How to Read Insider Buying Signals

An insider purchase is one of the most-watched data points in fundamental investing - it is the rare case where someone with an information advantage is required to tell you, in public, that they just bet their own money on the stock. But the raw Form 4 feed is noisy: it mixes equity grants, tax-withholding sales, and pre-scheduled plan trades in with the genuine conviction buys. Here is how to separate signal from noise.

1. Look at the role

Purchases by the CEO, CFO, or Chairman have historically outperformed buys by less-informed insiders - the CFO in particular tends to know the numbers first. 10% owners such as activist funds also carry weight, because they make large, research-driven directional bets. InsiderAlpha weights officer buys (1.5×), 10% owner buys (1.6×), and director buys (1.2×) accordingly.

2. Size relative to compensation

A $50,000 purchase from an executive earning $20M a year is a rounding error. The same $50,000 from a small-cap director earning $200K is a real statement of conviction. Always read the dollar value of a buy against the insider's pay and existing stake, not in isolation.

3. Cluster buying

When two or more insiders buy in the same short window, the signal is far stronger than any single purchase. Cluster buys frequently precede positive earnings surprises and strategic announcements because they suggest the insiders are reacting to the same shared, non-public read on the business.

4. Filter out Rule 10b5-1 plan trades

Pre-scheduled Rule 10b5-1 trades are calendar-based, not opportunistic - the insider set them up months in advance. InsiderAlpha detects 10b5-1 markers in Form 4 footnotes and flags them so they don't pollute your buy signal.

5. Recency and liquidity

The edge decays fast. A 24-hour-old buy in a liquid mid-cap is actionable; a 30-day-old buy in a thinly-traded micro-cap is not. We apply a recency bonus that decays over 48 hours and require real liquidity (50-day average volume ≥ 100,000 shares) before a name reaches the plan.

The base rate: buying is rare, and that is the point

Across the Form 4 filings InsiderAlpha tracks, insider sales dwarf insider buys. In the 12 months through late July 2026, insiders reported roughly 74,000 open-market sale transactions worth about $171 billion, against roughly 15,900 open-market purchases worth about $39 billion. That is more than four sales for every buy.

This imbalance is structural. Most insider equity arrives as compensation, through grants and option exercises, so selling is the default behavior for an executive managing personal finances. An open-market purchase reverses that default: the insider is choosing to increase exposure with outside cash. The scarcity of real buying is what gives the signal its power, and it is why a screen built on purchases alone starts from a much cleaner sample than one that tries to interpret the flood of sales.

How much time do you have to act?

Less than most people assume. In InsiderAlpha's 2026 data, the median open-market purchase is filed 2.8 calendar days after the trade, and 90% are filed within 7 days. The disclosure itself is therefore already a few days behind the insider's decision. Academic work and our own evaluation data both point the same way: the abnormal return after disclosure decays quickly, with the strongest edge in the first hours and days after the Form 4 hits EDGAR. Watching filings in real time, rather than reading a weekly digest, is not a nice-to-have. It is most of the trade.

Putting it together

The highest-conviction setup combines all five: a large, open-market purchase (code P), by a top officer or 10% owner, that is meaningful relative to their pay, made alongside other insiders, in a liquid stock, within the last day or two - and not flagged as a 10b5-1 plan trade. That is exactly the combination InsiderAlpha's score is built to surface.

What to ignore

Equity awards (code A), option exercises paired with same-day sales (M + S), and tax-withholding dispositions (F) are compensation mechanics, not signals. Routine, scheduled selling is also weak - insiders sell for diversification, taxes, and liquidity all the time.

A quick pre-trade checklist

1. Is it an open-market purchase (code P), not a grant, exercise, or tax-withholding event?
2. Is the buyer a senior officer, director, or 10% owner rather than a peripheral insider?
3. Is the dollar amount meaningful against the insider's likely pay and existing stake?
4. Is the trade free of a Rule 10b5-1 plan marker?
5. Did other insiders at the same company buy in the same window?
6. Is the filing fresh, ideally under 48 hours old?
7. Is the stock liquid enough to enter and exit without moving the price?

A purchase that clears all seven checks is uncommon. In a typical week the raw feed contains hundreds of purchase rows, and only a handful survive this filter. That is the correct outcome: the goal is a short list you can actually research, not a long list that flatters the screen.

How InsiderAlpha turns this into a score

The platform's confidence score is a direct implementation of the factors above. Role sets the multiplier: officers 1.5x, 10% owners 1.6x, directors 1.2x. Dollar size maps to a base score that rises from 20 points at $10K to 95 points at $10M and above. A recency bonus of up to 15 points decays linearly over the first 48 hours after filing. A cluster bonus adds 15 points when two or more insiders buy the same ticker within 48 hours. Technical checks add smaller amounts when the setup is orderly, for example an RSI between 30 and 70.

Hard filters run before any scoring. Purchases under $10,000 are excluded, and so is any name without real market data or without at least 100,000 shares of 50-day average volume. A high insider-conviction score can never override a failed liquidity check, because a signal you cannot exit is not a signal.

Common mistakes

• Treating every Form 4 as a buy signal. Most Form 4 rows are compensation mechanics or sales. The purchase code alone removes the majority of the feed.
• Chasing stale filings. A two-week-old buy has already been seen, and mostly priced, by everyone watching EDGAR.
• Ignoring liquidity. Insider buying in a thin microcap can be impossible to trade at anything near the insider's price. This is why InsiderAlpha excludes names that fail its volume and volatility checks entirely.
• Reading 10b5-1 sales as bearish. Half of insider selling is pre-scheduled. A calendar decided the timing, not the insider's outlook.
• Anchoring on share count instead of dollars. 100,000 shares of a $0.50 stock is a $50,000 bet, not a large one.

Frequently asked questions

Do insider buys actually predict returns?

On average, yes, modestly, and with wide variation. The academic evidence (Lakonishok & Lee 2001; Cohen, Malloy & Pomorski 2012) finds the effect concentrates in opportunistic, non-routine purchases. See our evidence summary for details.

What counts as a large purchase?

Context decides, but as a rough scale InsiderAlpha's scoring treats $10K as a minimum worth looking at, and dollar size scores rise steeply through $100K, $1M, and beyond. Below $10,000 we exclude the trade from scoring altogether.

How many purchases are pre-scheduled plan trades?

Few. In the first quarter of 2026, only 2.6% of the open-market purchases we tracked carried a Rule 10b5-1 marker, versus roughly half of insider sales. Purchase signals survive plan filtering largely intact.

See this week's top insider buys → · Are insider buys actually bullish? →

This article is informational and is not investment advice.