TL;DR: Insider buying is one of the most consistently profitable signals in equity research, but only when filtered correctly. Most insider transactions are noise: option exercises, RSU vesting, tax-driven sales, and routine sells by long-tenured executives. The signal lives in a small subset of patterns, including cluster buys with the P transaction code, purchases scaled large relative to compensation, and first-time buyers acting against price weakness. This guide covers the SEC Form 4 fields that matter, the patterns academic research has validated, and how AI tools like NowNews surface meaningful insider activity automatically.
If you want to test AI-powered insider activity tracking on a watchlist, start a free 7-day NowNews trial no credit card required.
Insider transactions are one of the few publicly available data sources where the people generating the signal are also the people who know the most about the underlying business. Officers, directors, and 10% shareholders are required to disclose every trade in their company's stock within two business days of execution, through SEC Form 4 filings. The filings are free, the database (EDGAR) is free, and the academic literature is consistent: insider buying predicts positive future returns more reliably than almost any other public-data signal.
The problem is that the raw data is overwhelming. Hundreds of Form 4 filings hit EDGAR every trading day, the vast majority of which are not signal. Most insider transactions reflect routine compensation events: option exercises, RSU vesting, tax withholding, scheduled 10b5-1 plan sales. These transactions are mandatory disclosures with no informational content. The skill in tracking insider trading is filtering for the small subset of trades where an insider is making a genuine, conviction-driven decision with their own money.
This guide is the practical workflow for that filtering. It covers the Form 4 fields that matter, the patterns that validate as signal, and the patterns that look impressive but are actually noise.
Why insider buying predicts returns (and selling usually does not)
Academic research dating back to the 1960s consistently shows that insider buying outperforms the broader market. The intuition is straightforward: executives know more about their company than any outside analyst, and when they buy with their own money, that decision reflects information that is often not yet reflected in the price.
Insider selling is much harder to interpret. Insiders sell for many reasons unrelated to their outlook on the company:
- Diversification of concentrated wealth
- Tax planning at year-end
- Personal expenses (homes, college, divorces)
- Estate planning
- Liquidity needs
- 10b5-1 plan execution scheduled months in advance
- Tax withholding when options or RSUs vest
A CEO with 90% of their net worth in company stock who sells $5 million is probably diversifying, not signaling concerns. The same CEO buying $5 million is making a directional bet, because they did not have to do anything. Buying is intentional. Selling is often mechanical.
This asymmetry is why insider tracking platforms and academic papers focus heavily on buying activity. The selling signal exists, but it is much weaker on average and requires more context to interpret correctly.
The SEC Form 4 fields that actually matter
A Form 4 filing has more than a dozen fields, but only a few carry interpretive weight. Understanding which fields matter saves enormous time when scanning hundreds of filings.
The transaction code (column 3)
The transaction code is the single most important field on the form. It tells you what kind of transaction the filing represents:
| Code | Meaning | Signal value |
|---|---|---|
| P | Open-market purchase | High. The insider paid market price with their own money. |
| S | Open-market sale | Mixed. May reflect outlook or unrelated personal reasons. |
| A | Grant or award | None. Routine compensation. |
| M | Exercise or conversion of derivative | None. Routine. |
| F | Tax withholding (shares surrendered) | None. Routine. |
| G | Gift | None. Estate planning. |
| J | Other | Variable. Read the footnotes. |
Code P is what insider trackers screen for. It represents a deliberate decision to allocate personal cash to the company's stock. Codes A, M, F, and G are filings the SEC requires but contain almost no informational signal about management's view.
This single distinction filters out roughly 80 to 90 percent of all Form 4 filings as noise. Free trackers like OpenInsider and SECForm4 let you filter by transaction code as a default search parameter.
The relationship code (top of filing)
The form indicates whether the insider is an officer, director, or 10% beneficial owner, and their specific title. This matters because not all insiders carry equal informational weight:
- CEO and CFO purchases carry the highest signal. These are the people closest to operational and financial reality.
- Independent director purchases carry moderate signal, especially for first-time buyers, but small purchases often reflect minimum ownership requirements set by the board.
- Officer purchases (VPs, divisional heads) carry context-dependent signal.
- 10% beneficial owner activity can be either an activist building a position (high signal) or a fund's mechanical rebalancing (no signal). Read the footnotes to distinguish.
The transaction amount and value
The size of the purchase relative to the insider's compensation is what separates token gestures from genuine conviction. A CEO earning $10 million annually buying $50,000 worth of stock is making a token gesture, often executed for optics. The same CEO buying $2 million is making a meaningful directional bet.
