Form 4 Transactions in 2025: Compensation, Market Trades, Ownership, and Filing Lag

The captured Form 4 corpus contains 382,436 transaction rows dated in 2025 across 169,322 filings. Grant or award code A accounts for 105,618 rows. Option exercise code M accounts for 83,712, sale code S for 83,119, tax-withholding code F for 50,506, and purchase code P for 22,570. Open-market or private purchases therefore represent 5.9 percent of the transaction rows. The extract covers 5,194 issuers and 48,636 reporting owners. Direct ownership appears on 328,138 rows. Median transaction-to-filing lag is two calendar days among valid non-negative date pairs, but calendar days do not measure Form 4 business-day compliance.

1. Cohort and counting rules

The analytical cohort includes transaction rows whose recorded transaction date falls within 2025. One filing can contain several rows, and one economic event can appear through acquired and disposed legs. Counts are therefore transaction-row counts rather than trades, people, or filings. The full source contains 970,528 rows, including sparse pre-2024 coverage, partial 2026 data, and invalid or future transaction-date values. Those periods remain outside the analytical cohort.

2. Transaction codes define the ledger

SEC code A identifies grants, awards, and other acquisitions. Code M identifies exercise or conversion of derivative securities. Code S identifies open-market or private sales, code P identifies open-market or private purchases, and code F identifies payment of exercise price or tax liability through securities. Codes D, J, G, and C describe additional dispositions, other transactions, gifts, and conversions. A code must be read with direction and security details.

3. Purchases are a small part of reported activity

Purchase code P accounts for 22,570 rows, or 5.9 percent of the cohort. Sale code S accounts for 83,119 rows, or 21.7 percent. Grants and option exercises together account for 189,330 rows, or 49.5 percent. Tax-related code F adds another 13.2 percent. The transaction record is therefore dominated by compensation, exercises, sales, and tax mechanics rather than purchases alone.

4. Acquired and disposed legs are separate rows

The cohort contains 180,300 acquired rows and 202,136 disposed rows. Option exercise code M contributes 41,593 acquired rows and 42,119 disposed rows. Grants are almost entirely acquired, while sales and tax withholding are almost entirely disposed. Summing all rows as independent insider decisions would therefore double-count some linked mechanics.

5. Filing, issuer, and owner counts use different units

The 382,436 rows occur in 169,322 filing accessions from 48,636 reporting-owner identifiers across 5,194 issuer identifiers. One owner can file for several issuers, and one filing can report several securities or transaction legs. Direct ownership appears on 328,138 rows, while indirect ownership appears on 54,298. These fields describe reported ownership form and do not establish beneficial-control structure.

6. Calendar-day filing lag is not a compliance test

Among 381,666 rows with valid transaction and filing dates and a non-negative interval, the median lag is two calendar days and the ninetieth percentile is four. A total of 241,837 rows, or 63.4 percent, fall within two calendar days. Form 4 timing requirements use business-day and transaction-specific rules, so this calculation describes recorded date spacing rather than legal compliance.

7. The extract cannot identify trading plans or motive

The captured columns include transaction code, date, security, shares, price, direction, ownership form, issuer, owner, accession, and source URL. They do not include filing footnotes or a validated Rule 10b5-1 plan indicator. The cohort is also not joined to clinical-trial readouts, earnings events, or market returns. It cannot establish whether a transaction was planned, discretionary, informed, or event related.

8. Supported interpretation

The record supports three conclusions within its field limits. Purchases are a small minority of 2025 transaction rows. Compensation, exercises, sales, and tax mechanics account for most rows, and exercises can produce linked acquired and disposed entries. Calendar-day filing lag is short for the median row but cannot determine compliance. Any event study must define transaction codes, join filings to external events, inspect footnotes, and avoid treating every row as an independent decision.

9. Limitations

Coverage is sparse before 2024, and partial 2026 data contain date anomalies. The 2025 cohort can include amendments and repeated economic mechanics across rows. Code labels simplify transaction instructions and require filing context. Price and share fields are stored as text and are not valued here. Calendar-day lag does not implement weekends, holidays, filing exceptions, or business-day rules.