Insider Purchases and Earnings: Timing Matters

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.

A director who buys two weeks before earnings is either brave, lucky, or in possession of a calendar.

The thesis is simple enough to fit on a dealing-room napkin: insider purchases made shortly before an earnings announcement may carry more information than purchases made far from one. The reason is also simple enough to make compliance officers reach for aspirin. As the next reporting date approaches, executives and directors are closer to fresh operating information, but they are also more likely to be constrained by closed periods, internal dealing codes, and market-abuse rules. If a premium exists for buys made 14 days before earnings, it sits precisely where information content and legal friction collide.

This article examines that idea as a quantitative-finance problem, not a morality play. The target metric is T+90 excess return conditional on distance to the next earnings event. In plain English, we want to know whether stocks bought by insiders 14 calendar days before earnings go on to outperform over the next 90 trading days, after adjusting for a sensible benchmark. Since no article-specific query output has been provided from the 162k-filings database, the honest values for sample counts and realised returns are n/a. That is mildly unsatisfying, but still preferable to the industry standard of making numbers up with confidence.

Why earnings proximity might matter

The informational case

Insider trading research has long found that corporate insiders, especially when buying rather than selling, can be informative about future returns. The broad result is not controversial. Seyhun’s work documented abnormal returns following insider trades in the United States over long samples, and later studies refined the conditions under which those returns are stronger or weaker. The more interesting question now is conditionality. Insider buys are not all equal. A purchase by a founder after a 60 percent drawdown is not the same animal as a routine director top-up after vesting-related sales. Timing relative to earnings is one of the cleaner conditioning variables because it ties the trade to a recurring information event with a known date.

The intuition runs as follows. As a firm moves toward its earnings release, uncertainty about the quarter collapses internally before it collapses externally. Revenue is largely booked, costs are more visible, guidance debates are underway, and any operational surprise has probably already ruined someone’s weekend. If an insider chooses to buy during that interval, the trade may encode a stronger private assessment about the near-term narrative and the market’s likely reaction.

That is the bullish version. It is also incomplete.

The compliance case

Most jurisdictions do not regard “closer to earnings” as a charming source of alpha. They regard it as a reason to restrict dealing. Under the EU Market Abuse Regulation, persons discharging managerial responsibilities, PDMRs, face a 30-calendar-day closed period before the announcement of an interim financial report or year-end report, subject to narrow exceptions. Issuers often impose broader internal windows. In the United States, there is no MAR-style statutory closed period for all issuers, but companies commonly maintain blackout windows around quarter-end and earnings, and Rule 10b5-1 plans are meant to reduce the appearance and reality of opportunistic timing.

So if one tests “buys 14 days before earnings”, one is not merely testing information content. One is also testing the residual set of trades that still occur despite those restrictions. Depending on market, issuer policy, and insider role, that residual set may be unusually informative, unusually noisy, or simply unusually sparse.

The practical implication is awkward but important. A pre-earnings premium observed in US data may not travel neatly to France or the wider EU, because the legal opportunity set is different. A strategy that looks like “buy the insider buys 14 days before earnings” may partly be a strategy on governance regimes and compliance architecture.

How to test the earnings proximity premium properly

Define the event-time variable first

The cleanest setup is to compute, for every insider purchase filing, the number of calendar days from the trade date to the issuer’s next scheduled earnings announcement. Call this days_to_earnings. Then bucket trades into intervals such as:

• 0 to 7 days
• 8 to 14 days
• 15 to 30 days
• 31 to 60 days
• 61+ days

The article’s focal bucket is 14 days before earnings. One can define this narrowly as exactly 14 calendar days, or more usefully as a band, say 8 to 14 days, to avoid pretending that event-time is measured with atomic precision in a world of after-hours filings and timezone oddities.

The next step is to calculate T+90 excess return from the trade date. “Excess” needs a benchmark. There are several defensible options:

1. Market-adjusted return, subtracting the local broad-market index return over the same horizon.
2. Sector-adjusted return, subtracting a matched sector index.
3. Characteristic-adjusted return, subtracting the return of a size-book-to-market or size-momentum matched portfolio.
4. Factor alpha, estimating abnormal return relative to a factor model.

For a publication article, market-adjusted or sector-adjusted excess return is usually the right balance between rigour and readability. Factor models are useful, but they can turn a straightforward question into a hostage situation involving stale betas.