A useful rule of thumb: a purchase under 10% of the insider's annual compensation is usually optics. A purchase over 50% of compensation is a genuine conviction trade. The interesting middle ground is 10 to 50%, which requires context.
The post-transaction holdings
The form lists how many shares the insider holds after the transaction. A purchase that increases the insider's holdings by 5% is materially different from one that increases them by 50%. First-time buyers, where the post-transaction holdings appear modest because the insider had little before, are particularly notable. They suggest something has changed in the insider's thinking.
If you want to skip the manual filing analysis and let AI surface the meaningful patterns automatically, NowNews offers a 7-day free trial of the full platform.
The patterns that validate as actionable signal
Beyond the basics of P-codes and scaled-vs-compensation, certain specific patterns consistently produce above-average subsequent returns in academic and practitioner research.
Cluster buying
When multiple insiders at the same company buy stock within a short window (typically two to four weeks), the signal is much stronger than any single purchase. The interpretation is straightforward: independent decision-makers reaching the same conclusion almost simultaneously, using their own money, is unusual unless something is happening at the company that the broader market has not yet recognized.
A cluster buy involving the CEO, CFO, and two board members in the same month is one of the highest-confidence signals available in public data. Practitioner studies suggest these clusters precede positive 12-month returns roughly 65 to 75 percent of the time, well above market baselines.
Purchases against price weakness
Insiders who buy after their stock has declined significantly are signaling that the decline is overdone. They have nonpublic context that the outside market does not have, and they are putting personal money behind their disagreement with the price action.
This pattern is particularly powerful when:
- The decline was 20% or more
- The insider purchase occurs within 30 days of the price low
- Multiple insiders participate (cluster buying against weakness compounds the signal)
The reverse case (insiders buying into strength) is much weaker on average, often reflecting confidence-signaling rather than genuine value perception.
First-time buyers
An insider who has never previously bought stock in the open market and suddenly starts buying is a particularly noteworthy signal. The insider could have bought at any time previously and chose not to. The decision to buy now reflects either a change in the insider's outlook, a change in the price they consider attractive, or new information they want to position for.
First-time buyers tend to produce stronger subsequent returns than repeat buyers, partly because the signal is less likely to be a routine pattern and partly because the buyer's decision is by definition novel.
Purchases shortly before known catalysts
When insiders buy in the weeks before known catalysts (FDA decisions, contract awards, earnings, regulatory rulings), the trades are subject to additional scrutiny but are also informative. Insiders are explicitly prohibited from trading on material nonpublic information, but they can trade based on their general assessment of the business going into the catalyst.
The interesting cases are:
- Pre-earnings insider buys when the eventual report exceeds expectations
- Pre-FDA-decision insider buys at biotech companies with binary catalysts
- Pre-contract-announcement buys at defense or industrial names
These transactions are heavily monitored by the SEC. Insider buys very close to a major announcement that subsequently makes the buyer significant gains can trigger investigations. Insiders generally know this and are cautious. The buys that do happen in pre-catalyst windows therefore tend to be either innocuous (general optimism, not material info) or eventually problematic for the insider.
Activist and institutional 13D/13G filings
While not Form 4, the related Schedule 13D and 13G filings disclose any investor crossing the 5% ownership threshold. A 13D specifically signals an active investor (often an activist) intending to influence the company; a 13G signals a passive investor (often an index fund or large asset manager).
A 13D filing from a known activist is a major catalyst event in itself. The market typically prices the stock higher on the assumption that the activist will push for changes that unlock value. Tracking 13D filings is a separate workflow that complements Form 4 monitoring.
The patterns that look impressive but are noise
Equally important is knowing what to ignore. Several patterns look like signal but are actually mechanical or compensation-driven:
Option exercise followed by immediate sale (cashless exercise). Code M followed shortly by code S. This is routine compensation realization and contains no directional signal.
RSU vesting with tax-withholding sale. Code F or A followed by S, often on the same day. Mandatory tax withholding, no signal.
10b5-1 plan sales. These are pre-arranged sales scheduled months in advance under SEC Rule 10b5-1. The footnote on the form will indicate the plan adoption date. A sale executing under a plan adopted six months earlier reflects no current view.
Small "qualifying" director purchases. Independent directors often have minimum stock ownership requirements set by the board's governance policies. Small purchases that move them just over the threshold are governance-driven, not conviction-driven.
