A $10M Insider Sale Is Not Always Large: Measuring Sales Against Holdings

A $10 million insider sale sounds more consequential than a $250,000 sale. Economically, that comparison is incomplete.

The larger transaction may represent a small trim by a founder with a concentrated position. The smaller transaction may liquidate nearly everything a reporting owner held directly in the security. Form 4 provides the fields needed to distinguish those cases: shares disposed, shares beneficially owned after the transaction and direct or indirect ownership.

Using the complete July 29, 2026 InsiderAlpha Form 4 Research Database, we measured 66,043 mature direct-ownership sale events from 2007 through January 2026.

The result was weaker than the intuitive rule:

Larger holdings-relative sales were followed by slightly lower raw outcomes in some comparisons, but the relationship was small, non-monotonic and not stable enough to treat as a standalone bearish score.

The key result

The median 180-day security return declined from +4.91% after sales below 1% of estimated pre-sale holdings to +3.98% after sales between 20% and 50%. The final 50% to 100% cohort then rose to +4.42%.

Sale as share of estimated pre-sale holdings	Events	Issuers	Median disclosed value	Median 30D	Median 90D	Median 180D	Positive at 180D
Under 1%	6,871	420	$442,906	+0.94%	+2.50%	+4.91%	60.2%
1% to under 5%	12,761	548	$235,264	+0.84%	+2.55%	+4.65%	60.5%
5% to under 20%	22,893	589	$524,440	+0.59%	+2.30%	+4.47%	59.9%
20% to under 50%	15,573	583	$1,168,741	+0.80%	+2.42%	+3.98%	59.6%
50% to 100%	7,945	500	$1,443,695	+0.65%	+2.32%	+4.42%	60.3%

The sequence is not a return staircase. Across all events, the Spearman rank correlation between sale fraction and the 180-day outcome was -0.010. Relative size barely ranked subsequent absolute returns.

What the percentage measures

For each owner event, we estimated the directly reported position before the sale:

estimated pre-sale holdings = shares sold + shares beneficially owned after

The primary feature is:

sale fraction = shares sold / estimated pre-sale holdings

If an owner sold 10,000 shares and reported 190,000 held afterward, the estimated sale fraction is 5%. If the same 10,000 shares left no directly owned shares, the observed fraction is 100%.

This is not the percentage of the person's net worth, total economic exposure or fully diluted ownership of the issuer. It describes one reported security and ownership form. The official Form 4 instructions require transaction amounts, post-transaction beneficial ownership and direct or indirect ownership to be reported separately.

Earlier research used holdings-relative sales to distinguish potentially informed selling from liquidity-motivated trades. The 2004 study Some Insider Sales Are Positive Signals examined an earlier 1987–2002 sample and a different portfolio design. The current test is a more recent transaction-level replication, not a like-for-like reproduction of that paper.

The buy-side companion, A $1M Insider Buy Is Not Always Large, applies the same economic question to reported purchases.

Sample construction

We applied the following rules:

1. Start with transaction code S, which the SEC defines as an open-market or private sale.
2. Begin on January 1, 2007 and require mature 30-, 90- and 180-day security-price outcomes.
3. Keep current, non-derivative rows with positive sold shares and non-negative post-transaction holdings.
4. Restrict the primary sample to direct ownership (D). Indirect positions can belong to different trusts, partnerships or other accounts whose denominators should not be combined casually.
5. Aggregate price tranches to one ticker-reporting owner-transaction date-security title-ownership nature event.
6. Sum sold shares and disclosed value, use the maximum reported post-transaction holding, and use the median price outcome where source rows share an event.
7. Calculate returns from the transaction-date market-close baseline, using market_return_30d, market_return_90d and market_return_180d.

The final sample contains 66,043 events across 615 tickers and 9,285 reporting owners, dated January 31, 2007 through January 16, 2026.

Coverage narrows materially from the 499,814 code S rows in the complete release. The study requires direct post-transaction holdings, non-derivative classification and a mature 180-day endpoint. It therefore describes the canonical holdings-covered subset rather than every historical sale.

