Search Fund Return Dispersion: Headline Aggregates Mislead

Search fund return dispersion describes how widely investment outcomes differ, including losses that a favorable aggregate can obscure. Concentrated exposure leaves an investor vulnerable to the businesses they actually own, while exceptional outcomes elsewhere can pull a reported aggregate upward. Read the spread of outcomes and the population behind it before treating any headline as representative.

Among 337 search-acquired companies, 26.4% had a reported total or partial loss and 8.3% fell in the highest reported ROI band¹. The study’s chart includes operating companies with unrealized values as well as terminal outcomes¹. CapitalPad calculated the shares from Stanford GSB’s 2026 study, Exhibit 7.

Acquisition outcomes are spread across losses and gains

Study outcomes among 337 search-acquired companies, operating and exited: total or partial loss 26.4%; 1–2x 20.2%; 2–5x 27.6%; 5–10x 17.5%; 10x or more 8.3%. Outcomes are spread rather than clustered around an aggregate.

337 search-acquired companies · Share of companies (%)

Total or partial loss26.4%

1–2x return20.2%

2–5x return27.6%

5–10x return17.5%

10x or more return8.3%

Source: Stanford GSB 2026 Search Fund Study, via CapitalPad statistics page (updated September 14, 2026); 337 search-acquired companies, operating and exited¹. Study figures for the asset class, not the performance of any fund, network or firm.

Figure data

Acquisition outcomes are spread across losses and gains

Outcome	Share of 337 acquired companies
Total or partial loss	26.4%
1–2x return	20.2%
2–5x return	27.6%
5–10x return	17.5%
10x or more return	8.3%

The figure’s “returned” wording describes a reported ROI band, not necessarily cash proceeds. Operating-company values in the chart are unrealized¹.

What dispersion means

Dispersion concerns the distance between outcomes. An aggregate compresses those differences into a summary. That summary can be useful for describing a research population, but it cannot show how every participant fared or which outcome an investor will experience.

Consider the distinction without assigning hypothetical returns. A research population can contain businesses that lost capital, businesses that returned capital with modest gains, and businesses that produced exceptional gains. Combining them can yield an attractive aggregate even though the experiences of their owners differed sharply.

The practical risk is reading that aggregate as the result of a typical investment. The word “average” often encourages this shortcut, especially when a recap moves quickly from historical research to an investment opportunity. A published aggregate, an arithmetic mean and a median are different descriptions. The original methodology must establish which description applies.

Concentration makes the distinction personal. In Investing & Building a Portfolio of SMBs on The SMB Investor podcast, Grant Hensel describes business owners whose wealth depends heavily on the company they operate. The relevant concern is dependence on that business: an adverse outcome can affect much of their financial position.

That observation does not establish that diversification preserves expected returns, eliminates losses or reproduces an academic aggregate. It identifies a risk category. An investor can acknowledge concentrated exposure without claiming to know the right portfolio size or the future outcomes of a broader portfolio.

How exclusions change an aggregate

Stanford GSB’s 2026 search-fund study reports that excluding the top 10 percent of funds by ROI leaves an aggregate ROI of about 2.1x². This is an academic asset-class result, not a return earned by any particular investor.

The study separately describes removing funds with ROI of 10x or more². That exclusion uses a return threshold. Removing the highest-ranked share uses a position within the distribution. These are different filters, and the resulting populations must not be treated as interchangeable.

This distinction matters because an exclusion is part of the claim, not an optional footnote. Changing the filter changes what remains in the calculation. A reader who remembers only that “the winners were removed” can accidentally compare unlike results or attribute a result to the wrong population.

The reported residual aggregate is also not a forecast for an investor who misses exceptional outcomes. It describes a retrospective calculation with an identified exclusion. It does not establish what a prospective collection of investments will return, how that collection will be assembled or whether its outcomes will resemble the historical remainder.

Nor does removing exceptional results make them irrelevant. The point of examining the exclusion is to understand how much an aggregate depends on the upper end of its observed distribution. Both the full population and the filtered population answer descriptive questions. Neither supplies a personal return expectation.

Reading search fund return dispersion without a forecast

The useful reading task is to keep the reported result attached to its definition. If a figure travels into a presentation without the population, measurement basis and exclusion, it has lost information needed to interpret it.

The following distinctions help preserve that information:

Research statement	What it helps describe	What remains unanswered
An aggregate combines observed outcomes.	It summarizes the population under the stated method.	It does not show the result experienced by a typical participant.
A calculation excludes the highest-ranked funds.	It examines the remainder after a rank-based exclusion.	It does not identify future exceptional outcomes.
A calculation excludes outcomes above a stated return boundary.	It examines the remainder after a threshold-based exclusion.	It does not necessarily remove the same observations as a rank-based filter.
A breakdown separates loss outcomes from gain outcomes.	It shows how observations fall into reported categories.	It does not establish an individual investor’s probability of loss.

