How to read SEC Form 4 insider buying (and what's actually a signal).

If you've ever read a finance Twitter thread that ends with "the insider just bought 50,000 shares" and felt vaguely compelled to investigate further, you've encountered SEC Form 4. It's the filing every corporate insider — directors, officers, anyone with 10%+ ownership — must submit to the SEC within 2 business days of any trade in the company's stock. It's been mandatory since 1934. The signal is real. The noise around it is what kills retail traders.

This post is the field guide: what Form 4 actually contains, which 90% of filings to ignore, and which of the remaining 10% are worth reading closely. Tapeline's Smart Money sub-score — one of the six named factors in the composite (see the methodology) — does not apply these filters: it nets every disclosed transaction in its window. The logic below is for reading the filings yourself, whatever tool you use.

Corrected 17 September 2026: an earlier version of this post said the Smart Money sub-score does this filtering automatically, that Tapeline Premium shows "filtered" Form 4 activity across the full universe, that a trade is at most 48 hours old when its Form 4 reaches you, and that the filings reliably predict something. The sub-score nets every disclosed non-derivative Form 4 transaction in its window by signed dollar value, without filtering out 10b5-1 plan sales, option exercises or small trades; /app/holdings lists the newest Form 4 transactions for the stocks we score, with a purchases-only filter; a filing can reach Tapeline a few days after the trade; we have no evidence that Form 4 filings predict price moves; and it called codes P and S "open-market" purchases and sales, when each covers a private transaction too.

What Form 4 actually is

A Form 4 filing has six things you care about:

1. The insider's name and role — CEO, CFO, director, 10%+ owner. Role matters; we'll get to why.
2. Transaction code — a one-letter code from a fixed table. The ones that matter for "is this a signal": P (a purchase on the open market or in a private sale), S (a sale of either kind), A (grant — almost never meaningful), F (tax-withholding sale — almost never meaningful).
3. Number of shares — raw count, not dollar amount. You compute the $ from price.
4. Price per share — the executed price.
5. Date of trade — not the filing date. The filing window is 2 business days, so the trade can be up to two business days older than the filing.
6. Shares held after transaction — total post-trade. This is the field most retail traders ignore and the one that determines whether the trade is a signal or noise.

The 90% that's noise — what to ignore

1. Anything that isn't transaction code P or S

Form 4 has 30+ transaction codes. Most of them — grants, vestings, exercises, gifts, withholdings — are not voluntary market activity. An insider getting shares via an automatic restricted-stock vesting tells you nothing about their view of the company's valuation. They didn't choose to acquire the shares; the comp plan did. Filter to code P (a purchase) and code S (a sale) only. Everything else is HR paperwork dressed as a filing.

2. 10b5-1 sales

10b5-1 plans are pre-arranged trading schedules executives use to sell stock systematically without being accused of insider trading. A CFO who set up a 10b5-1 plan in March 2025 that triggers a sale of 10,000 shares every quarter is selling mechanically — it's not their reaction to current information. These show up as code S sales but are explicitly marked "pursuant to 10b5-1 plan" in footnotes. Filter them out; they're noise.

3. Tiny purchases relative to ownership

A board member who already owns 500,000 shares buying 100 more is not a signal. They're rounding error in their own portfolio. The relevant ratio is shares purchased ÷ shares held after. Anything under 1-2% is meaningless.

4. Director purchases at companies with mandatory share-ownership requirements

Many large companies require directors to own at least N× their annual cash retainer in stock. When a newly-appointed director makes a small open-market purchase, they're often just complying with the requirement — not expressing a view on valuation. The clue: it's their first purchase, it's small, and it happens within 90 days of board appointment.

The 10% that matters — what's actually a signal

1. Cluster buying

The single highest-information Form 4 pattern. Multiple different insiders — usually CEO + CFO + at least one director — all making open-market purchases (code P) within a tight time window (say 30 days), at meaningful sizes (1%+ of their existing holdings each). This is hard to fake and hard to explain via comp-plan mechanics. When you see it, the people closest to the company's actual numbers have collectively decided the stock is mispriced low.

2. CEO purchases at a meaningful percentage of net worth

If the CEO buys $1M of stock and their compensation suggests a net worth in the $50M range, that's 2% of their net worth in one position. That's a real bet. Cross-check via the proxy statement (DEF 14A) for total compensation history.

3. CFO buying when others are selling

The CFO has the cleanest, earliest view of the company's actual financials — quarterly closing, working-capital trends, cash-flow forecast. When a CFO buys against a sector tape that has other peers selling, it's an unusually strong dissenting signal.

4. First-time purchases at companies that haven't seen insider buying for 12+ months

Companies in steady-state mode often see zero insider open-market activity for stretches. When that dry spell breaks with a meaningful purchase, something has changed in management's view. Pull the most-recent 4-5 Form 4s and check the gap.

How Tapeline scores this automatically

Tapeline's Smart Money sub-score — one of the six named factors in the composite — nets the disclosed Form 4 purchases and sales in a recent rolling window by their signed dollar value, and maps that balance onto a 0-100 scale. What it reads, and the statutory lags it inherits, are set out on the Smart Money factor page.

The result is a 0-100 sub-score that lands in the composite. Recent insider buys are also displayed on Tapeline Premium at /app/holdings — Form 4 data per ticker, sorted by date. Read alongside the composite score, not in place of it.

Where the signal breaks down

Three honest caveats:

• Filing lag. By the time a Form 4 is filed, the trade can be up to two business days old, and Tapeline re-checks each stock on SEC EDGAR about every two days, so on Tapeline it can be a few days older still. The market often already moved.
• Selling is less informative than buying. Insiders sell for personal reasons (diversification, house purchase, divorce) that aren't tied to their view of the company. Buying is almost always a directional view; selling is mixed.
• Small-cap signal-to-noise is worse than large-cap. Microcap insiders trade more frequently for personal-liquidity reasons. Looking for cluster buying helps but doesn't eliminate the noise.

The pitch

Form 4 is one of the few insider disclosures retail traders can read in full — the raw data is public, the filing is mandatory, and most retail traders don't read it. The hard part isn't access; it's filtering, and that part stays with you. Tapeline surfaces the filings as (1) a sub-score in the composite, which nets every disclosed transaction without the filters above, and (2) on Premium, the transactions themselves at /app/holdings, where you can narrow the list to purchases (code P).

30-day Premium trial — $0 today. Browse the most recent Form 4 transactions across the stocks we score, or filter to one ticker for its last 90 days. The trial takes a card; the first charge is on day 30 and one click cancels before then.