Sales for visible personal events. Sales coinciding with publicly known executive home purchases, divorces, or other major personal financial events are often unrelated to outlook.
Spousal or trust transactions. Form 4 requires disclosure of trades by spouses and certain trusts. These often reflect estate planning rather than the insider's market view.
A repeatable workflow for tracking insider activity
The buy-side and serious retail standard is roughly:
Step 1 — Set a watchlist. You cannot meaningfully monitor every Form 4 filing in the market. Define a watchlist of 30 to 100 names you actually follow.
Step 2 — Filter for transaction code P. Free tools (OpenInsider, SECForm4) and paid tools both let you screen by transaction code. Discard everything that is not P unless you have a specific reason to look at sales.
Step 3 — Apply size filters. Filter for transactions with dollar value above a meaningful threshold (commonly $250,000 for large-caps, $50,000 for small-caps). This eliminates the smallest-token purchases.
Step 4 — Check for cluster activity. For any name with a P-code purchase, check whether other insiders have also bought in the prior 30 days. Multi-insider clusters elevate the signal strength.
Step 5 — Scale against compensation. For the meaningful purchases, look up the insider's compensation in the most recent proxy statement. A purchase scaled large versus compensation is the highest-quality signal.
Step 6 — Cross-reference with price action. Note whether the purchase occurred during a price drawdown, a flat period, or a rally. Drawdown-context purchases are the strongest pattern.
Step 7 — Check for institutional 13D/13G activity. A coincident insider buying cluster and a new 13D filing from an activist is one of the strongest setups available.
This workflow runs in 15 to 20 minutes per name when there is activity to evaluate, and it scales to a daily review of any signals on a 50-name watchlist.
NowNews integrates insider activity surfacing into the broader news intelligence flow, so meaningful insider buying that coincides with news catalysts gets prioritized in the Impact Feed automatically.
Common mistakes when interpreting insider activity
Several patterns produce most of the avoidable misreads:
Treating all sales as bearish. Insider sales are weak signals on average, dominated by personal-finance noise. Selling clusters at extreme magnitudes can be informative, but most sales are not.
Treating small director purchases as conviction. Many director purchases are governance-driven minimums. Look at scale relative to the director's wealth and prior pattern.
Ignoring 10b5-1 plan footnotes. A scheduled sale under a 10b5-1 plan adopted six months ago is mechanical. Always check the form's footnotes for plan-adoption dates.
Reacting to a single Form 4 in isolation. A single purchase is a data point. Patterns require context: compare against the insider's history, the company's recent insider activity, and the price action.
Forgetting that insiders can be wrong. Even high-quality insider buys precede negative returns roughly 30 to 35 percent of the time. The signal is probabilistic, not deterministic. Insider tracking improves win rates; it does not eliminate losses.
Anchoring on famous insiders. A CEO with name recognition is not necessarily a more accurate predictor than a CFO at a smaller company. The signal is in the pattern, not the person's brand.
How AI changes insider activity analysis
Manually tracking Form 4 filings even for a 30-name watchlist is feasible but tedious. The mechanical tasks (scanning filings, filtering by code, scaling against compensation, cross-referencing 10b5-1 plans) are exactly what AI tools handle well.
Modern insider activity analysis platforms automate four specific tasks:
Real-time filing ingestion. Every Form 4 in EDGAR is parsed within seconds of filing, structured into the relevant fields, and matched against watchlists. The analyst no longer scans EDGAR; they review pre-filtered alerts.
Automated signal scoring. Filings are scored on factors such as transaction code, size relative to compensation, timing relative to price action, and presence of cluster activity. Low-score filings get suppressed; high-score filings get surfaced.
Cluster detection. AI tools maintain a rolling 30-day window per company and detect when multiple insiders converge on buying activity, automatically flagging the cluster.
Cross-reference with news flow. Insider buying that coincides with news catalysts is a particularly valuable combined signal. Tools like NowNews overlay insider activity on news events, so a CFO purchase the day after an analyst downgrade becomes visible as a contrarian conviction signal.
Historical base rates. For each insider and each company, the system maintains statistics on how past similar trades performed. An insider with a track record of well-timed purchases carries different weight than one whose past trades were near peaks.
The combination of these capabilities reduces what was a 30-to-60-minute daily review across a watchlist to a 5-minute scan of pre-filtered, pre-scored signals. This is what makes systematic insider tracking practical for individual analysts and small funds.
