The Effects of Insider Filings on Stock Market Returns

Note on figures cited below. Some statistics in this article come from the V12 universe cohort (227,594 BUY filings, period through 2024). Since 2025-01-01 the live out-of-sample BUY universe (28 markets) shows a 52.96 percent win rate and a +8.59 percent mean 90-day return on 20,688 filings. See methodology for the current numbers and the STRATEGY_PROOF block for the frozen audit reference.

One filing can move a stock more in an afternoon than a quarter of polished investor-relations prose.

The filing hits the tape, then the market starts arguing

Insider disclosures are one of the few corporate signals that arrive with a blunt factual core. Someone close to the issuer bought, sold, exercised, transferred, or otherwise moved economic exposure. The filing is public. The timestamp is public. The market then does what markets do, it interprets, discounts, overreacts, underreacts, and occasionally shrugs.

This article focuses on the narrow but useful question: what does average abnormal return look like by day after publication, from T+0 to T+30?

That framing matters. It is not the same as asking whether insiders earn excess returns when they trade. It is not the same as asking whether insiders possess private information. Those are older and broader literatures. Here the event is the publication of the filing, because publication is when the rest of the market can react. If you care about tradeable price impact, that is the clock that counts.

The awkward part is that we do not have a fresh article-specific extract from Insiders Trades's 162k-filings database for this piece. So there will be no heroic chart claiming that France buys earned exactly x basis points on T+2. Where article-specific numbers are unavailable, I will say n/a. Better an honest blank than a decorative fiction.

Still, there is plenty to say, because the shape of post-publication returns is constrained by market microstructure, disclosure law, and a large academic literature on insider trading and post-disclosure drift.

Why publication day is the real event

Under the EU Market Abuse Regulation, persons discharging managerial responsibilities, and closely associated persons, must notify transactions once a threshold is met, and the issuer and competent authority must receive that notification promptly. Public disclosure follows on a short timetable. In practice, this means the market often receives a relatively crisp public signal, even if the underlying trade happened earlier.

That distinction is not a footnote. If an executive bought shares on Monday but the filing is published on Thursday, any event study anchored on the trade date risks mixing private-information trading with public-information reaction. An event study anchored on publication date captures something cleaner: the market's response to the disclosure itself.

Why daily microstructure is messy even when the rulebook is clean

A daily abnormal-return series from T+0 to T+30 sounds straightforward. It is not. The filing may appear before the opening auction, during continuous trading, or after the close. The stock may be liquid enough to absorb the signal immediately, or illiquid enough that the first day's print is mostly spread and inventory adjustment. The filing may coincide with earnings, guidance, a capital raise, or a board reshuffle. And what looks like a "sale" may be a tax-driven disposal after vesting, a derivative unwind, or a family-office transfer.

Markets are very good at turning a simple input into a statistical hygiene problem.

What the literature says about the shape of post-disclosure returns

The broad academic result is familiar: insider purchases are generally more informative than insider sales, and the strongest predictive content often appears in smaller firms, less-followed firms, and more opportunistic trades. That is the headline. The daily path is where things get interesting.

T+0 to T+2, the market's first pass

On publication day, T+0, the market reacts to the filing's sign, size, and context. Purchases by senior executives, especially open-market buys, are usually read as a positive signal. Sales are more ambiguous. An executive can sell for any number of boring reasons, including taxes, diversification, or school fees. Markets know this, which is why sale filings often generate weaker or less consistent abnormal returns than buy filings.

In microstructure terms, T+0 is where order imbalance, spread setting, and auction discovery do most of the work. If the filing lands before the open, much of the reaction may be incorporated in the opening auction. If it lands intraday, the path can be jagged, because market makers widen spreads, liquidity providers reassess inventory risk, and slower investors process the news in sequence rather than as a single unified intelligence.

The literature on post-event drift suggests that immediate incorporation is often incomplete. That does not mean markets are asleep. It means information diffusion is costly and uneven. Some investors ignore insider filings. Some parse them badly. Some cannot trade quickly in smaller names without moving the price against themselves. The result is that T+1 and T+2 often still matter.

T+3 to T+10, signal quality starts to separate from noise

By the first full week after publication, the initial "headline reaction" has faded and a more selective process takes over. This is where the market begins to discriminate between filings that were genuinely informative and filings that merely looked dramatic in PDF form.

Open-market purchases by a chief executive after a drawdown tend to carry more informational weight than an automatic sale under a pre-arranged plan, or the European equivalent of a routine disposal linked to vesting and tax settlement. Clustered buying by multiple insiders can strengthen the signal. Tiny trades can weaken it. So can transactions that leave economic exposure largely unchanged.

For average abnormal returns, this middle stretch often determines whether the event study shows a one-day pop or something closer to a short drift. If abnormal returns remain positive through T+5 or T+10 for purchase filings, that is usually interpreted as evidence of slow incorporation or limited attention. If the effect reverses quickly, the market may have initially overreacted, or the filing may have been confounded by other news.