Read the metric label with the same care. A multiple describes a relationship between value and invested capital under the source’s definition. Internal rate of return incorporates the timing of cash flows. A multiple alone does not describe how long the capital was committed, and an annualized measure does not describe the full spread of outcomes.

The study’s labels should therefore remain attached to the figures. Avoid casually relabeling an aggregate as a median, a realized investor return or another familiar metric. Similar-looking labels can conceal differences in what was measured.

The study reports aggregate investor cash flows rather than an average of individual fund returns, and its calculations include operating companies with estimated values. Those distinctions matter when deciding what further information belongs in an investment conversation.

What acquired-company outcomes show

CapitalPad’s statistics page calculates that 26.4 percent of the 337 search-acquired companies in Stanford GSB’s 2026 Exhibit 7 had a reported total or partial loss¹. The chart includes both operating companies with unrealized values and terminal outcomes¹.

The denominator is acquired companies in that chart. It is not all searches, all capital commitments or all investors. The statement also combines total and partial losses; it does not say that every observation in the loss category lost all invested capital.

Those boundaries prevent a common misreading. An acquisition outcome requires a business to have been acquired. A search that ended without buying a business presents a different question. Treating an acquisition loss share as a failure rate for the entire search process erases that distinction.

The guide to broken searches and acquisition risk addresses that earlier stage. Here, the narrower issue is what the acquired-company population can tell us about the spread of reported outcomes after acquisition. Keeping those questions separate allows each source to retain its actual scope.

A share of acquisitions is also different from a share of invested capital lost. The cited category describes observations, while a capital-loss measure would require information about amounts invested and recovered. The combined category does not provide that calculation, and this article does not estimate it.

Operating-company values in the chart are unrealized estimates. A reader cannot assume the breakdown represents the eventual outcomes of those investments or the cash proceeds received by investors.

The downside lesson is specific: loss outcomes appear within the stated acquisition population. It would go beyond the evidence to turn that observation into a prediction about a prospective opportunity, a platform or a particular LP’s experience.

What an LP cannot infer

On The SMB Investor podcast, in Inside the Search Fund Playbook, John Steinberg discusses Stanford research as evidence about the search-fund model. The distinction relevant here is between an asset class and a specific investment. This paraphrase carries no numerical claim from the episode.

Historical asset-class research cannot establish the quality or prospective result of a particular fund, sponsor or business. It can inform questions about the model while leaving the individual opportunity unassessed. Moving from the research population to a specific investment requires evidence about that investment.

The same boundary applies when comparing ways of participating in SMB investing. A description of search funds cannot simply be transferred to every independent sponsor or direct co-investment. Our search-fund versus independent-sponsor comparison explains the roles; the research population still determines where a statistical claim belongs.

The evidence here does not supply an allocation recommendation, a required number of investments, a return target or a simulation. It also does not show that an investor can obtain the historical mix of outcomes by selecting opportunities now. Retrospective identification of exceptional results differs from knowing them in advance.

These figures are not SMB Investor Network results. They do not describe the performance of investments available through any particular platform. Past asset-class performance does not guarantee future results, and a partnership attribution does not change the scope of academic research.

Questions to keep beside search fund return figures

Use a short reading note to preserve context when a statistic is useful enough to repeat:

• Name the source chain. Identify the academic research and the vendor recap through which the claim was obtained.
• Keep the population attached. Record whether the statement concerns funds, searches, acquisitions or another defined group.
• Preserve the exclusion. Copy the distinction between a rank-based removal and a return-threshold removal accurately.
• Check the measurement label. Record what the source calls the result and which methodological details remain unverified.
• Keep loss categories intact. Do not turn combined total and partial losses into a total-loss claim.
• Record the missing context. Note which businesses are still operating and when their values were measured.

These notes organize public evidence. They are not a sponsor scorecard or a procedure for approving an investment. Their value is traceability: another reader should be able to find the same claim and understand the limits attached to it.

Episodes cited

• John Steinberg, Inside the Search Fund Playbook, The SMB Investor podcast. This article uses the observation only for the distinction between asset-class research and a specific investment.
• Grant Hensel, Investing & Building a Portfolio of SMBs, The SMB Investor podcast. This article uses the observation only for the qualitative concern about concentrated business exposure.

Published in partnership with SMB Investor Network.

Continue with search-fund study comparability, return to the search-fund returns research guide.