Private Markets · Manager Selection
What Actually Predicts Returns in Private Equity, Venture & Private Debt
Alternative investments — private equity, venture capital, private credit, real assets — are simply the parts of the market that sit outside daily-traded stocks and bonds: capital committed for years, no live price to watch, and a wide spread of outcomes.
They are not a free lunch, and the returns are not a given. In principle, investors are paid for the lock-up and the work it takes; in practice that compensation is uneven and far from guaranteed. The reasonable case for a modest allocation — diversification, access to companies and credit that never list, returns that don't move in lockstep with public markets — sits against real costs: illiquidity, higher fees, complexity, and a long wait to find out whether it worked.
What does set the asset class apart is dispersion — the gap between a good manager and a poor one is far wider, and more persistent, than in public markets. The work, then, is less in the allocation than in the selection. That is what this primer examines: drawing on more than 95,000 private funds, the factors that have actually moved returns, and how much the evidence supports each.
The findings
We tested roughly ninety factors against realised returns across 13,814 funds with a reported IRR, correcting for multiple testing. A reviewer pass then caught a genuine mechanism error and an overstatement — at points 6 and 10 — both since corrected. What remains are the observations that hold.
1 — Larger funds tend to return less
Across every asset class, larger fund size correlates with lower IRR. A $10bn fund must deploy $10bn; there are not enough great deals at that size, so mediocre ones get included. The penalty becomes visible above $5bn and severe above $10bn. Consistent with Kaplan, Harris & Jenkinson.
view — we treat mega-flagships above $5bn with some scepticism for return-seeking allocations; sized for stability, not alpha.
2 — A published target IRR has some signal
Only about 5% of funds publish a target IRR, but when they do, the rank correlation with realised IRR runs +0.33 to +0.46 across asset classes. The honest framing: willingness to publish is itself a signal of confidence and accountability (selection), and the targets carry real predictive power (causal). Our best estimate is roughly 70% causal, 30% selection.
view — we read a published target as a statement about the manager's conviction and accountability — not merely a number to back-test.
3 — Earlier funds in a series tend to outperform later ones
A manager's Fund III tends to edge out their Fund VI or Fund X. The correlation is small (−0.07 to −0.12) but consistent across asset classes, and matches Kaplan & Schoar's persistence-decay results. The mechanism is largely mechanical: successful early funds attract bigger follow-on raises — and larger funds, per point 1, tend to return less.
view — on balance we lean toward Fund III–V from a smaller top-quartile manager over Fund X from a giant.
4 — Industry and region carry signal; "strategy" alone does not
Industry exposure and geographic focus are the two categorical factors that are statistically significant across all three asset classes. A "buyout" fund in healthcare looks nothing like "buyout" in technology. Supported by Cressy, Munari & Malipiero.
view — we tend to favour industry-specialised managers operating in their wheelhouse over generalists.
5 — Standard fees track quality; they don't cause it
Managers who hold full standard economics — the familiar "2-and-20" — rather than discounting to raise capital, correlate weakly but positively with IRR (+0.12 to +0.14). It reads best as a quality signal: the strongest managers don't need to cut fees, and tend to deliver. The stronger framing — "never fee-shop in PE" — overstates what the correlation supports.
view — we wouldn't reject a strong PE manager over premium fees; equally, we wouldn't pay premium fees for an unproven one.
6 — Low hurdles signal elite brands; high hurdles signal mid-pack
In VC, low-hurdle funds outperform high-hurdle ones (Spearman −0.38, n=50; partial −0.39 after controlling for size and vintage). The earlier behavioural story — "high hurdles force early exits and kill returns" — was mechanically wrong: IRR rises when distributions come early, yet the hurdle-versus-MOIC correlation is just +0.03. The real driver looks like selection. Elite brands do not need to offer hurdles to attract LPs; mid-pack institutional VC does. The hurdle proxies for fund quality, not behaviour.
view — there's no need to screen out low-hurdle funds — they are often the elite brands. And a high hurdle isn't necessarily LP-friendly virtue; it usually signals the GP is competing on terms.
7 — The firm-age effect is composition-driven
Manager firm age correlates negatively with VC returns (−0.25) — but the interpretation is nuanced. Older VC firms split into elite survivors that keep their edge and zombie franchises that should have wound down; zombies outnumber survivors in aggregate, producing the negative correlation. Sequoia, Index, Benchmark and Kleiner Perkins are the survivors, not the rule.
view — we wouldn't apply firm age as a blanket screen; the question is whether a given franchise is a survivor or a zombie.
8 — EBITDA targets are a label check, not a return predictor
VC funds that target larger-EBITDA companies tend to underperform — but the effect is partly definitional, because those funds are usually growth equity wearing VC clothing. It reads better as a screen for mislabelled mandates than as a statement about venture return drivers.
view — where a "VC" fund's stated target EBITDA looks like growth equity, we reclassify it before benchmarking.
9 — Private debt is structurally lower-return
The private-debt universe median is 9.7% IRR, the top quartile 13.1%, the top decile around 18%. Dispersion is far tighter than in equity — a standard deviation near 7.5%, against roughly 14% for PE and 25% for VC. It is an income-producing asset class; credit yields are bounded by rates plus an illiquidity premium.
view — we don't apply PE benchmarks to private debt: top-quartile is 13%+, 10–12% is solid, and 20%+ implies equity-like risk wearing a credit label.
10 — In private debt, size matters little; selection dominates
Private debt shows a weak negative correlation between manager AUM and returns (−0.07, marginally significant). The full picture by AUM bucket is modest:
| Manager AUM | Median IRR |
|---|---|
| Sub-$1bn | 10.4% |
| $1–5bn | 10.5% |
| $5–20bn | 10.0% |
| $20–100bn | 9.8% |
| >$100bn | 9.5% |
view — we wouldn't dismiss mega-platforms as a class. Within the private-debt segment, selection matters more than size.
11 — Strategy choice drives the private-debt return ladder
Among defensible private-debt managers, IRR varies sharply by strategy — and within a strategy, smaller-borrower books tend to outperform (max-EBITDA correlation −0.15, pointing to structural pricing power in lower-middle-market lending).
| Strategy | Representative managers | Median IRR |
|---|---|---|
| Venture debt | Western Tech · Vistara · Inveready | 15–29% |
| Distressed | EAAA · Silver Point · Black Diamond · Anchorage | 16–26% |
| Specialty mezzanine | VSS · Centerfield · NewSpring | 12–25% |
| Direct lending | Northern Arc · Plexus · Perceptive | 12–17% |
view — in private debt we treat strategy selection as setting the ceiling — chosen deliberately, before the manager.
Das Family Office · Investment Office
Speak with us
To discuss how a private-markets allocation might sit within a specific portfolio — or to receive future research as it is published — leave a note below. Enquiries reach the team directly. For accredited investors.