Cluster insider buying explained

For this public guide, cluster insider buying means two or more distinct Section 16 reporting persons at the same issuer reporting code-P purchases with transaction dates inside a 14-calendar-day window.

The definition has four deliberate boundaries:

• Distinct people. Several rows from one reporting person are not a multi-actor cluster.
• The same issuer. Purchases across unrelated companies do not belong together.
• Code P. Grants, exercises, withholding, and gifts do not qualify as purchases for this screen.
• Transaction dates. Filing dates can bunch activity that occurred on different days.

Fourteen days is an editorial screening window, not a law of markets. An event on day 15 does not lose all relevance, and two purchases on adjacent days do not become important by calendar magic.

The SEC’s Form 4 instructions↗ (opens in a new tab) define reporting fields, deadlines, ownership concepts, and transaction codes. They do not stamp a set of Forms 4 as a “cluster” or assign the group a bullish meaning. Cluster is a research label used to organize several public events into one question.

That distinction prevents category drift. The underlying facts remain individual filings: reporting owners, transaction dates, codes, amounts, prices, holdings, ownership forms, and footnotes. The cluster is an interpretation placed on top of those facts.

A meaningful cluster should plausibly represent more than one decision. Multiple insiders may receive the same board-approved grant on the same day, have shares withheld under the same compensation process, or appear together in a joint/group filing. Those events can be related by administration rather than independent judgment.

Same-day code-A grants are not automatically a meaningful cluster. Neither are several code-M exercises triggered by the same vesting calendar. The existence of several names tells you how many people are listed; transaction code and footnotes tell you what happened.

Look for signs of independence:

• separate code-P purchases rather than one joint transfer;
• different transaction dates or execution prices consistent with separate orders;
• no common compensation or administrative explanation in the footnotes;
• distinct reporting persons rather than duplicate ownership vehicles for one person.

A cluster of three code-P purchases is a different object from three acquisitions labeled A, M, or G. The acquired/disposed column alone cannot distinguish them. Code P says the transaction was an open-market or private purchase; the other codes describe grants, exercises, gifts, and related mechanics.

Private purchases can also deserve more context than ordinary exchange executions. Price, counterparty information when disclosed, indirect ownership, and footnotes may affect the economic read. “P” clears the first mechanical test. It does not end due diligence.

If one apparent member of the cluster is later corrected in a Form 4/A, recalculate the group. A pattern is only as stable as its source rows.

Role diversity

Purchases spanning operating officers, financial leadership, and directors can reflect a broader set of vantage points than several purchases by people with the same role. That may raise notability. It does not prove coordination, shared information, or correctness.

Transaction size

Add dollar amounts only after verifying price and shares, then keep relative context. A cluster of small policy-sized purchases can have a large aggregate headline while barely changing any participant’s exposure. Compare post-transaction holdings and prior behavior.

Timing proximity

A two-day grouping is generally tighter than a fourteen-day grouping, all else equal. Filing delays can distort the picture, so use transaction dates. Also check what company or market event sits inside the window; several purchases after a broad selloff may share context without sharing a conclusion.

If similar groups appear after every annual grant cycle or ownership-compliance deadline, repetition may weaken the idea that the latest cluster is unusual. If code-P purchases by different people appear irregularly and cluster around comparable conditions, repetition may strengthen the case for attention.

Historical comparison needs honest denominators. Two prior clusters are two cases. A neat story drawn through both does not become a track record. Record what happened, how the cluster was defined at the time, and what would have counted as a contrary outcome.

One insider is a data point. Several may be a pattern. Neither is a promise.

Before rewarding the cluster, check for:

• code-S sales by the same buyers before or after the purchases;
• larger sales by other reporting persons in the same window;
• exercise-and-sell sequences that leave little net exposure;
• code-F withholding misread as discretionary selling;
• indirect transfers that duplicate rather than expand ownership;
• amendments that change an amount, code, or date;
• a disclosed ownership policy that explains similar purchase sizes.

Contradictions do not automatically negate the cluster. They describe the evidence stack. A mixed stack deserves a mixed conclusion.

Even a high-quality cluster proves only that several qualifying purchases occurred close together. It does not prove undervaluation, shared conviction, or future performance. For the broader failure modes, use the skeptic’s checklist.

Buteon groups nearby events by issuer and tests distinct actors, codes, transaction dates, roles, size, ownership change, prior behavior, and offsets. A cluster can raise notability because several evidence points align. Conviction still depends on the wider stack.

Exact weights and scoring thresholds remain proprietary. The public analytical principle does not: independent, high-quality, economically meaningful purchases close in time are more interesting than administrative coincidence. Interesting is where research begins.