Four in Five Form 4 Filers Never Bought a Share

Of the 55,114 insiders who filed at least one Form 4 transaction dated in 2024 or 2025, 45,885 made no open market purchase in those two years. That is 83.25%, and it is a floor rather than a ceiling, for reasons set out below. Officers are the worst group, not the best.

They were filing the whole time. The letter A on a Form 4 means the share count went up, and it does not say why. Across 2026 so far, we hold 132,368 lines reporting an insider acquiring stock, and 12,554 of them carry code P, the purchase code. Most feeds surface the direction and let the reader supply the word “bought”. Our position: printing a code A grant and a code P purchase under one heading called insider buying is a category error, and it is the single most common way a Form 4 gets misread.

Every figure here comes from the 3.5 million Form 4 transactions we hold, filed between January 2015 and 21 August 2026, and was measured on 22 August 2026. The window behind each number is stated with it, because two of them are not the same window and the difference matters.

How many Form 4 filers never buy their own company’s stock?

Take every insider with at least one Form 4 transaction dated in 2024 or 2025. 55,114 people. 9,229 of them made at least one open market purchase, counting non-derivative code P rows only, which is the narrowest available reading of a purchase. 45,885 did not.

The rows lean the same way. Those two years produced 600,051 transaction rows, and 493,643 of them, 82%, came from the people who never bought. Four in five of the names moving through an insider feed are generating paperwork, not positions.

The obvious objection is that the non-buyers filed once and vanished, and that the serious insiders are hiding underneath them. Raise the bar and the figure holds. Of the 6,742 insiders with 20 or more transaction rows in those two years, 83.46% never bought. Of the 1,329 with 50 or more, 81.57%. Screening for activity does not find you buyers. It finds you people with a lot of paperwork.

Do officers buy more often than directors?

They buy less often, while filing three times as many rows at the median.

Role, 2024 to 2025	People	Never bought	Share	Median rows
Officers	28,776	24,591	85.46%	9
Directors, non-officer	22,994	18,967	82.49%	3
10% owners only	2,590	1,790	69.11%	3

Roles are assigned officer first, then director, then 10% owner, so nobody is counted twice. The table covers the 54,360 people who carry one of the three.

The natural counter is that officers are paid in stock, so grants and vesting bury whatever purchases they make. It fails on its own arithmetic. Never bought is a yes or no per person, and volume cannot dilute a yes or no. Nine median rows against three give an officer three times as many chances for one row to be a P. More filing should surface more buyers. It surfaces fewer.

One group sits apart. Among people flagged only as 10% owners, 69.11% never bought, sixteen points below the officers. Those are largely funds and holding entities rather than employees, and they are the one cohort whose stake did not arrive as pay.

What does “Acquired” mean on a Form 4?

It means the share count went up. That is all it means. Table I carries a column headed “4. Securities Acquired (A) or Disposed of (D) (Instr. 3, 4 and 5)”. The letter in it is a direction flag. The decision sits in a different field on the same row, the transaction code, which the form defines on itself in General Instruction 8.

Code	What the form calls it	Who is on the other side
P	Open market or private purchase of non-derivative or derivative security	The market, or a private party
A	Grant, award or other acquisition pursuant to Rule 16b-3(d)	The issuer
M	Exercise or conversion of derivative security exempted pursuant to Rule 16b-3	The issuer
F	Payment of exercise price or tax liability by delivering or withholding securities incident to the receipt, exercise or vesting of a security	The issuer
S	Open market or private sale of non-derivative or derivative security	The market, or a private party

Definitions from General Instruction 8, code letters moved into their own column (SEC, Form 4); the counterparty column is ours.

An A in column 4 can come from a P, from an A or from an M. The letter cannot tell you which. Congressional disclosure loses its meaning at a different point in the document, but the failure rhymes: the amount is a bracket and the date is a deadline.

How many Form 4 acquisitions are actual purchases?