T+11 to T+30, persistence, reversal, or statistical dust

By the time you reach T+30, the daily microstructure story has become a medium-horizon information story. If cumulative abnormal returns are still building, the filing likely contained information that the market digested only gradually. If they flatten, the signal was absorbed early. If they reverse, the initial interpretation was too generous, or the event was contaminated by transient liquidity effects.

This is where many practical strategies go to die. A signal that looks excellent on a cumulative chart can be untradeable after spreads, market impact, and the simple fact that many filing-driven moves happen in names where liquidity is thin and patience is expensive.

That is also why average abnormal return by day is more useful than a single cumulative number. The daily series shows when the market moved, which is often more valuable than knowing only how much it moved by T+30.

The mechanics behind the curve, auctions, spreads, and delayed digestion

If you want to understand the daily path from T+0 to T+30, start with the plumbing rather than the poetry.

Opening auctions do more work than many event studies admit

A filing disclosed before the market opens can be reflected almost instantly in the opening auction price. In that case, much of the "day 0" abnormal return is really an auction effect, not a continuous-trading effect. This matters because auction prices aggregate overnight information under a different mechanism than intraday trading. The opening print can be efficient in one sense, but still leave room for drift if not all investors participate or if the stock opens with limited depth.

For researchers, this creates a practical issue. A daily close-to-close event study may understate the immediacy of the reaction if the key adjustment happened at the open, and overstate intraday digestibility if it attributes the full move to a single trading day.

Spreads and liquidity can manufacture apparent alpha

Small and mid-cap stocks often show stronger insider-related anomalies. That sounds exciting until you remember that these are also the names where bid-ask spreads are wider, depth is thinner, and stale pricing is more common. Some of the measured abnormal return around filings may reflect compensation for illiquidity or simply the mechanics of crossing the spread.

This is not a reason to ignore the signal. It is a reason to ask whether the signal survives realistic execution assumptions. A buy filing followed by a 1.5 percent abnormal move over three days is less impressive if the spread is 80 basis points and your own order nudges the book.

Publication timing creates hidden T+1 effects

Many disclosure systems timestamp publication at the document level, but not always with the granularity a microstructure analyst would like. If a filing appears after the close, the economically relevant reaction starts on the next session. If your event study labels the calendar date of publication as T+0 regardless of market hours, you can dilute the true day-0 effect and artificially inflate T+1.

This is one reason daily event studies often show a suspiciously strong next-day continuation. Sometimes it is slow information diffusion. Sometimes it is just a filing posted at 18:42.

The market reads the form, then reads the footnotes

Not all insider filings are equal on first inspection. The headline fields, insider name, role, transaction type, volume, and price, are machine-readable enough. The crucial context often sits in the notes. Was the trade discretionary? Was it linked to an option exercise? Did the insider increase net exposure or simply reshuffle claims? Was the transaction part of a broader compensation event?

Human and machine readers both process headline fields faster than footnotes. That asymmetry can produce a two-stage reaction: an initial sign-based move on T+0, then a refinement over T+1 to T+5 as context is digested.

Europe gives you a cleaner timestamp, but not a perfect experiment

The European disclosure regime is often a better starting point for publication-date studies than older US insider datasets, because the public disclosure process is more standardised across venues. "Better" is not the same as "perfect".

MAR helps because the public event is observable

Article 19 of MAR gives researchers a practical anchor. Once the disclosure is public, the market can react. That sounds obvious, but in empirical finance, obvious is a luxury. The clearer the public timestamp, the less likely you are to confuse private-information trading with public-information processing.

This is especially useful when comparing average abnormal returns by day. If your event date is clean, then a T+0 spike followed by T+1 drift can be interpreted with more confidence as public-information assimilation rather than leakage.

But issuer practice still varies

The filing may be posted on the issuer website, disseminated through a newswire, captured by a vendor, and indexed by a regulator-facing system on slightly different clocks. If your dataset relies on one source of timestamps and your market data relies on another, the event window can be misaligned by hours or, in ugly cases, by a full session.

That is not a theoretical nuisance. It directly affects the estimated shape of the T+0 to T+30 curve.

Sales remain the awkward cousin

The literature has long found that insider purchases are more informative than sales. Under MAR or SEC rules, that asymmetry does not disappear. If anything, richer disclosure can make the asymmetry more visible, because many sales are revealed to be routine or mechanically linked to compensation events.

For a publication-date event study, this implies that average abnormal returns for sale filings may be weak, noisy, or close to zero after controlling for context. A simple pooled "all filings" chart can therefore hide the only effect investors actually care about.

How to build the daily event study without embarrassing yourself

A good publication-date study is less about fancy statistics than about refusing easy mistakes.

Step one, classify the event properly

At minimum, separate open-market purchases from open-market sales. Then identify exercises, awards, gifts, transfers, and tax-related disposals. If the filing taxonomy is messy, use the notes field and net-exposure logic where possible. A disposal that accompanies an exercise but leaves the insider with materially more shares is not economically the same as a clean bearish sale.

Role matters too. Chief executives and finance directors often carry more signal than non-executive directors, though that is an empirical question, not a moral hierarchy.

