Not All Insider Buys Are Equal: How We Filter Signal From Noise

Most insider-trading screeners have a dirty secret: they show you everything — routine purchases, pre-announced plans, calendar buys — and call it "insider activity." The academic evidence says this is worse than useless. It's actively misleading.

The Research Most People Miss

In 2012, Lauren Cohen, Christopher Malloy, and Lukasz Pomorski published "Decoding Inside Information" in the Journal of Finance. They split insider trades into two categories:

Routine trades — predictable, calendar-based purchases that happen at roughly the same time every year. A director buying $50K every March. A CFO adding shares every time his RSUs vest.

Opportunistic trades — irregular, conviction-driven purchases with no historical pattern. A CEO buying $800K for the first time in three years. A CFO doubling down after a stock drops 40%.

The results were stark:

Category	Monthly Alpha
Routine	0 bps
Opportunistic	82 bps (~10% annualized)

Zero. Routine insider trades generate literally zero excess returns. The entire edge comes from opportunistic trades.

Why This Matters More for Cluster Buys

Alldredge & Blank (2019) showed that insider purchases made within 2 days of a colleague's purchase generate 2.1% abnormal returns over the following month — nearly double solitary trades.

But here's what nobody talks about: that 2.1% figure blends routine and opportunistic traders together. If you could isolate clusters where every participant is making an opportunistic trade, the true signal is likely significantly higher than what's been published.

The research understates the opportunity because it doesn't isolate cluster + opportunistic together. That's the gap we're closing.

What "Routine" Looks Like in Practice

A director at a mid-cap buys roughly $47K every March — 2022, 2023, 2024, 2025, 2026. No variation. No reaction to price. Just a calendar event. This carries zero information content about the business.

Compare that to a CEO making her first purchase in three years — $800K — while the CFO independently buys $200K the same week. No historical pattern. No 10b5-1 plan. This is a high-conviction cluster buy.

Every insider-tracking tool on the market — OpenInsider, Finviz, even paid services — shows both of these identically. Same table. Same formatting. No label distinguishing routine from opportunistic.

How We Filter

ClusterDesk applies two filters before a cluster ever reaches your inbox:

1. 10b5-1 Plan Detection

Pre-announced trading plans (Rule 10b5-1) are by definition uninformed. The insider committed to the trade months ago — it carries no information about current business conditions.

We scan SEC Form 4 footnotes for 10b5-1 plan disclosures and exclude those filings from cluster detection. A "cluster" where two of three buyers are on pre-set plans isn't a cluster — it's a coincidence.

2. Historical Pattern Detection

We track purchase history for every insider. If the same person buys at roughly the same time (±30 days) in 2+ consecutive years, future purchases at that time are flagged as routine and penalized in scoring.

This is computationally expensive — it requires maintaining a rolling history per insider across all filings — but it's the difference between showing you everything and showing you what matters.

The Takeaway

The edge in insider trading data isn't access. Everyone has access to SEC Form 4 filings. The edge is knowing which filings carry information and which are noise.

If your insider-trading tool doesn't distinguish routine from opportunistic, you're not reading signal. You're reading a calendar.

---

References:

• Cohen, L., Malloy, C. & Pomorski, L. (2012). "Decoding Inside Information." Journal of Finance, 67(3), 1009–1043.
• Alldredge, D. M. & Blank, B. (2019). "Do Insiders Cluster Trades With Colleagues?" Journal of Financial Research, 42(2), 331–360.
• Lakonishok, J. & Lee, I. (2001). "Are Insider Trades Informative?" Review of Financial Studies, 14(1), 79–111.