Form 4 Derivative Table vs Non-Derivative: Table I and II

Why Form 4 has two tables in the first place

Section 16 of the Securities Exchange Act requires officers, directors, and 10 percent owners to report changes in beneficial ownership on Form 4 within two business days of the transaction. The form's designers faced a structural problem: insiders do not just own shares, they own rights to shares. A stock option, a warrant, and a convertible note all change an insider's economic exposure without changing their common share count.

So the form separates the two exposure types. Table I covers non-derivative securities: instruments held outright, almost always common stock. Table II covers derivative securities: anything whose value derives from an underlying security and converts into it at a stated price on a stated schedule. The columns differ because the questions differ. For stock you need the date, transaction code, share count, price, and shares owned after. For a derivative you also need the conversion or exercise price, the date it becomes exercisable, its expiration date, and the title and amount of the underlying shares.

The split also reflects the filing family's division of labor. Form 4 captures transactions within two business days. Form 3, due within 10 days of becoming an insider, snapshots initial holdings across both categories. Form 5, due within 45 days of fiscal year end, sweeps up exempt or missed transactions. Compare that cadence with institutional disclosure, where a 13F arrives 45 days after quarter end, and Form 4 looks like what it is: one of the freshest ownership signals in the public record. For how the forms fit together, see 13F vs 13D vs 13G vs Form 4, and the SEC's investor site keeps a plain overview of Forms 3, 4, and 5.

Table I vs Table II: what each one actually reports

The fastest way to internalize the difference is side by side:

Table I is where decisions live. A code P row means the insider wrote a personal check for stock on the open market, which is the single most studied event in the insider literature; academic work has generally found open market purchases to be the rows with predictive content, while sales are far noisier. A code S row needs context (diversification, taxes, a pre-set trading plan) before you read it as bearish. Each row also carries an ownership form flag, direct or indirect, so holdings through trusts and family vehicles are distinguishable.

Table II is where structure lives. A row there tells you the insider holds, received, or disposed of an instrument that can become stock: at what strike, from what date, until what expiry, and over how many underlying shares. None of that is a market order. Almost all of it is compensation machinery or capital structure mechanics moving on a schedule that was set years earlier.

Inside the derivative table: options, RSUs, and conversions

Three instrument families dominate Table II in practice. Employee stock options are the workhorse: granted with code A, exercised with code M (or X for certain exercises), and defined by the strike and expiration columns. Those two columns are the real economics. An option deep in the money that gets exercised years before expiry reads differently from one exercised in its final month, where the calendar, not a view on the stock, forced the action.

Restricted stock units are the messy case, and any parser that pretends otherwise will misclassify filings. Some issuers report stock-settled RSUs in Table I at grant, treating them as stock acquired with code A. Others report them in Table II as a derivative (a right to receive shares, conversion price of zero) that converts into Table I stock at vesting. Both treatments are accepted practice, which means the same economic event can appear in either table depending on the company's filing conventions. A tracker keyed to one pattern silently miscounts the other.

Conversions are the third family. When a convertible note or convertible preferred turns into common, Table II records the disposal of the convertible with code C and Table I records the acquisition of the shares. The headline share count can be enormous, and the insider decided nothing that week; the conversion terms were fixed when the instrument was issued. The same applies to warrants exercised by early investors and to dispositions back to the issuer under code D.

You can verify any of this against the raw documents through EDGAR full text search, or more simply on the SEC's filing index for the issuer.

One exercise, three rows: how naive trackers double-count

Follow a single routine event, an executive exercising options and selling part of the proceeds, through the form. The filing will typically show: a code M row in Table II disposing of the options, a code M row in Table I acquiring the underlying shares, and a code S or F row in Table I selling shares or surrendering them to cover taxes. One decision, three rows, shares appearing twice.

A parser that sums "shares acquired" across both tables counts the exercise twice and may report the event as net insider buying. A parser that counts row volume reports three transactions where one happened. A dashboard that ignores codes entirely shows a large acquisition followed by a sale and calls it suspicious. The common misreadings:

The fix is mechanical once you see it: treat Table I as the ledger of share-level activity, use Table II to classify why shares moved, dedupe M pairs into a single exercise event, and score conviction only on rows where the insider chose the trade and the price, which in practice means open market P purchases and context-checked S sales. This is exactly the filtering behind insider buying cluster analysis: clusters are only meaningful if every row in them survived this classification.

Pulling clean insider data into an agent

Parsing both tables, normalizing issuer-specific RSU conventions, and deduplicating exercise pairs is exactly the kind of work an LLM should not improvise at query time. Arkolith ingests Form 4 filings into a normalized stream (51,000+ insider transactions tracked alongside the Q1 2026 institutional dataset of 1,824 filers and 1.87 million long positions), with every datapoint carrying provenance back to its SEC EDGAR accession number, so an agent can cite the filing instead of hallucinating one.

From the REST side, resolve the issuer first, then walk its data:

curl -H "Authorization: Bearer YOUR_KEY" "https://arkolith.com/api/v1/search?q=NVIDIA"

To cross-check insider activity against the institutional book on the same name, pull a fund's holdings by CIK:

curl -H "Authorization: Bearer YOUR_KEY" "https://arkolith.com/api/v1/funds/1067983/holdings"

The same data is exposed as MCP tools, so a Claude or ChatGPT agent can call it natively; setup takes a few minutes via the quickstart. For human-readable views, the per-ticker insider pages (for example TSM) separate purchase signal from plumbing, and the stock pages put insider flow next to institutional ownership. One honest limitation to carry into any model: Form 4 tells you what insiders did, on a two business day delay, and Table II tells you what they hold the right to do. Neither tells you why, and sales in particular resist clean interpretation.

Frequently asked questions about the Form 4 derivative table

Are RSUs reported in Table I or Table II of Form 4?

It depends on the issuer's filing convention. Many companies report stock-settled RSUs in Table I at grant as acquired stock, while others report them in Table II as a derivative with a conversion price of zero that converts to common at vesting. Both are accepted practice, so a robust parser has to handle both patterns for the same economic event.

Does an option exercise count as insider buying?

Not in the conviction sense. A code M exercise converts a years-old compensation grant into shares at a pre-set strike, and it is frequently paired with a same-day sale or tax withholding. Academic work has generally found that open market purchases (code P) are the rows with predictive content, not exercises.

What does the conversion or exercise price column tell you?

It defines the real economics of the derivative position: how far in or out of the money the instrument is, and together with the expiration column, whether an exercise was discretionary or forced by the calendar. An exercise years before expiry on a deep in-the-money option reads very differently from one in the final month before expiration.

Can you ignore Table II entirely when tracking insiders?

Almost, but not quite. The trade signal lives in Table I, yet you need Table II to deduplicate exercise pairs, to classify why Table I shares moved, and to see structural context such as a large overhang of vested, exercisable options or expirations that will force activity. Ignore it and your Table I counts will be wrong.

This article explains public filings and data concepts. It is not investment advice.