How to Read SEC Form 4 Insider Filings

Anatomy of a Form 4 filing

A Form 4 filing has five sections that matter for analysis. The reporting person section identifies who filed: their name, their relationship to the company (officer, director, or 10%+ beneficial owner), and their title if they hold one. The issuer section identifies the company by name, ticker, and CIK number.

The transaction detail lives in two tables. Table I covers non-derivative securities, which is where common stock purchases and sales appear. Each row includes the transaction date, a single-letter transaction code, the number of shares, the price per share, and whether ownership is direct or indirect. Table II covers derivative securities: stock options, warrants, convertible notes, and restricted stock units. Derivative rows include exercise prices, expiration dates, and the number of underlying shares.

The footnotes section is where conditions and context hide. This is where 10b5-1 plan disclosures appear, where gift purposes are noted, and where vesting schedules are described. Shibui extracts five signal flags from these footnotes automatically, so you do not need to read them manually for most analysis.

Transaction codes explained

Every Form 4 transaction carries a single-letter code that classifies the type of trade. Eight codes cover the vast majority of filings. Understanding what each code means is the first step to filtering signal from noise.

For buy/sell signal analysis, code P (open-market purchase) is the strongest indicator because the insider chose to spend their own money. Code S (open-market sale) is informative but noisier, since many sales are pre-arranged or compensation-related. Codes A, M, G, F, J, and C are compensation or administrative events that dilute the signal if included in screening filters.

Five signal flags that separate noise from information

Not every Form 4 sale is a bearish signal, and not every purchase is a conviction buy. Shibui extracts five boolean flags from Form 4 footnotes that identify routine or non-discretionary transactions. Filtering on these flags strips away the noise and isolates the trades where the insider made an active decision.

The 10b5-1 flag alone accounts for 951,764 transactions, about 15% of all insider filings. Tax withholding adds another 210,000 (3%). Together, these two flags cover roughly 18% of all insider transactions. Filtering them out before screening for insider sentiment removes the single largest source of noise in the data.

Cluster buying: what 3+ insiders buying within 30 days means

A single insider purchase could be routine: a new director making an obligatory buy, or an executive exercising options and holding the shares. But when three or more insiders at the same company independently file Form 4 open-market purchases within a 30-day window, that is a stronger signal. Each person made a separate decision to risk their own capital at roughly the same time, which suggests shared conviction about the company's value relative to its current price.

Shibui detects cluster buying by scanning Form 4 filings across all 10,000+ NYSE and NASDAQ companies. It filters to transaction code P (open-market purchase), excludes derivative transactions, removes superseded (amended) filings, and groups by company within a rolling 30-day window. The result is a list of companies with recent multi-insider buying activity, ranked by the number of distinct buyers or total dollar value.

Walkthrough: reading a real Form 4

To see how parsed Form 4 data looks in practice, query a specific company's recent insider filings. Each record returned maps directly to a section of the original EDGAR document. The insider's name and title come from the reporting person section. The transaction date, code, shares, and price come from Table I (non-derivative securities). The ownership nature (direct or indirect, and if indirect, the entity name) comes from the ownership column. The signal flags (is_10b5_1, is_tax_withholding, and others) come from parsed footnotes.

A typical response includes a mix of transaction types. You will see code A (awards) when executives receive RSU grants, code F (tax withholding) when those RSUs vest and shares are surrendered for taxes, code M (exercise) when options are exercised, and code S (sale) when shares are sold on the open market. The signal flags tell you which sales were pre-arranged under a 10b5-1 plan and which were discretionary. For a company like MSFT, the majority of insider activity is compensation-related (codes A, F, M), with a smaller number of open-market purchases (code P) or discretionary sales.

Each record also includes an accession number that links back to the original filing on EDGAR. If you need the full document, including footnotes that go beyond the extracted signal flags, you can follow that link to the SEC filing directly.

Direct vs indirect ownership

Every Form 4 transaction is classified as either direct or indirect ownership. Direct means the insider holds the shares in their own name. Indirect means the shares are held through another entity on the insider's behalf. Common indirect ownership structures include family trusts, a spouse's brokerage account, LLCs, partnerships, and estates. The insider still has economic exposure to the position, which is why the transaction is reportable.

Of the 6.2 million priced insider transactions in the database, 876,000 (14%) are classified as indirect ownership. This distinction matters for two reasons. First, indirect transactions sometimes represent estate or tax planning rather than a view on the stock. A transfer from an insider's personal account to their family trust is technically a Form 4 event, but it does not change the insider's economic position. Second, some large holders structure their ownership through LLCs or holding companies, so filtering to "direct only" would miss their activity entirely. The ownership nature field tells you which entity holds the shares, so you can make that judgment per transaction.

Questions

What is SEC Form 4?

SEC Form 4 reports changes in ownership by corporate insiders: officers, directors, and 10%+ shareholders. It must be filed within two business days of the transaction. Each filing includes the insider's identity, their relationship to the company, the transaction date, shares traded, price per share, and a code indicating the type of transaction (purchase, sale, award, exercise, gift, or tax withholding).

What do Form 4 transaction codes mean?

Form 4 uses single-letter codes: P is an open-market purchase (the insider spent their own money), S is an open-market sale, A is a compensation award, M is an option exercise, G is a gift, F is tax withholding on vesting equity, J is another type of acquisition or disposition, and C is a conversion of derivative securities. For screening insider sentiment, P and S are the most informative codes.

How do I tell if insider selling is meaningful?

Shibui extracts five signal flags from Form 4 footnotes: is_10b5_1 (pre-arranged plan, 951,764 transactions), is_auto_plan, is_gift, is_estate, and is_tax_withholding (210,000 transactions). Filtering out transactions where any of these flags are true isolates the discretionary trades where the insider chose to sell outside of a scheduled plan or compensation event.

What is cluster buying on Form 4?

Cluster buying is when three or more insiders at the same company file Form 4 open-market purchases within a 30-day window. A single insider buying could be routine, but multiple insiders independently spending their own money at similar prices suggests shared conviction about the company's value. Shibui detects cluster buying across all 10,000+ NYSE and NASDAQ companies.

What is direct vs indirect ownership on Form 4?

Direct ownership means the insider holds shares in their own name. Indirect ownership means shares are held through another entity on the insider's behalf, such as a family trust, a spouse's account, or an LLC. Of the 6.2 million priced insider transactions in the database, 876,000 (14%) are indirect. Both are reportable because the insider retains economic exposure.