A handful of deals carry a venture fund's return

The record

In a January 2001 NBER working paper, economist John Cochrane measured returns on individual venture financing rounds using a maximum-likelihood model that corrects for the fact that failed startups rarely report a value. Even after that correction, he found the underlying distribution of returns to an IPO or acquisition to be highly skewed, with an average return of 698% and a standard deviation of 3,282% around it. In his words, "there are a few truly outstanding returns of thousands of percent and many more modest returns of 'only' 100% or so." A quarter-century later, Carta's Q1 2026 fund performance report, covering more than 2,700 funds it administers, describes a similar shape at the fund level: across vintages from 2017 through 2024, 90th-percentile net IRR exceeds 20%, while the 75th percentile never rises above 15.5%.

What the sources establish

The two sources measure different things — one prices individual financing rounds from a historical sample, the other benchmarks whole fund vehicles today — but both find that a mean or average figure is pulled upward by a small share of outcomes rather than reflecting a typical result. Cochrane's paper notes that because the distribution is so skewed, arithmetic average returns are "much higher than geometric averages," meaning a quoted mean overstates what a representative dollar experienced. Carta's percentile breakdown makes the same point in fund terms: most vehicles in a given vintage cluster well below the top performers, and the report states that "only a small minority of vehicles are achieving the sorts of performance that many LPs expect."

Scope and revision

Cochrane's estimate covers financing-round outcomes for a historical sample of venture-backed companies and explicitly measures IPO-or-acquisition events, not whole-fund cash returns; it says nothing about fees, fund duration, or rounds that never reached either exit. Carta's percentiles cover only funds administered on its platform, are not a full census of U.S. venture funds, and describe unrealized as well as realized performance since they are IRR figures drawn from reported NAVs. Neither source states a single power-law constant that applies across periods; both show concentration without quantifying it as one formula, and combining them into a single claim about "venture returns" would overstate what either individually supports.

The decision in front of you

Editorially: a founder pitching for capital, or an angel sizing a check, should treat an investor's stated target return as a portfolio-level bet on a small number of outsized outcomes rather than an expectation set for any individual company, including the one being pitched.

• Is the return being cited a mean, which skew inflates, or a median, which better describes a typical outcome?
• Does the claim describe individual deal returns or whole-fund performance, and over what vintages?
• What share of the underlying sample failed to reach even a return of capital?

Both records point the same direction: venture returns concentrate in a minority of outcomes, at the deal level and at the fund level, which is a reason to be skeptical of any single quoted average that does not also disclose its spread.