High versus low relative size

To simplify the comparison, we classified sales below 5% as low-relative-size and sales from 20% through 100% as high-relative-size.

Cohort	Events	Median fraction	Median disclosed value	Median 180D	Positive at 180D
Under 5%	19,632	1.73%	$266,000	+4.76%	60.4%
20% to 100%	23,518	38.97%	$1,242,011	+4.12%	59.8%

The raw difference is -0.64 percentage points. That direction is consistent with the idea that a larger reduction may carry more information, but the economic separation is small and the positive-outcome rates differ by only 0.6 points.

The within-issuer test

Relative sale sizes are not randomly distributed across companies. Founder-controlled issuers, mature compensation programs, equity-backed executives and recurring sellers can occupy very different parts of the distribution.

We therefore isolated 483 tickers with at least three low-relative-size and three high-relative-size events. Each ticker received one vote through this calculation:

median 180D outcome after high-relative-size sales - median after low-relative-size sales

Matched-issuer diagnostic	Result
Matched issuers	483
Median high-minus-low spread	-0.54 pp
Issuers where the high-relative-size median was lower	51.6%
Issuer-bootstrap 95% interval	-1.22 to +0.49 pp

The interval crosses zero, and high-relative-size sales produced the lower median in only slightly more than half of matched companies. The release does not establish a stable within-company penalty from position reduction alone.

Did complete reported exits behave differently?

The upper bucket contains 1,756 events where sold shares equaled estimated pre-sale holdings and the directly reported post-transaction position was zero.

Upper-bucket cohort	Events	Median 180D	Positive at 180D
50% to under 100%	6,189	+4.51%	60.6%
Exactly 100%	1,756	+4.23%	59.3%

An observed 100% is best described as an apparent exit from that directly reported security position. It does not prove that the person eliminated every derivative, indirect or otherwise unobserved economic exposure.

Time-period sensitivity

The high-minus-low direction was negative in both broad periods, but remained modest:

Period	Under 5% median 180D	20%–100% median 180D	Raw difference
2007–2015	+2.50%	+1.41%	-1.09 pp
2016–2026	+5.05%	+4.58%	-0.47 pp

This consistency prevents the pooled result from being explained entirely by one era. It still does not transform the feature into a strong isolated signal.

What the result changes

Nominal value and relative position change answer different questions:

• disclosed dollars measure transaction scale;
• sale fraction measures how much of the reported direct position was reduced;
• post-transaction holdings show what remained in that security and ownership form;
• none of those fields reveals motive, personal liquidity needs or every source of economic exposure.

A serious insider-sale model should retain relative size, but combine it with transaction sequence, trading-plan language, owner history, role, issuer context and related rows in the same filing.

The companion transaction-code study shows why code semantics come first. The screener implementation guide demonstrates how to preserve owner and accession identity before ranking an event.

Limitations

• Outcomes are absolute security-price changes, not market- or sector-adjusted alpha.
• Returns begin from the transaction-date close, not the first tradable price after EDGAR acceptance.
• Post-transaction holdings describe a reported security and ownership form, not total wealth.
• The holdings requirement excludes a large portion of validated legacy sales.
• Code S includes private as well as open-market sales under the SEC definition.
• The analysis does not identify Rule 10b5-1 plans, exercise-linked sales or tax motives.
• Repeated observations from the same owner and issuer are not fully independent.
• Bucket boundaries were selected before interpreting the results, but alternative boundaries can change descriptive medians.

This is an observational event study, not an executable trading strategy or investment recommendation.

Conclusion

A $10 million insider sale can be a small trim. A much smaller sale can close an entire directly reported position. Holdings-relative size is therefore essential context.

But context is not the same as a universal bearish multiplier. In this release, larger position reductions were followed by slightly lower raw outcomes, while rank correlation was nearly zero and the matched-issuer interval crossed zero.

The defensible use is to preserve the feature and combine it with filing semantics, owner history and related transactions—not to assume that every large percentage sale predicts a drawdown.

Researchers can inspect the 64-field schema in the free 10,000-row sample or query targeted Form 4 records through the InsiderAlpha API.