A single corporate insider buying shares on the open market is interesting. Three insiders at the same company buying within the same week is a different kind of event. Both show up as Form 4 filings on EDGAR. Both are legal. Only one of them has a consistent track record in the academic literature as a signal worth paying attention to. This article explains why the cluster is the stronger signal, what makes it mechanically different from a lone purchase, and how to verify a cluster yourself using nothing but official SEC data.
This is not investment advice. It is an explanation of how a specific disclosure pattern works and why researchers have treated it the way they do.
What a Form 4 actually tells you
Officers, directors and holders of more than ten percent of a company's stock are required to report changes in their ownership to the SEC. The reporting form is Form 4, and under the rules adopted after the Sarbanes-Oxley Act it has to be filed within two business days of the transaction. The SEC describes the form and the obligations behind it on its Forms 3, 4 and 5 page at investor.gov.
The two-business-day deadline matters for this topic more than almost anything else. It is what makes clusters observable in near real time. If insiders had a month to report, a group of purchases spread across one week would blur into a month of filings and the clustering would be far harder to see. Because the window is short, the filing dates track the trade dates closely, and a burst of purchases shows up as a burst of filings.
Each Form 4 lists the transaction date, the transaction code, the number of shares, the price, and the insider's holdings after the trade. The transaction code is where most beginners go wrong. Code P means an open-market or private purchase. Code A means a grant or award. Code M means an option exercise. Code F means shares withheld for taxes. Only code P reflects an insider choosing to spend their own cash on the stock at the market price. Everything in this article refers to code P purchases.
Why a single buy is a noisy signal
Insiders buy for many reasons that have nothing to do with their view of the company's prospects.
Some companies have minimum ownership guidelines for directors. A newly appointed board member who buys shares in the first few months may simply be meeting a policy. Some executives buy a token amount after a sharp drop because it looks good in the press. Some buy to offset the optics of a large sale the previous quarter. Some are rebalancing a personal portfolio for reasons unrelated to the company.
Then there is the matter of size. A chief executive with tens of millions of dollars in stock who buys a few thousand dollars worth of shares is not expressing much conviction. The purchase is real and it is reported, but as a fraction of their existing exposure it is close to zero.
A single buy also has a selection problem. There is only one person's judgment behind it. That person may be wrong, may be motivated by something you cannot see, or may be the one optimist in a management team that is otherwise quietly worried.
None of this means single buys are worthless. Larger purchases by senior officers, especially after a period of no insider activity, have historically been more informative than small ones. But the signal-to-noise ratio of a lone filing is low, and treating every code P filing as bullish is a reliable way to get burned.
Why a cluster is different
A cluster is usually defined as several distinct insiders at the same company making open-market purchases within a short window, commonly a few days to a couple of weeks. The exact thresholds vary between researchers and between data vendors, and there is no official SEC definition. The idea is consistent though: independent decisions by independent people pointing the same way at the same time.
Several things change when you move from one buyer to several.
First, idiosyncratic motives wash out. One director may be meeting an ownership guideline. It is much less likely that the chief financial officer, two directors and the general counsel all have unrelated personal reasons to buy in the same week. When motives are diverse but behavior is identical, the common factor is usually a shared view about the company.
Second, information access is pooled. Different insiders see different parts of the business. The CFO sees cash flow and the balance sheet. A division head sees order books. An independent director sees board-level strategy and audit discussions. When people with different vantage points reach the same conclusion, the conclusion is better supported than when one person reaches it alone.
Third, the timing itself carries information. Insiders are restricted from trading during blackout periods around earnings and while in possession of material nonpublic information. Purchases tend to concentrate in open trading windows shortly after results are published. A cluster inside such a window suggests that several people looked at the same freshly released numbers, saw how the market reacted, and decided the reaction was wrong.
Academic work on this goes back decades. Studies of insider trading in aggregate have repeatedly found that purchases are more informative than sales, because sales have many liquidity-driven explanations while purchases have few. Later work went further and found that the informativeness of insider purchases rises when multiple insiders act together, and rises again when the buyers are not routine traders. The intuition is the same as above: consensus among informed people, acting outside of a predictable schedule, is a stronger signal than a single trade. Precise return figures depend heavily on the sample period and methodology, so this article does not quote them. The direction of the finding has been stable across studies.