Step two, anchor on publication timestamp, not trade date

For this article's angle, publication is the event. If the filing appears after the market close, shift the effective event to the next trading session or run a specification that distinguishes pre-open, intraday, and after-close releases. Otherwise your T+0 and T+1 estimates will be a polite fiction.

Step three, choose a benchmark and then stop worshipping it

A market model, sector-adjusted return, matched-firm return, or factor model can all work. The benchmark matters, but less than many papers imply. In short windows, contamination and timing dominate benchmark choice. In longer windows out to T+30, benchmark specification becomes more important, especially in volatile markets.

The sensible approach is robustness. Report the daily average abnormal return under more than one benchmark. If the sign and shape survive, the result is probably real enough to discuss.

Step four, cluster your standard errors and de-duplicate your events

Insider filings cluster. Multiple insiders may file on the same issuer and same day. One executive may file several lines tied to a single economic action. Treating each line item as an independent event is a reliable way to manufacture significance.

Aggregate to issuer-day where appropriate, then test sensitivity to alternative grouping rules. It is less glamorous than a heatmap, but more useful.

Step five, inspect the tails

Average daily abnormal return can be distorted by a handful of extreme events, especially in small caps. Winsorisation, medians, and distributional plots are not optional niceties. They are the difference between a signal and a hostage situation.

What a plausible T+0 to T+30 pattern usually looks like

Without a live article-specific extract, we cannot print Insiders Trades's own daily return curve here. We can still describe the pattern that a careful researcher would expect to test for.

For purchase filings

The standard expectation is a positive abnormal return on T+0, some continuation through T+1 to T+5, then either a tapering drift or a flattening cumulative profile out to T+30. The strongest version of this pattern should appear in discretionary open-market buys, in smaller or less-followed firms, and when multiple insiders buy within a short interval.

A weaker but still meaningful pattern is a muted T+0 with stronger T+1, often caused by after-hours publication or slower digestion in less liquid names.

For sale filings

The expectation is weaker and more conditional. Routine sales may show little to no abnormal return. Opportunistic or unusually large discretionary sales may produce negative T+0 or short-horizon drift, but the average effect is often diluted by benign motives. If a study finds sale filings to be as informative as purchases on average, the first question should be whether compensation-related disposals were properly filtered.

For mixed or pooled samples

The curve often looks disappointingly flat. That is not because insider disclosures do not matter. It is because the sample design has averaged a strong positive subgroup with a weak or noisy subgroup and then declared the corpse representative.

What matters to investors, and what remains unresolved

The practical value of a daily post-publication study is not academic ornament. It tells you whether the signal is immediate, delayed, or mostly eaten by the opening print.

If the effect is mostly T+0, speed matters more than conviction

A strategy that depends on same-day reaction is a market-access problem. You need reliable ingestion of filings, timestamp discipline, and enough liquidity to trade without donating the edge to the spread. This is where many elegant backtests become charitable contributions to brokers.

If the effect drifts through T+5 or T+10, selectivity matters more than speed

A slower pattern is more interesting for discretionary investors and lower-turnover systematic strategies. It suggests the market underreacts to some filing types, especially those requiring context to interpret. Here the edge lies in classification quality, not raw latency.

T+30 is where economics should beat cosmetics

If cumulative abnormal returns persist to T+30 after realistic controls, that is more compelling than a noisy T+0 pop. But it also invites harder questions. Is the filing revealing undervaluation, signalling confidence, or simply coinciding with other good news? A publication-date event study can show association and timing. It cannot, by itself, settle motive.

That leaves a useful open research agenda. Which combinations of insider role, transaction type, firm size, and disclosure timing produce the cleanest post-publication drift? And how much of that drift survives once you account for spreads, auction mechanics, and the dreary but important fact that other people can read filings too?

The next concrete step is simple enough to state and annoyingly hard to execute: build the T+0 to T+30 average abnormal return curve separately for discretionary open-market buys and routine compensation-linked sales, split by pre-open versus after-close publication, and test whether the difference survives realistic execution costs. If it does, you have a market signal. If it does not, you have a literature review with better branding.

Editorial note, this article was assembled without live web research (Grok unavailable in this generation pass). Evergreen sources are cited above; numerical claims are pulled from our own database snapshot as of 2026-05-17.

Insiders Trades backtest cohort

Our analysis of the UK market reveals a notable absence of priced trades within the BUY cohort, resulting in a mean 90-day return and win rate that remain undefined. In contrast, when examining the broader global universe, which encompasses 227,594 BUY filings across 17 markets, we observe an aggregate mean 90-day return of -1.70% and an aggregate Sharpe ratio of -0.08. These figures suggest that, on average, insider disclosures may not lead to positive stock price movements, reinforcing the notion that market reactions to such disclosures can be muted or even negative. The lack of data from the UK highlights potential limitations in the availability of actionable insights within this specific context. It is essential to consider the sample size and time horizon when interpreting these results, as they may not fully capture the complexities of insider trading impacts across different markets.

Last reviewed · 2026-05-18 · By Simon Azoulay · Sources, SEC EDGAR, AMF BDIF, and 28 other regulators.