Filter the trades that should not be there

Insider datasets are full of transactions that look like signal and behave like plumbing. A proper test should exclude, or at least separately tag:

• option exercises without open-market purchase intent,
• automatic dividend reinvestment,
• transfers between beneficial ownership entities,
• tax-related dispositions,
• trades under pre-arranged plans where identifiable,
• tiny notional purchases that barely cover lunch.

The focus should be discretionary open-market buys. If the database does not fully classify transaction type, the article should say so plainly. The worst backtests in finance are usually built on immaculate code and dirty labels.

Choose the unit of observation carefully

A single filing can contain multiple line items, and multiple insiders can buy the same stock on the same day. That creates a choice. One can analyse at the transaction level, the insider-day level, or the issuer-day level. Each has trade-offs.

• Transaction level preserves granularity but can overweight clustered filings.
• Insider-day level reduces duplicate noise.
• Issuer-day level is often best for return prediction because the traded asset is the stock, not the person.

A sensible compromise is to aggregate same-day buys within issuer, then compute a value-weighted or equal-weighted signal. If several insiders buy within the same issuer and same event-time bucket, that cluster itself may be informative and worth a separate indicator.

Make earnings dates boringly accurate

This sounds trivial until it is not. Earnings announcements can be before market open, after market close, delayed, revised, or announced through local channels not captured cleanly by generic vendors. Event-time misclassification is fatal here because the entire thesis hinges on distance to earnings. If the earnings calendar is wrong by even one day, the “14 days before” bucket becomes a random-number generator with a tie.

For France and the EU, issuer investor-relations pages and regulated-information feeds are often more reliable than generic calendars. For US names, SEC 8-Ks and issuer releases can help validate dates. If the earnings timestamp is unavailable, use the announcement date and accept that some trades near the boundary will be noisy.

What the premium could mean, if it exists

Information, yes, but also selection

Suppose the 8 to 14 day bucket shows stronger T+90 excess returns than the 31 to 60 day bucket. The lazy interpretation is “insiders know more near earnings”. The better interpretation is “the subset of insiders who are both allowed and willing to buy near earnings differs from the broader insider population”.

That distinction matters. Near earnings, the denominator shrinks. Many insiders are prevented from trading by policy or prudence. Those who do trade may be concentrated in smaller firms with looser windows, in jurisdictions with different constraints, or among non-executive directors who have a different information set from the CFO but still enough conviction to act. The measured premium may therefore reflect a selection effect rather than a pure increase in information precision.

A useful falsification test is to compare the pre-earnings premium across governance regimes. If the effect is strongest where blackout policies are weakest, that supports an information-opportunity story. If it survives in stricter regimes but only for certain insider roles or exception cases, that points to a more nuanced mechanism.

The market may underreact differently near earnings

There is also a microstructure angle. A buy close to earnings may be more likely to be ignored or discounted by outside investors because they assume, reasonably, that insiders should not be trading aggressively at that time. The market might treat the filing as noise, a compliance artefact, or a small personal allocation choice. If so, any genuine information in the trade is incorporated only gradually, producing a stronger T+90 drift.

This is not guaranteed. The opposite can happen. A pre-earnings purchase can attract disproportionate attention precisely because it is unusual, causing a faster price response and leaving less for the next 90 days. Which effect dominates is an empirical question.

Why T+90 is a sensible horizon

A 90-trading-day horizon is long enough to capture post-earnings drift, guidance revisions, and second-order analyst updates, but short enough to remain plausibly tied to the original information event. T+5 is too short and mostly measures filing attention. T+252 starts to blend into the general mess of business quality and market regime.

There is one caveat. If the trade occurs 14 days before earnings, then much of the T+90 window lies after the earnings announcement itself. That is not a bug. It is the point. The signal is whether the insider’s pre-event purchase predicts the market’s medium-term reassessment after fresh public information arrives.

The obvious traps, and a few less obvious ones

Closed periods can invert the sample

In Europe under MAR, a strict 30-day closed period before interim and year-end reports means that a “14 days before earnings” sample may be tiny or populated by exceptions. That creates two problems. First, small samples are unstable. Second, exceptions are not random. They may involve unusual circumstances, such as severe financial difficulty, employee schemes, or other carve-outs. If one does not separate these, the estimated premium can be nonsense with decimal places.