Frequently asked questions
How quickly do insiders have to file Form 4 after a transaction?
Within two business days of the transaction. SEC rules require this rapid disclosure to ensure the public has near real-time visibility into insider activity. Filings happen at the end of the business day they are submitted, so a Monday transaction is typically visible to the public by Wednesday afternoon at the latest.
Is Form 4 data really free?
Yes. SEC EDGAR provides free real-time access to every Form 4 filing as it is submitted. Free trackers like OpenInsider, SECForm4, and StockInsider parse and display this data with basic filters at no cost. Paid platforms like InsiderScreener, GuruFocus, HedgeTrace, and NowNews add filtering, scoring, alerts, cluster detection, and news cross-referencing on top of the free underlying data.
What is a "cluster buy" and why does it matter?
A cluster buy occurs when multiple insiders at the same company purchase stock within a short window (typically two to four weeks). The signal is significantly stronger than any single purchase because independent decision-makers, each with personal capital, are reaching the same conclusion. Practitioner studies suggest cluster buys precede positive 12-month returns 65 to 75 percent of the time, well above market baselines. Cluster buys involving the CEO, CFO, and multiple directors are among the highest-confidence signals in public-data analysis.
Should I copy insider trades?
In general, no. Insider buying is informative but probabilistic, with a 30 to 35 percent failure rate even on high-quality signals. Treat insider buys as research starting points: they identify names worth investigating, not names to buy mechanically. Combine with fundamental analysis, valuation work, and an honest assessment of why the insider might be buying. Mechanical copying of insider trades has a much weaker historical track record than using insider activity as one input among several.
Are 10b5-1 plan sales informative?
Generally no. 10b5-1 plans are pre-arranged trading schedules adopted months in advance, designed precisely to insulate insiders from accusations of trading on material nonpublic information. A sale executing under a plan adopted six months ago reflects the insider's view six months ago, not today. Always check the Form 4 footnotes for plan-adoption dates. Plan-driven sales are nearly always noise, not signal.
Does NowNews track insider trading activity?
NowNews integrates insider activity into its broader news intelligence flow. When meaningful insider buying coincides with news catalysts (earnings beats, FDA decisions, contract awards), the combined signal is surfaced in the Impact Feed. The platform's Leader Statements module also tracks executive commentary that often correlates with insider trading patterns. The 7-day free trial includes full access to these features.
How do activist 13D filings differ from regular Form 4 filings?
Schedule 13D and 13G are filed when an investor crosses the 5% ownership threshold, not when an insider transacts. 13D specifically signals an active investor (often an activist) intending to influence company strategy; 13G signals a passive investor (often an index fund). Both are major events. A 13D from a known activist is itself a market-moving catalyst, often producing 10-30% one-day moves. Form 4 (insider trades) and 13D/13G (large external positions) are complementary signals: the former tracks management conviction, the latter tracks external pressure.
Can insider buying be wrong?
Yes. Even high-quality insider purchases (cluster buys with large size relative to compensation) precede negative returns roughly 30 percent of the time. Insiders are sometimes overly optimistic about their own company. They can be wrong about industry-wide cyclical pressure. They can buy into a fundamentally weakening business out of loyalty or sunk-cost reasoning. Insider tracking improves win rates above market baselines; it does not eliminate losses. Use it as a probability-shifting input, not a deterministic signal.
The bottom line
Insider trading data is one of the few publicly available signals where the people generating the activity also know the most about the underlying business. The signal is real, but most of the raw data is noise: option exercises, RSU vesting, tax-driven sales, and 10b5-1 plan executions contain no directional content. The skill is filtering for transaction code P, scaling against compensation, identifying cluster buys, and giving extra weight to first-time buyers and purchases against price weakness.
Doing this manually for a 30-name watchlist is feasible but tedious. AI tools that ingest Form 4 filings in real time, score them by signal strength, detect clusters automatically, and cross-reference with news flow turn a 30-minute daily review into a 5-minute scan of pre-filtered alerts. NowNews integrates this kind of insider activity surfacing into its news intelligence flow, so meaningful insider buys that coincide with news catalysts get prioritized automatically.
If you want to see AI-powered insider activity tracking in practice, NowNews offers a 7-day free trial of the full platform. Try it on a watchlist and compare the AI-surfaced signals against your own EDGAR scanning.
This article is updated as insider trading patterns and tools evolve. Last reviewed: April 2026. Have a specific insider activity case you think should be discussed? Contact us.