Private Market Returns: Read the Spread, Not the Average

The asset-class average is the least useful number a self-directed allocator reads. In Cambridge Associates' vintage tables as of June 2020, the median vintage-level upper-to-lower-quartile gap among legacy US private equity funds (a category broader than buyout) across reported 2001-to-2014 vintages is 12.3 points, and the venture gap is about the same size at 13.2 points¹². Real estate isn't far behind at 10.5 points³. Over the 20 years through June 2022, public equity managers sit about 1 point apart⁴. Read a single average return for "private equity" and you've thrown away the number that actually describes the risk.

I build the software lower-market sponsors and investors use, and I read a lot of vintage tables and benchmark books doing it. This post is my own arithmetic on three of Cambridge's published tables, not a house view on any asset class. Every number below carries the evidence id it came from.

The average hides the fund you will own

Same vintage years, same asset class: the lower-quartile cutoff for legacy US private equity funds averaged 7.1 percent a year and the upper-quartile cutoff 20.7 percent². That's not two different asset classes. It's the same benchmark category, the same years, a different manager.

Fund IRRs don't compound into a clean multiple the way a savings-account rate does, so treat that 13.6-point gap as a rate spread, not a dollar promise². But the point stands: these private-market tables show a substantial spread between upper- and lower-quartile cutoffs within each strategy.

Takeaway: these private-market tables show substantial within-strategy spreads in fund outcomes.

How wide is the spread, by asset class

Cambridge's vintage tables for 2001 to 2014 show the same shape across three strategies, at different widths. US private equity (legacy definition: buyout, growth, subordinated capital, PE energy): average lower quartile 7.1 percent, median 13.3, upper quartile 20.7, a 12.3-point median spread². US venture: 2.1 / 8.6 / 16.4, a 13.2-point spread¹. Real estate: 4.1 / 9.5 / 14.9, a 10.5-point spread³.

Compare that with the rest of the market. Over 20 years to mid-2022, US large- and small-cap growth public fund managers spread by 1.2 points⁴. Senior private credit spread by 4.3 points⁴. Multi-stage venture spread by 18.2 points, and early-stage venture by 22.4⁴. A separate PitchBook cut of 3,368 funds agrees on the ordering: venture's IRR standard deviation ran 34.2 percent, real estate, growth, buyout and secondaries ran 17.0 to 21.1 percent, and credit ran 9.2 percent⁵. In the figure below, "buyout" is shorthand for Cambridge's broader legacy US private equity benchmark category².

Same vintage years, different fund

A horizontal range chart with five rows on a scale from minus 5 to 25 percent net IRR. US PE legacy averages a lower quartile of 7.1, a median of 13.3 and an upper quartile of 20.7, a 12.3 point spread with no vintage below zero. US venture averages 2.1, 8.6 and 16.4, a 13.2 point spread, and the lower quartile fell below zero in 6 of 14 vintages. Real estate averages 4.1, 9.5 and 14.9, a 10.5 point spread. Senior private credit and US growth public funds show spread width only, from a different source: 4.3 points and 1.2 points.

US buyout median, 2001 to 2014 average: 13.3 percent.

Averages of vintage-year quartiles, not a ranking of asset classes. Credit and public rows are a different source and show spread width only. Net IRR.

0:00 / of 0:16

Source: our calculation from Cambridge Associates vintage-year quartile tables, 2001 to 2014 average (dispersion-cambridge-vintages.py)¹²³. Opto Investments citing S&P and Burgiss, 20 years to Jun 30, 2022, spread only⁴.

Figure data

Same vintage years, different fund

Strategy	Lower quartile	Median	Upper quartile	Spread (upper minus lower)	Vintages with lower quartile below zero
US PE legacy (buyout, growth)	7.1%	13.3%	20.7%	12.3 pts	0 of 12 vintages
US venture	2.1%	8.6%	16.4%	13.2 pts	6 of 14 vintages
Real estate	4.1%	9.5%	14.9%	10.5 pts	3 of 14 vintages
Senior private credit (20y to 2022)	—	—	—	4.3 pts	—
US growth public funds (20y to 2022)	—	—	—	1.2 pts	—

Averages of vintage-year quartiles, not a ranking of asset classes. Credit and public rows are a different source and show spread width only. Net IRR.

The figure's private-equity spread refers to Cambridge's legacy US private equity sample, which includes buyout, growth equity, subordinated capital and PE energy, and covers reported 2001-to-2014 vintages. For private equity and venture, the quartile markers are averages across vintages and the spread is the median vintage-level gap¹².

Takeaway: dispersion is a property of the strategy, and credit and public funds are a different kind of decision from buyout and venture.

Same spread, different floor

Legacy US private equity and venture have similar interquartile spreads, but they don't share a floor. Venture's lower quartile fell below zero in 6 of 14 vintages; legacy US private equity's never did across 12 vintages with quartiles reported¹². The Kauffman Foundation found 62 of its 100 venture funds failed to beat public markets after fees, and no fund over $500M in that portfolio returned more than 2x⁶.

Dispersion also moves with the era. For an illustrative, hypothetical comparison outside E770's measured sample, suppose a 1990s venture spread were 59 to 76 points; that would be far wider than the 2001-to-2014 average¹. And the averages aren't uniformly bad news: across large datasets, average net PE multiples cluster around 1.55 to 1.63, roughly matching public indices since 2006⁷, and a separate study of nearly 1,400 funds found average buyout beat the S&P 500 by 20 to 27 percent over fund life⁸.

Takeaway: the width of the spread tells you how much manager selection matters; the floor tells you what a wrong pick costs.

