Reading Form 4 Filings: Which Insider Buys Predict Returns

Most insider buying is noise. The academic literature has been clear on this for decades: in aggregate, insider purchases show modest outperformance, but that average hides a wide spread between high-signal buys and filings you should ignore entirely. If you treat every Form 4 buy alert as bullish, you're going to underperform a coin flip.

Here's a framework for sorting the signal from the noise.

What a Form 4 actually reports

A Form 4 is filed within two business days of a transaction by an officer, director, or 10%+ holder. The filing itself is mechanical — what matters is the transaction code in the table at the bottom. The codes that show up most often:

• P — Open-market purchase. The insider wrote a check.
• S — Open-market sale.
• A — Grant or award (free stock from the company).
• M — Exercise of a derivative (usually options converting to common).
• F — Shares withheld by the company to cover taxes on a vest.
• G — Gift.
• J — Other (read the footnote).

Only P is a real purchase. Everything else is the insider receiving compensation, paying taxes, or moving shares between pockets. Financial media — and even some screeners — routinely report option exercises as "insider buying." They are not. An exercise often precedes a sale and tells you nothing about conviction.

Which Form 4 buys carry real signal

The research consensus (Lakonishok & Lee, Cohen-Malloy-Pomorski, and the long line of work since) points to a small set of attributes that separate predictive buys from noise:

1. Open-market (code P), not 10b5-1. A 10b5-1 plan is a pre-scheduled trading program set up months in advance. Buys executed under one are noted in the footnotes or a checkbox. These are not opportunistic — they're calendar-driven. Discount them heavily.

2. Material size relative to the insider's wealth. A CFO making substantial compensation who buys a small amount of stock is doing PR. A CFO buying a meaningful amount — or a significant percentage of their reported net liquid worth — is making a statement. The dollar amount in isolation tells you little; size relative to compensation and existing holdings tells you a lot.

3. The right insider. The hierarchy of predictive power, roughly:

• CFO purchases > CEO purchases. CFOs see the numbers earliest and tend to be more cautious traders.
• Operational officers (COO, segment heads) > board directors. Directors often buy for optics.
• Avoid weight on 10%-holder filings unless the holder is an unaffiliated outsider — a sponsor topping up its existing block is not the same signal as a fundamental-driven outsider taking a new position.

4. Cluster buying. One insider buying is interesting. Three or four insiders buying within a 30-day window — especially across functions (a CFO and a segment head and an independent director) — is the highest-quality signal in the dataset. Clusters historically outperform single-insider buys by a meaningful margin.

5. Buying into weakness, not strength. A purchase after a significant drawdown when the tape is ugly carries more information than a purchase made when the stock is breaking out. The insider is implicitly saying the market is wrong; that's a real call.

6. First-time or unusual buyers. An officer who has never bought stock on the open market in their tenure, suddenly putting in a six-figure order, is a different signal than a director who buys stock regularly at intervals as a matter of policy. Check the insider's prior Form 4 history — most filers have one.

Buys that look bullish but aren't

A few patterns that fool retail screeners constantly:

• CEO "confidence buys" after a guide-down. Often announced via press release the same day as a downward revision. These are PR. The CEO knows the optics matter and the personal cost is small relative to comp. Returns following these are roughly market.
• Director buys at IPO lockup expiry or in the first year post-IPO. Often part of a pre-arranged commitment.
• Buys by insiders at companies with heavy buyback programs. When the company is repurchasing stock, the insider's economic exposure is rising mechanically; a token purchase adds little.
• Spousal or trust filings without context. Read the footnotes. A transaction in a family trust the insider doesn't directly control is weaker than a direct holding.
• Buys made days before a scheduled earnings release. Counterintuitively, these are often less informative because the trading window opens routinely after earnings — anyone buying then is buying with the same public information you have.

A practical scoring rubric

When a Form 4 alert hits, run it through five questions:

1. Is the transaction code P? (If no, stop.)
2. Is it open-market, not 10b5-1? (Check the footnotes.)
3. Is the dollar size meaningful versus the insider's comp? (Use the latest proxy.)
4. Is the buyer a CFO, operational officer, or part of a cluster?
5. Is the stock weak or contested right now?

Three or more yeses, and the filing is worth diligence. One or two, it's noise. Zero, ignore.

This isn't a buy signal on its own — insider buying is a prior, not a thesis. It tells you someone with better information than you thinks the stock is mispriced. You still have to figure out why.

What to watch next

• Pull the last 12 months of Form 4s on any name you own. Look for the pattern, not the headline trade. SEC EDGAR is free; openinsider.com aggregates well.
• Read the proxy to ground-truth insider compensation before judging trade size.
• Set alerts for cluster buying rather than single filings — most paid services let you filter on this.
• Track follow-through. When you flag a high-signal buy, log it and check the stock 6 and 12 months later. Calibrating your own hit rate is more useful than any backtest someone else publishes.