The SEC has made insider transaction data publicly available for decades. It's free. It's updated within two business days of every trade. It represents the disclosed buying and selling activity of people who, by definition, know their companies better than anyone on the outside.
And most investors either don't use it at all, or use it badly.
The "don't use it at all" crowd usually has one of two objections: they've heard that insiders are often wrong, or they think the data is too hard to parse on a raw EDGAR search. Both objections are at least partially right, which is exactly why it's worth learning how to do this correctly instead of either ignoring it or naively following every purchase.
FIRST: STRIP OUT THE NOISE
Most of what shows up in raw Form 4 data is not a signal. Before you can use the data, you need a reliable way to filter for transactions that represent actual discretionary conviction rather than routine compensation events or tax mechanics.
| Transaction | Code | Worth tracking? |
|---|---|---|
| Open market purchase | P | Yes: the cleanest bullish signal |
| Open market sale | S | With heavy context (see below) |
| Stock award / grant | A | No: compensation event, not a choice |
| Shares withheld for taxes | F | No: automatic, not discretionary |
| Option exercise | M | No: mechanical, often followed by sale |
| Gift / charitable donation | G | No: not a market signal |
Once you're only looking at P and S codes, you've eliminated probably 70-80% of the noise. What's left is actually worth analyzing.
A PRACTICAL FRAMEWORK
Here's a step-by-step process for evaluating insider transactions that most people who've been doing this for a while converge on, in some form:
Identify the transaction type
Is it a P (open market purchase)? If yes, proceed. Is it anything else? Unless it's an unusually large S-code sale at a strange time, move on. Don't spend time analyzing grants, tax events, or option exercises. They're not telling you anything about conviction.
Check the title of the buyer
Not all insiders are equal. A CEO or CFO buying is more meaningful than a non-executive director. A director who also operates in the business (an executive chairman, for instance) carries more weight than a purely independent director. The closer to the operational reality of the company, the more the purchase reveals.
Size it relative to their position
A $100,000 purchase means something very different depending on who's making it. Look at the "securities owned after" field on the Form 4. If this purchase increases their holdings by 5%, that's a real commitment. If it's a rounding error on a position they've held for years, it's probably noise. Also consider their compensation: a $500K purchase from someone earning $2M/year is materially different from the same dollar amount from someone earning $200K/year.
Look for other buyers at the same company
This is the most important step. One purchase gets a watchlist flag. Multiple purchases at the same company within 30-60 days, especially from different executives or directors, is a cluster and deserves real attention. Check back through the last 60-90 days of filings before you make any judgment about whether a purchase stands alone or is part of a broader pattern.
Check when they last bought or sold
Track the insider's history. A first-time open market buyer (someone who has never bought stock in their own company before) is a more significant signal than someone who buys regularly regardless of price. Conversely, if an executive has been selling steadily for two years and just bought a small amount, that's weaker than it looks on the surface.
Understand the fundamental backdrop
Insider data is a signal, not a thesis. Before acting on anything, you need a basic answer to: why is this stock trading where it is, and why might it be wrong? Insiders buying after a sector selloff is different from insiders buying into ongoing fundamental deterioration. The data tells you they see value, but it doesn't tell you they're right, and it doesn't replace understanding the business.
THE SELLING QUESTION
Insider selling gets a lot of attention and is mostly misread. The standard line is true: there are a hundred reasons to sell stock and only one reason to buy. Diversification, a house purchase, a divorce, an estate plan, a donation. Insider sales are frequently mundane.
That said, there are patterns in selling worth paying attention to:
- Cluster selling from multiple insiders , especially outside of a scheduled earnings window, can be meaningful. The same logic that makes cluster buying significant applies in reverse.
- A CEO or CFO selling a large percentage of their total position is more notable than a routine trimming. If someone goes from holding 2 million shares to 400,000 in six months, that's worth investigating.
- Sales that happen right after a big run-up, timed precisely at local price highs, sometimes suggest the insiders know the momentum is likely to fade. Not always. Sometimes it's just good luck and portfolio management, but it's worth flagging.
