Decoding Inside Information," Journal of Finance 67 (2012)
- Kang, Kim, and Wang, "Cluster Trading of Corporate Insiders" (working paper)
- Antón, M., Cohen, L., and Polk, C., "Best Ideas," Review of Financial Studies (2021)
- Belmont, Sacerdote, and Van Hoek, "Relief Rally: Senators as Feckless as the Rest of Us at Stock Picking," NBER Working Paper 26975 (2020)
- Belmont, Sacerdote, Sehgal, and Van Hoek, "Do Senators and House Members Beat the Stock Market? Evidence from the STOCK Act," Journal of Public Economics 207 (2022)
- Eggers, A. and Hainmueller, J., "Capitol Losses: The Mediocre Performance of Congressional Stock Portfolios," Journal of Politics (2013)
- Our earlier coverage, Do politicians actually beat the stock market? What the data says
- Our earlier coverage, Every trade a CEO makes in their own stock goes public within 2 days. Here's how to read a Form 4
- Our earlier coverage, What is a 13F filing? How to track what hedge funds are buying (and what it leaves out)
This is general market commentary and opinion, not investment advice. Academic findings about average returns across many trades and many years do not guarantee any individual trade, filing, or copied position will be profitable. Always do your own research and consider speaking with a licensed financial professional before making any investment decision.