In the US, the issue is less statutory and more policy-driven. Blackout windows vary by issuer. Some companies allow trades for a short period after quarter-end but before earnings under certain conditions, though many do not. Again, heterogeneity in permission sets contaminates simple pooled tests.

Filing date versus trade date

Many databases are built around filing dates because those are easy to collect. For this signal, that is not good enough. The event-time variable must be anchored to trade date, while any investable implementation must consider filing date because that is when the market learns about the trade. These are different questions:

• Trade-date analysis asks whether insiders possess useful information.
• Filing-date analysis asks whether outside investors can capture it.

A proper article should ideally show both. The trade-date version is academically cleaner. The filing-date version is what readers can actually trade, subject to slippage and liquidity.

Earnings date endogeneity

Insiders may know that earnings will be brought forward, delayed, or accompanied by guidance changes. If the public calendar is stale, the measured days_to_earnings can be wrong in a way that is correlated with the insider’s information. That is a subtle source of bias. It will not ruin every test, but it should make one modest.

Small-cap concentration

Insider signal strength is often stronger in small and less-followed firms. Those same firms may have more variable reporting practices and less rigid internal trading controls. If the 14-day premium is really just a small-cap premium wearing a lanyard, sector and size controls should expose that. At minimum, stratify results by market capitalisation.

A research design worth publishing

Baseline specification

If the data were fetched, the baseline table I would want is straightforward:

1. Sample of discretionary open-market buys.
2. Sort by days_to_earnings buckets.
3. Compute average and median T+90 excess return.
4. Report counts, hit rates, and value-weighted notional.
5. Control for size, sector, country, insider role, and filing lag.

A regression form could look like:

[ ExcessReturn_{i,t+90} = \alpha + \beta_1 NearEarnings_{i} + \beta_2 Size + \beta_3 Momentum + \beta_4 Sector + \beta_5 Country + \beta_6 InsiderRole + \epsilon ]

where NearEarnings is an indicator for the 8 to 14 day bucket, or a spline over event-time distance. One can also test non-linearity. It would not be surprising if the relationship were hump-shaped rather than monotonic. Extremely close trades, say 0 to 3 days, may be rare and distorted by restrictions. Moderately close trades, say 8 to 14 days, may be the sweet spot. Trades far from earnings may simply be less tied to a concrete catalyst.

Interaction terms that actually matter

Three interactions are especially worth testing.

Insider role × earnings proximity

CEOs and CFOs are more informed, but also more restricted and more scrutinised. Non-executive directors may trade with less direct operational knowledge but perhaps more freedom in some firms. If the premium is concentrated among one role, that tells you something useful about mechanism.

Buy size × earnings proximity

Large discretionary purchases, especially relative to the insider’s prior holdings or annual cash compensation, tend to be more informative than token buys. A 14-day premium that appears only in large buys is more credible than one driven by symbolic purchases.

Prior stock performance × earnings proximity

Contrarian insider buys after sharp drawdowns often predict rebounds. If the 14-day effect survives after controlling for prior six-month return, it is less likely to be a simple “insiders buy weakness” story.

Event-study sanity checks

Before getting lost in regressions, run a plain event study. Plot cumulative abnormal returns from filing date to T+90 for each event-time bucket. If the curves are indistinguishable until day 40 and then one suddenly levitates, your benchmark or sample construction probably needs a stern conversation.

A second useful plot is trade frequency by days_to_earnings. In a strict MAR environment, activity should collapse in the final 30 days for covered insiders. If it does not, either the sample includes many non-covered persons, the transaction coding is broad, or the earnings calendar is wrong. None of those are fatal, but each changes the interpretation.

What we can say now, before the numbers arrive

The prior from the literature

The literature gives a cautious prior in favour of conditional insider-buy alpha, but not a free pass for this exact variant. Broadly, insider purchases outperform on average, especially in smaller firms and when trades are clustered or opportunistic. There is also a large literature on post-earnings announcement drift, showing that the market does not fully incorporate earnings news immediately. Those two facts make the “pre-earnings insider buy” hypothesis plausible. They do not prove that 14 days is the magic number.

The closest intellectual neighbour is not simply insider trading research, but the literature on attention, disclosure timing, and underreaction. A pre-earnings buy could be informative because the insider anticipates a positive earnings surprise, because they expect guidance to improve, or because they know the market’s current expectations are stale. T+90 is where those channels tend to reveal themselves.