The 10b5-1 problem: Many insider sales happen through pre-arranged 10b5-1 trading plans, where a broker automatically sells shares on a schedule set months in advance. These look like regular S-code sales in the Form 4. If you're analyzing selling, always check whether the sale notes a 10b5-1 plan in the footnotes. Automatic plan sales carry essentially no signal value.
WHAT INSIDER DATA DOES AND DOESN'T TELL YOU
It's worth being clear-eyed about the limits here, because overselling this data is what leads people to get burned.
Insiders know their company's internal state better than anyone. Backlog, customer conversations, margins, product pipeline, headcount trends. They have visibility that even the most diligent sell-side analyst doesn't. When they buy, they're often expressing a view on that internal reality.
Insiders don't know the macro, the timing, or what the market will care about. A CFO can be completely right that their company's underlying business is healthy and still watch the stock drop another 40% because interest rates moved, the sector rotated out of favor, or a competitor launched a product that changes the competitive dynamics. They see the inside of the company clearly and the outside not much better than you do.
The time horizon matters. Most academic research on insider trading looks at 6-12 month forward returns. That's the window over which the signal tends to be cleanest. If you're looking at this data and thinking in days or weeks, you're probably going to be disappointed, and probably frustrated.
Where insider data is most powerful: as a confirmation tool. You've done your fundamental work. You think the stock is undervalued. You see strong insider buying. Now you have two independent sources of signal pointing the same direction. That's a better entry point than either signal alone.
THE OTHER SIGNAL: Form 4 insider positions
Form 4 tells you what company insiders are doing. But there's a second category of ownership data that most retail investors overlook entirely: Form 4 filings. Every investment manager overseeing more than $100 million in equity assets is required to disclose their full portfolio to the SEC every quarter. Berkshire Hathaway, Citadel, Bridgewater, Appaloosa — all of them, every 90 days.
Unlike Form 4, which captures real-time transaction data, the Form 4 is a quarterly snapshot. It tells you what a fund held at the end of the quarter, typically disclosed 45 days after quarter-end. The lag matters — by the time you see it, the position may have changed — but the signal is still meaningful because it reveals conviction that was sustained across months, not a quick trade.
What makes Form 4 data useful
The real value isn't in any single Form 4 filing. It's in the change between quarters. When multiple institutions independently add to the same position in the same quarter — without coordinating — that convergence is a signal. Each fund made its own analysis, ran its own models, and arrived at the same conclusion: this stock is worth owning more of.
- New positions are the most interesting signal. A fund that has never owned a stock initiating a position means an analyst team did enough work to build conviction from scratch and push it through an investment committee. That's not a small thing.
- Additions (increasing an existing stake) suggest the fund's thesis is playing out, or they want more exposure at current prices.
- Lack of buying after a dip is worth watching for the inverse reason — if insiders aren't buying after a significant pullback, it may suggest they don't see value at current prices.
Form 4 vs Form 4: Form 4 tells you what insiders with deep company knowledge are doing with their own money right now. Form 4 tells you what professional investors managing billions decided was worth holding at the end of last quarter. Both signals are strongest when they point the same direction.
How to use it alongside Form 4
The most interesting setups are when executives and directors are buying concurrently — especially across different roles. A CEO and CFO both buying in the same week, in the open market, after a decline, is a powerful signal. These are people with full visibility into the business making personal financial bets.
Conversely, watch for divergence: insider buying alongside heavy options activity or unusual short interest can signal contested views on the stock. Context always matters — no signal is conclusive on its own.
The limits of Form 4 data
- The 45-day lag is real. Q4 data is disclosed in mid-February. A fund could have exited the position entirely in January. You're always looking at a historical snapshot, not a live view.
- Large passive funds are noise. When Vanguard or State Street shows up as a holder, it almost always reflects index fund flows, not a discretionary call. Filter on active managers with real track records.
- Small changes at large funds may be irrelevant. A fund adding 50,000 shares of a stock they hold 10 million of is not a meaningful signal. Look at percentage changes, not absolute numbers.Form 4 filings are the primary signal for tracking insider conviction. That kind of convergence is hard to manufacture.