How are Search Fund Investors Really Faring?
Daniel Lazier, Jacob Thomas, A. J. Wasserstein. (2025)
Yale Case, no number printed, October 27, 2025
Sample. "we used a valuable data set provided by 12 investors representing 23 funds" (p5), masked A to L, not named. "We solicited a wide range of serial search fund investors for data, and some chose to participate while others did not." (p6). "our data includes 1,192 observations in which an investor made an entrepreneur selection decision and a subset of 768 observations where an investor had to make a capital allocation decision." (p5). Observations are investor-deal pairs, not unique companies: "Some of our observations are duplicated, meaning investor B saw the same opportunity as investor J." (p5). "The broken searches represent $17 million in capital invested out of a total of $700 million deployed." (p9). Time window, vintages, geography and response rate: not stated. Included and excluded. Includes broken searches (the 1,192) and completed deals (the 768). Realized and unrealized are mixed: "If the deal is still operating, we use a current mark provided by the investor." (p6); the split is not stated. MOIC only (some investors did not share IRRs). Gross of fund fees. Traditional versus self-funded mix of the sample: not stated. Deals the investors passed on are unobserved. Headline numbers. "the 1,192 entrepreneur picks resulted in 768 actionable deals, yielding a 64% conversion rate." (p9). "Out of 1,192 entrepreneurs, 612 (51%) deals ultimately generated a positive return of at least 1.0x MOIC." (p9). "only 31% of the 768 completed deals generated a MOIC greater than 3.0x" (p9). "The mean MOIC including broken searches is 2.0x; the mean MOIC excluding broken searches is 2.78x. The median MOIC including broken searches is 1.0x, while the median MOIC excluding broken searches stands at 1.6x. Lastly, the weighted average MOIC for all investors across all funds is 2.50x." (p9). "The percentage of MOIC observations above 10.0x is just 2% (a one-in-fifty outcome)." (p7 to p8). Across the 12 investors' lifetime results: "The mean and median MOICs are 2.58x and 2.39x, respectively." and "Our maximum is 4.32x, which is lower than all the aggregate MOICs reported across different Stanford studies in Exhibit 1." (p11). "only five out of 23 funds have griffin exposure." (p15). Against Stanford (authors' own comparison, p18 to p20). "our sample data set consists of 1,192 observations, which is 14% of the potential Stanford observations." "The macro numbers are nearly identical, with similar acquisition rates of 63% and 64% and similar gain and loss rates at 69% and 68% and 31% and 32%, respectively." "we are seeing far fewer griffins (3%) than in the Stanford data set (11%)." "our reported MOIC (2.5x) is significantly lower than Stanford's (4.5x)." "We do not fully know why this is the case and cannot reconcile and disentangle the differences." Limits. Authors: "All our data comes from a dozen investors, which could introduce sampling bias."; "It is possible that investors with poorer results might have opted not to engage in the project."; "they are only directional at best and contain some inherent flaws." (p6). Reader's notes: the text calls the 768 a "768-company sample" (p15) although they are investor-deal observations with duplicates, and the number of unique companies is never given; the 95% interval on the mean treats duplicated observations as independent; griffin share is printed as 2% in one place and 3% in another; an unknown share of MOICs are unrealized investor marks; "at least 1.0x" counts exactly 1.0x as a positive return. The 4.5x is Stanford's number, not Yale's.