How to read an insider transaction

A single insider filing is a sentence in a language with only five words: who, what, how much, when, and under which rules. Most reading mistakes come from skipping one of the five.

The five fields that matter in every filing

Every disclosure regime we ingest, from SEC Form 4 to MAR Article 19 notifications, reduces to the same skeleton:

1. The person: name and role (CEO, CFO, director, or a person closely associated with them).
2. The nature: purchase, sale, award, exercise, pledge, gift.
3. The volume and price: how many shares, at what unit price.
4. Two dates: when the trade happened, and when it was published.
5. The instrument: ordinary shares, derivatives, or something more creative buried in a footnote.

The rest of this guide walks through each field in the order a practitioner should read them, with links to the deeper studies this series has published on each one.

Nature: the only field with a built-in prior

Start with what kind of transaction it is. The academic record, from Seyhun onward, is consistent on one asymmetry: insider purchases carry information on average, insider sales mostly do not. An executive can sell for a divorce, a house, a tax bill, or simple diversification. There is exactly one reason to buy with personal cash.

On US Form 4 filings the nature is encoded in a transaction code: P is an open-market or private purchase, S a sale, A a grant or award, M an option exercise, F shares withheld for taxes, G a gift. Only code P is a discretionary buy; a Form 4 that looks bullish can be an automatic vesting event once you check the code. European MAR notifications describe the nature in free text, which is messier but carries the same distinction.

Two refinements matter:

• Option exercises are not purchases. An M followed by an S is an insider converting compensation into cash, not a conviction trade.
• Pre-scheduled plans dilute the signal. In the US, trades executed under a 10b5-1 plan were committed to months in advance, so they say little about what the insider believes today. The filing usually flags this; the footnotes are where such details hide.

Who: role is a feature, not a ranking

A filing signed by a CFO is not automatically worth more than one signed by a director, but role does change the prior. Finance chiefs sit closest to earnings quality and covenant reality; chief executives have the broadest strategic view and the loudest signalling effect. We compared the two head-to-head over a 90-day horizon in CEO vs CFO buys; the honest summary is that role matters conditionally, entangled with firm size, trade size, and timing.

Also watch for the widest role bucket: persons closely associated (family members, holdings, foundations). Under MAR they file too, and a purchase routed through a spouse or a family holding is still a purchase decided at the same dinner table.

Finally, look up the person, not just the title. A buyer with a strong personal track record across past filings deserves more attention than a first-time buyer. Every insider page on this site aggregates that history per person.

How much: three denominators, one useful number

A 100,000 EUR purchase is enormous in a 50M EUR micro cap and a rounding error at a mega cap. Raw amounts mislead; you need denominators:

• Percent of market cap. The same cash buys a very different share of the company across the cap spectrum. This is one reason insider signals behave differently in small caps: the same conviction moves the needle more where coverage is thin.
• Relative to the insider's own economics. A CEO adding 2 percent to an already huge stake is topping up. A CFO putting several months of salary into the open market is making a statement.
• Relative to the insider's past trades. A buyer who usually files 5,000 EUR purchases and suddenly files 500,000 EUR is telling you something changed.

When: two dates and a legal delay

Every filing has a transaction date and a publication date, and the gap between them is set by law: two business days for a US Form 4, three business days for a MAR notification in the EU, with other regimes documented in our per-market guides. Backtest or screen on the publication date; the market cannot trade information it has not seen. The gap itself is informative: consistently late filers are a red flag regulators watch, a pattern we covered in the 3-business-day rule and enforcement asymmetry.

Context around the date matters as much as the date:

• Closed periods. Under MAR, dealings by executives are prohibited during the 30 calendar days before interim or year-end results. A purchase landing just AFTER results, when the window reopens, is common and mechanically clustered; one landing well inside an open window is a freer choice.
• Earnings proximity. Buys shortly before earnings announcements are rarer and historically carry a different return profile; we measured this in the earnings proximity premium.
• Drawdown context. A buy after a 40 percent slide is a contrarian statement about intrinsic value. A buy at the all-time high is momentum agreement. Same nature, same size, different meaning.

The strongest pattern: several insiders, one stock

If one field combination deserves a shortcut, it is this: multiple distinct insiders buying the same stock within a short window. Independent wallets reaching the same conclusion compound in a way one large trade does not. The pattern goes back to the cluster-trade literature and it is the single most reliable screen we run; the full 17-market treatment is in our cluster trades study, and the live version is the cluster buys screener.

Symmetrically, one insider buying while two others sell is not a cluster. Net the flows before celebrating.

From reading to scoring

Everything above can be done by hand for one filing. At scale it becomes a scoring problem: weigh nature, role, relative size, timing, and clustering into one rank. That is what our composite score does, and its weights are public and audited, not folklore: see the composite score audit for why each component earns its weight, the methodology page for definitions, and the performance page for the current out-of-sample results with their test window. This pillar deliberately quotes no performance numbers: they live in one place, where they are kept current.

Go deeper

This article is the pillar of our "reading an insider transaction" series. The spokes each take one field further:

• CEO vs CFO buys: which role wins on T+90
• Cluster trades across 17 markets
• The earnings proximity premium
• When do insiders themselves exit?
• Form 4 footnotes: what insiders hide in legalese
• Small-cap insider edge
• The composite score audit
• Self-dealing red flags

To apply it live: the screener filters every field discussed here, and the performance page shows what the resulting strategy has done out of sample.