Fewer than one in ten. Every code level figure here comes from a single window, transactions dated in 2026 up to 21 August, so every share below is measured against the same set of filings.

Of the 132,368 acquisition lines in that window, 61.45% are code A, a grant or an award, and 23.93% are code M, an exercise or conversion. Code P accounts for 12,554, so 90.52% of acquisition lines record something other than a purchase. The document level gap is wider, because a filing usually bundles several lines: 91,780 filings report an acquisition and 7,952 contain a code P anywhere. That leaves 91.34% of filings showing an insider acquiring stock with no purchase in them at all.

Code A names the rule it claims exemption under, and the rule text settles what kind of event it is:

(d) Acquisitions from the issuer. Any transaction … involving an acquisition from the issuer (including without limitation a grant or award) … shall be exempt if: (1) The transaction is approved by the board of directors of the issuer, or a committee … composed solely of two or more Non-Employee Directors … (17 CFR 240.16b-3)

The conditions run past what we quoted, but the first line sets the shape: an acquisition from the issuer, signed off by the board or a committee of non-employee directors. The insider’s part was to receive it.

Code M is a different event, and money does move on it. The price field on an M line is the exercise price the insider actually pays. It is not a price they went out and met in the market that day.

Does a disposal on a Form 4 mean the insider sold into the market?

Often not, and the codes say which cases are which.

Start with the pattern that looks least like compensation. Across 2026 so far, officers and directors account for 28,330 sale days, counted once per person per date. 5,181 of those days, 18.29%, also carry an option exercise by the same person on the same date. On 82.80% of those 5,181 days the person sold no more shares than they exercised that day. That is one event wearing two rows, and a feed that prints the sale and drops the exercise has shown one leg of two.

The second pattern never reaches the market. Rule 16b-3 exempts “the disposition to the issuer of issuer equity securities” where “the terms of such disposition are approved in advance” (17 CFR 240.16b-3). To the issuer, approved in advance. That is code F: shares handed back to the company to cover the tax on a vesting event, or to pay an exercise price. Of the 124,861 filings in that window, 11,038 contain nothing but code F, which is 8.84%. Roughly one filing in eleven discloses a payroll mechanism and nothing else.

Could the real figure be higher than 83.25%?

Yes, and we publish the lower one on purpose.

Every choice in the definition points the same way. A purchase counts only as a non-derivative code P line, which is the narrowest reading available. If someone’s only activity in the window is in derivatives, they still count as a filer and they still count as not having bought. Both decisions can push 83.25% down and neither can push it up, so treat it as a floor: the share who never bought is at least that, and on a wider definition of a purchase it would be higher still.

We also quote windows rather than trends, and that is deliberate. Every code level figure here comes from 2026, stated as one window a reader can go and check. We publish no decade long trend line for insider buying, here or anywhere, because a line drawn across ten years of Form 4s mixes changes in how insiders file with changes in what they actually did, and almost nobody who publishes one separates the two.

One direction check, which is not a restatement of the headline. Across 2026 up to 21 August, 44,642 insiders filed, 4,073 bought, and 90.88% did not. That is roughly eight months against twenty four, so it gives everyone less time to buy and the two figures are not comparable. Narrow it to people with non-derivative activity and it is 89.89% of 40,276. The direction holds in both windows.

How do you read a single Form 4?

Four checks, all answerable from the document itself:

Read the transaction code before the A or D letter. The letter says which way the share count moved. The code says what moved it.
Look for a P anywhere in the filing. If no line carries one, nothing was bought on the market, whatever the summary above it says.
On a disposal, check whether every code is an F. Then the counterparty is the company and the size follows from a vesting schedule.
When a sale shares a date with an exercise, compare the share counts before reading the sale as a decision to cut a position.

Per person code histories sit in our insider directory, and our methodology page says which codes we count and which we drop. If you would rather have the feed itself filtered this way, code P and nothing else, that is what an account is for. Either way, next time you see a filing quoted as insider buying, open the original on EDGAR and find its transaction code. One letter decides whether the story survives.