What would count as convincing evidence

Given the legal and selection issues, I would regard the following as convincing:

• a positive and statistically reliable T+90 excess return for the 8 to 14 day bucket,
• persistence after controlling for size, sector, country, insider role, and prior momentum,
• stronger effects for larger discretionary buys,
• similar direction on filing-date implementation, even if attenuated,
• no obvious concentration in a handful of illiquid microcaps,
• and a sensible cross-market pattern consistent with local dealing rules.

What would not convince me is a single pooled average with no controls, no role split, and no discussion of blackout windows. That is not a result. It is a request to be fooled.

What might falsify the thesis

Three outcomes would seriously weaken the idea.

First, if the premium disappears once trades under plans, option-related transactions, and tiny notionals are removed. Second, if the effect is entirely explained by small-cap exposure or by buying after drawdowns. Third, if filing-date returns are flat, implying that any information advantage exists only on trade date and is not monetisable by outside investors.

A fourth possibility is more interesting than fatal: the premium may be positive in the US but absent or reversed in France and the EU because MAR-style restrictions remove the cleanest opportunities. That would not kill the general concept. It would simply mean the signal is regime-specific, which is often how markets remind us that law is part of data-generating process.

Implementation notes for an investable version

Trade date is research, filing date is portfolio construction

If this article were being turned into a live strategy, the sequence would be:

1. Detect a qualifying insider buy filing.
2. Map the trade date to the next earnings date.
3. Keep only buys in the chosen event-time bucket.
4. Enter on filing date close or next open.
5. Hold for 90 trading days, or test alternatives around the next earnings cycle.

That implementation will almost certainly underperform the trade-date backtest. It should. Reality is rude like that.

Capacity and liquidity

Pre-earnings insider buys may cluster in smaller names, where spreads are wider and position sizing is constrained. If the signal only works in names where one cannot deploy serious capital without becoming the market, that is not useless, but it is not the same thing as scalable alpha. Capacity analysis should therefore report median market cap, median daily value traded, and implementation shortfall assumptions. Since no article-specific data extraction has been supplied, those values remain n/a here.

A practical extension

A promising extension is a two-stage filter:

• Stage 1: insider buy within 8 to 14 days before earnings.
• Stage 2: positive earnings surprise or constructive guidance at release.

This turns the signal into a confirmation framework. One can test whether the combination produces stronger post-event drift than either component alone. It is less elegant than the pure insider signal, but markets have never promised elegance.

The article we can write once the query runs

At present, the article-specific dataset was not fetched, so the empirical heart of the piece remains pending. That means no counts by bucket, no realised T+90 excess returns, no role splits, no France-versus-US comparison from the 162k filings archive, and no backtest chart worth printing. The correct values are n/a. There is no shame in that. There is some shame in pretending otherwise.

What we do have is a clear hypothesis, a defensible test design, and a list of ways the result could be real, spurious, or regime-dependent. That is enough to know what to ask of the database next:

• How many discretionary open-market buys occur 8 to 14 days before earnings, by market and role?
• What is their T+90 excess return on both trade-date and filing-date bases?
• Does the effect survive controls for size, sector, and prior drawdown?
• Is the premium stronger in clustered buys?
• Does it vanish in MAR-heavy samples where closed periods should suppress activity?

Those are not clerical questions. They are the difference between a publishable anomaly and a compliance artefact with a nice chart.

The payoff is straightforward. If the 14-day bucket survives these tests, you have a concrete event-conditioned insider signal worth monitoring in production. If it does not, the next step is equally concrete: widen the lens to a full event-time curve around earnings and let the data tell you whether the premium lives at 21 days, after the release, or nowhere at all. That open question is better than a fake answer.

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 insider trading activity before earnings announcements reveals significant insights. With a dataset comprising 220,691 priced trades across 17 markets, the mean 90-day return stands at -1.70%, accompanied by a win rate of 41.82%. The annualized Sharpe ratio is -0.08, indicating that the risk-adjusted returns for this cohort are underwhelming. This aligns with the broader global universe context, where 227,594 BUY filings yield an identical mean 90-day return of -1.70% and a Sharpe ratio of -0.08. Such figures suggest that relying on insider trading signals prior to earnings announcements may not provide the anticipated advantage, challenging the notion that insider information consistently leads to profitable trading outcomes. It is important to note that these findings are derived from a substantial sample size and a 90-day time horizon, which may influence the overall interpretation of the data.

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