Every quarter I get a version of the same message. Someone screenshots a wall of Form 4 filings from a single company, all dated within a few days of each other, all a few weeks after the earnings call, and asks whether this many insiders buying at once means something. It looks like conviction. It looks like a group of people who know something deciding to act together. And most of the time it is nothing of the kind. It is a calendar doing exactly what the calendar always does.
The reason this trips people up is that the interesting part of insider trading data is not really the trade. It is the timing of when the trade was allowed to happen at all. Once you understand that layer, a lot of clusters that look like signal collapse into noise, and a much smaller set of trades that most people ignore start to look worth reading.
How trading windows actually work
Almost every public company of any size runs an insider trading policy on top of the legal rules. The legal rules say you cannot trade on material non-public information. The company policy goes further and says, in effect, we do not trust anyone to judge that line in the heat of the moment, so we are going to lock the doors during the risky part of the quarter and only open them during the safe part.
The risky part is the blackout period. It typically starts some weeks before the quarter ends and runs until a couple of days after earnings are released. The logic is simple. In the run up to a quarterly report, insiders are the people most likely to know how the numbers are shaping up, so that is exactly when they are not permitted to buy or sell. The doors stay shut until the results are public and the market has had a moment to absorb them.
Then the window opens. This open window usually runs for a few weeks after the earnings release, often something like the two or three days after the report through roughly the middle of the following month, before it closes again as the next quarter winds down. During that stretch, an insider who wants to transact can. Outside it, they generally cannot, at least not without a pre-arranged plan.
So when you see ten Form 4s land in the same week after earnings, the first question is not why did they all decide to buy now. It is when else were they even allowed to. The honest answer is usually not much of the quarter. The window is short, everyone shares the same window, so of course the filings pile up in it. That is calendar mechanics, not a huddle.
Why the clustering means less than it looks
There are a few reasons the same-week pile-up is weaker than it appears, and it helps to hold all of them in your head at once.
- Everyone shares one clock. The window is company-wide. The CFO, three VPs, and a couple of directors are all governed by the same policy, so their filings naturally bunch even if no two of them ever spoke about it.
- A lot of it is not discretionary at all. Grants vest on schedule, options get exercised as they approach expiry, and shares get sold to cover the tax bill on vesting. These show up as Form 4s and they cluster in the window too, but they carry roughly zero information about what the insider thinks. The transaction codes on the filing tell you which is which, and most people never look.
- Pre-arranged plans muddy the timing. Many executives trade through plans set up in advance, so the actual execution date can be weeks after the decision was made. A sale that prints in the open window might reflect a choice made a quarter earlier for reasons that have nothing to do with the latest results.
Put those together and the post-earnings cluster is mostly a shadow cast by policy and payroll. It can still contain real buys from people making a real bet. But you cannot see those by counting filings, because the count is dominated by mechanics.
The trades that actually deserve weight
Here is the part that flips the whole thing around. If the window forces the routine stuff into a predictable slot, then anything that happens outside the expected pattern is, almost by definition, unusual. And unusual is where the information tends to hide.
The trades I stop and look at are the ones that had to fight the calendar to exist. A purchase that lands well away from the normal open window. A filing that shows up during what should be a blackout, which usually means it went through a special pre-clearance or a rare exception, and someone senior signed off on the awkwardness of that. An open-market buy where the executive spent real personal cash rather than exercising options, especially when it is not tied to any vesting event. Those are the trades where a person chose to act at a moment the system was nudging them not to, and that choice is the signal.
None of this is a guarantee. Insiders are wrong all the time, and a well-timed buy can still go nowhere. But the base rate is different. A routine cluster tells you the calendar turned. An off-cycle purchase tells you someone made a deliberate decision at an inconvenient time, and that is a much shorter list of people to pay attention to.
A workflow for reading a cluster
When a batch of Form 4s shows up and someone asks whether it matters, this is roughly the order I go through.
- Check the transaction codes first. Separate open-market buys and sells from grants, option exercises, and tax-withholding sales. If what is left is thin, the cluster is mostly mechanics and you can move on.
- Locate the trade in the cycle. Is this the normal open window right after earnings, or is it at an odd point in the quarter? Normal window plus routine codes is close to no signal. Odd timing raises the weight immediately.
- Ask whether a plan explains it. Filings often flag pre-arranged plan sales. If a sale was scheduled in advance, do not read it as a fresh opinion on the business.
- Weigh buys over sells, and cash over options. People sell for a hundred reasons and buy for basically one. A discretionary open-market purchase with personal money carries far more than a sale of comparable size.
- Then look at who and how much. Only after the above does the identity of the buyer and the size relative to their existing holdings start to mean much.
The failure mode I see most often is treating the raw filing count as the signal and skipping straight to who. You end up excited about six insiders buying when five of them were just vesting on schedule and the sixth was covering a tax bill. The count was real. The conviction was imaginary.
This is a lot of tedious cross-referencing to do by hand across dozens of names, which is a big part of why we fold window context and transaction codes into the disclosure feeds on Blockcircle rather than leaving you to eyeball a wall of filings. Whatever tooling you use, the mental model is the one that matters. Ask when the trade was allowed to happen before you ask what it means. Most post-earnings clusters are the sound of a door swinging open on schedule, and the trades worth remembering are the ones that happened when the door was supposed to be shut.