The 10b5-1 complication
One mechanical detail cuts against a naive reading of clusters. Insiders can set up prearranged trading plans under Rule 10b5-1. Trades executed under such a plan were decided in advance and say nothing about the insider's current view. Since the 2023 amendments to the rule, Form 4 includes a checkbox indicating whether a transaction was made under a 10b5-1 plan, and the filing has to state the plan adoption date. The SEC's rule release and summary are available on sec.gov.
Most 10b5-1 plans involve sales, not purchases, so this is less of a problem for buy clusters than for sell clusters. It is still worth checking the box on each filing in a cluster. A cluster of three purchases where two were on a plan is really a cluster of one.
What a real cluster looks like on EDGAR
You can verify a cluster without any paid tool. EDGAR's full-text search and the company filing index both work. The process is tedious but straightforward.
- Open the company's filing list on EDGAR and filter to form type 4.
- Note the filing dates. A burst of several Form 4 filings within a few days is the first hint.
- Open each filing and check the transaction code in Table I. Discard anything that is not code P.
- Check the reporting person on each filing. Several filings from the same insider on consecutive days is one buyer making several trades, not a cluster.
- Check the 10b5-1 box and the transaction price. Purchases at the current market price are what you want. A purchase at a deep discount is often a private placement, not an open-market vote of confidence.
- Compare the shares bought to the holdings after the transaction. A buy that doubles someone's position means more than a buy that adds one percent to it.
If after those steps you still have three or more distinct insiders with code P purchases at market prices within roughly two weeks, you have a cluster in the sense researchers use the term.
Free aggregators also exist. OpenInsider and similar sites screen Form 4 filings and have cluster buy views. They are a reasonable starting point as long as you verify the underlying filings on EDGAR, since aggregators occasionally misclassify transaction codes or miss amendments.
Where clusters fail
Clusters are a stronger signal, not a guarantee. Some failure modes recur often enough to list.
- Insiders at a struggling company can be collectively wrong. Management teams that have run a business for a decade often underestimate how much the world has changed around them. Several of them buying together can reflect shared blindness rather than shared insight.
- Clusters in very small companies can be engineered. A few directors each buying a modest amount in a thinly traded stock can move the price and generate headlines. Size the cluster against the company's market capitalization and daily volume before taking it seriously.
- A cluster after a long decline is different from a cluster after a sharp one-day drop. The former often reflects a view that the market has slowly lost interest. The latter can be a reflexive response to a single bad headline and has a weaker record.
- Clusters that coincide with a dividend cut, a covenant waiver or a going-concern notice deserve extra skepticism. Insiders sometimes buy to signal confidence precisely when confidence is in short supply.
How this relates to other filing-based signals
The logic behind clusters shows up in other parts of the disclosure system too. Congressional stock trades are disclosed individually under the STOCK Act and its 45-day rule, which is covered in detail in the 45-day rule and why it matters. A single member of Congress buying a stock is noisy for exactly the reasons a single insider buy is noisy. Several members on relevant committees buying the same name within a reporting period is a very different pattern.
Institutional holdings follow the same logic at a slower cadence. A single fund adding a position in its quarterly 13F means little. Many unrelated funds adding the same position in the same quarter is the institutional version of a cluster. The reporting lag that makes 13F data slower to act on is explained in 13F deadlines and the 45-day lag.
In every case the principle is identical. Agreement among independent, informed parties carries more information than any one of them acting alone. Form 4 just happens to be the filing where that agreement becomes visible fastest, because the two-business-day deadline leaves almost no gap between the decision and the disclosure.
If you want the institutional version of this signal already assembled, the Smart-Money 13F Consensus report tracks which stocks multiple tracked funds added in the same quarter, drawn directly from EDGAR filings and updated each reporting cycle. You can see the current edition at /reports/13f.html.
Want the signal instead of the raw filings? Get a free report preview. Prefer the tool to the write-up? Browse all data feeds or connect the free MCP server.