Small funds, small deals: wider tails, same middle

"Small funds outperform private equity" is one of the most repeated lines in this space, and it's an average-of-a-skewed-distribution statement. Across 10,276 funds and $8.7 trillion committed, mean PME (a public-market-equivalent multiple) falls with size: 1.53 for the smallest quartile down to 1.29 for the largest⁹¹⁰. Median PME barely moves: 1.17, 1.13, 1.14, 1.18 across the same four quartiles⁹¹⁰. The standard deviation of PME is 57.6 percent lower in the largest-fund quartile than the smallest¹⁰.

Small funds beat large funds on average. The typical fund does not.

Four fund-size quartiles by Kaplan-Schoar PME, smallest to largest. Mean PME falls from 1.53 to 1.41 to 1.34 to 1.29. Median PME stays flat: 1.17, 1.13, 1.14, 1.18. The standard deviation of PME is 57.6 percent lower in the largest quartile than the smallest.

Mean PMEMedian PME

Dispersion 57.6 percent lower in the largest quartile than the smallest [E776].

Source: Brown, Fermand, Hu, Maxwell, Lundblad, Volckmann, Scale, Scope, and Speed in Private Capital Funds, UNC Institute for Private Capital (2024), Kaplan-Schoar PME by fund-size quartile, 10,276 funds¹⁰.

Figure data

Small funds beat large funds on average. The typical fund does not.

Fund-size quartile	Mean PME	Median PME
Q1 (smallest)	1.53x	1.17x
Q2	1.41x	1.13x
Q3	1.34x	1.14x
Q4 (largest)	1.29x	1.18x

Kaplan-Schoar public market equivalent; above 1.00x beat public equities.

Buyout specifically: small funds averaged 1.38x versus 1.28x for large¹¹. At the deal level, small- and mid-cap buyouts averaged 2.8x gross versus 2.4x for large-cap, and small-cap deals carried both the largest share of sub-0.5x losses and the largest share of 5x-plus winners¹². The typical outcome and the average outcome are different questions, and small funds are exactly where they diverge most.

Takeaway: "small funds outperform" is true of the average and roughly false of the typical fund; the gap between those two facts is the tail risk you're taking on.

Where the lower middle market sits

A lower entry price may contribute to the higher average posted by small deals. GF Data's 2025 sample shows average TTM EBITDA multiples stepping down with size: 5.9x at $10M to $25M in total enterprise value versus 10.0x at $100M to $250M¹³. Cheaper entry may be a tailwind.

The SBIC sample also shows a wide range of outcomes. Small Business Investment Company funds, a studied small-business investment population, averaged a 16.9 percent IRR with a standard deviation of 10.4 points and a range from negative 13.4 to positive 73.3 percent¹⁴. Recent SBIC vintages (2021 and later) are still sitting at a 1.0x multiple, which the researchers attribute to the ordinary J-curve¹⁵.

Takeaway: smaller transactions had lower entry multiples, while the SBIC sample showed a wide range of outcomes¹⁴¹³.

Our guide to a private market portfolio covers how to map a sleeve like this against the rest of your holdings before you decide how many positions it needs.

Can you pick the top quartile?

Here's the finding that should change how you read a pitch deck. For post-2000 buyout funds, picking a manager because their last fund was top quartile shows little or no persistence, using only information available at the time you'd have made that decision. What little persistence exists is driven by the bottom quartile repeating, not the top¹⁶. Venture persistence does hold up better under the same test¹⁶.

Funds also take a while to settle. Cambridge notes that most funds need six-plus years to reach their final quartile ranking, and sit in two or three other quartiles first¹⁷. Kauffman found the same instability inside a single fund's life: aggregate reported IRR in its portfolio peaked around month 16¹⁸, and an early negative IRR predicted the final result no better than chance would¹⁸.

Takeaway: a "top quartile" claim in a pitch deck can reflect a ranking that changes before it becomes final, and the buyout study found modest persistence driven by the bottom quartile rather than the top¹⁷¹⁶.

How this breaks

• Reading a vintage-year median as "the return of private equity" and sizing a sleeve on it.
• Comparing an IRR quartile from one benchmark provider against another provider's quartile. Different samples, different definitions¹⁷⁴⁵.
• Treating a young fund's current quartile as settled, when most funds still have several years of re-ranking ahead¹⁷.
• Holding one small fund or one direct deal and expecting the average when the observed outcomes have wider low and high tails¹⁰¹².
• Reading concentration risk in a private sleeve the same way you'd read it in public stocks. Our concentration guide covers the difference between overlap and outcome variance.

Takeaway: every number in this post describes a distribution, and you own one draw per commitment.

What I'd check

I'm not an LP in any of these funds and I don't advise anyone on where to put money. Here's how I'd read this data if I were the one deciding on a sleeve.

1. I'd stop quoting a single return for "private equity." The median and both quartiles from one provider show the observed range for a specific strategy and vintage, including how far the bottom quartile sits below the average.
2. A bottom-quartile outcome shows how far a result can fall below the median in the observed sample. For venture, that means treating a negative IRR as a live possibility, since it happened in 6 of 14 vintages¹.
3. I'd treat "top quartile" in any deck as a question, not a fact: which benchmark, which date, how old was the fund when it was ranked¹⁷¹⁶.
4. For direct small-business deals, the observed outcomes include both sub-0.5x losses and 5x-plus winners¹².
5. I'd ask every manager for realized results next to the marks, and the losers next to the winners, before I read a single average return.

Related reading: how commitments actually turn into cash calls and distributions is the subject of the private equity J-curve, modeled for one household, the companion piece to this post.

Source notes

The vintage-table arithmetic in Figure 1 is our own calculation from Cambridge Associates' published benchmark books; Figure 2 uses fund-size PME data from Brown et al.¹⁰. This post makes no recommendation of any security, fund, manager or allocation.