One number matters more than you think: N, the number of startups you hold. In power‑law markets like pre‑seed/seed, return is a function of whether you include the rare outliers. If you want the market’s IRR, you need a portfolio big enough to reliably catch those outliers. The mistake most seed funds make isn’t price or pick—it’s N.
What “indexing” means in venture (vs. public markets)
In public equities, covered in this WSJ article, indexing means you buy a cap‑weighted basket and harvest the market’s return. In venture, you can’t mechanically buy “the market,” but you can approximate beta by holding a large, stage‑consistent sample of deals and avoiding concentration risk. Practically, that looks like 50–150 initial positions per fund at seed (or exposure to hundreds via a platform/FoF), not 15–30. The goal isn’t to out‑select; it’s to out‑include. Institutional Investor
The base rates that drive portfolio design
Power‑law outcomes at the deal level
The best long‑run dataset we have (Correlation Ventures) shows how skewed outcomes are: ~65% of financings lose money, ~25% return 1–5×, ~6% return 5–10×, and <4% return ≥10×. Only ~0.4% exceed 50×. The implication is blunt: if you run a small portfolio, the odds say you will miss the deals that drive the asset class. Medium
Seed vs. Series A/B dispersion
Earlier entry has a lower “hit rate” but drastically higher upside per hit. AngelList’s stage‑cut of unicorn investments: the median multiple at pre‑seed/seed was ~31× (vs. ~9× when you mix in later entries). That multiple compression is why seed requires more shots—and why small seed portfolios are fragile. AngelList
What is the “index” IRR at early stage?
No single source publishes an official “seed beta,” but triangulation from institutional datasets is instructive:
- Cambridge Associates (CA) U.S. Venture Capital—Early Stage Index (institutional‑quality funds): ~20.6% 5‑yr and ~18.7% 10‑yr annualized (as reported by American Century from CA’s indices). That’s higher than CA’s overall VC index over the same horizons. American Century Investments
- CA U.S. Venture Capital Index (all stages) as of 12/31/2024 shows ~15.1% 5‑yr and ~13.7% 10‑yr net to LPs—dragged by 2021–2023 vintage marks. Early‑stage consistently screens higher than later stages over multi‑year windows. Cambridge Associates
- Over very long windows, CA’s data still shows venture’s significant value‑add vs. publics (e.g., WSJ citing CA’s 25‑year VC return beating public benchmarks), but the last decade’s denominator effects and the 2021 bubble hangover compress headline IRRs. Read: long‑run attractive, short‑run cyclical. The Wall Street Journal
Takeaway for LPs: a reasonable “index‑like” expectation for seed‑heavy exposure sits in the mid‑ to high‑teens net IRR across cycles, with periods that print 20%+ and downturn windows that print single digits. Early‑stage > later‑stage on average, but with more dispersion.
The math: how big must N be to “catch the index”?
If p is the base‑rate probability that a seed check becomes a 10× outcome (use 3–4% from Correlation data), the probability a portfolio of N independent seed bets captures at least one 10× is:
- Using p = 4%, you need N ≈ 74 for a 95% chance of at least one 10×.
- Using p = 3%, you need N ≈ 99 for the same 95% target.
- For 50× hits (p ≈ 0.4%): N ≈ 748 for 95% confidence; N ≈ 500 gets you ~86% odds.
This is why 50–150 seed positions per fund (or exposure to hundreds via platforms/FoFs) is a rational design if your objective is beta capture rather than heroic alpha. An Institutional Investor Monte Carlo using these base rates concludes the “golden rule” for LPs is to build exposure to ~500 startups over time; 100 is the minimum to materially reduce miss risk. Medium
Translation: most 20–30 check writing seed funds are under‑diversified for the job of matching the market’s hit distribution. You can be skilled and still whiff, simply because your N is too small.
Seed vs. Series A/B: different game, different N
- Seed/pre‑seed: highest upside per winner; lower per‑deal hit rate; needs large N to avoid missing outliers. Best suited to index‑like or “index‑plus‑follow‑on” portfolio construction. Medium
- Series A/B: higher selection information, lower upside per winner, lower dispersion. Later‑stage/expansion funds historically post ~low‑teens net IRRs over long spans—good, but structurally lower than seed‑heavy portfolios. This is consistent with CA’s stage cuts and industry analyses. American Century Investments
Implication: the case for “more positions” is strongest at seed. At Series A/B, larger N still helps, but the marginal benefit shrinks because upside per hit is smaller and dispersion is lower.
6) IRR → MOIC: what does “hitting the index” look like?
For a 10‑year fund life, compounding (gross) looks like:
| IRR | 10‑yr MOIC |
| 15% | ~4.05× |
| 20% | ~6.19× |
| 25% | ~9.31× |
| 30% | ~13.79× |
So a seed “beta” in the high‑teens to ~20%+ net translates to ~4–6× net MOIC depending on fees, pacing, and recycling. Hitting 25%+ net requires either fortuitous timing, alpha in follow‑ons/ownership, or unusually rich vintage effects. (CA’s recent 5–10‑year VC horizons underscore how 2021–2023 marked valuations can pull headline IRRs down; that’s cyclicality, not a refutation of the base case.) Cambridge Associates
“Won’t indexing blunt the upside?” The data says: bigger portfolios often do better
- AngelList LP cohort analysis: median IRR rises with portfolio size; a 100‑investment portfolio’s median outperformed a single‑deal portfolio by ~900 bps per year, and investors who used the Access Fund (broad index exposure) “typically and in expectation outperform” those cherry‑picking. Worth noting this is even with the adverse selection bias of AngelList, which misses many of the extreme outliers. AngelList
- Institutional simulations: ultra‑diversified funds had higher median returns and dramatically lower left‑tail risk; the 5th‑percentile outcome for 500‑deal portfolios ≈ the median for 15‑deal portfolios. In a hits‑driven market, diversification is not a concession—it’s a weapon. Institutional Investor
Important nuance: “Index‑like” at seed doesn’t mean undifferentiated. The scalable pattern that works in practice is Index + Rules: broad entry plus systematic follow‑ons into traction, and enough ownership to let a small number of names drive returns. (Accelerator platforms and some YC‑focused funds operationalize this.) rebelfund.vc Worth noting the YC itself invests in 500+ startups per year at the Pre-Seed stage.
Why seed indexing feels contrarian (and why it persists anyway)
- Narrative gravity: LPs and GPs like the story of concentrated “picking winners.” But the modal seed outcome is loss; the math rewards inclusion over precision. Medium
- Operational load: writing 100+ checks/yr requires automation, standardized docs, and back‑office scale. Platforms exist largely to solve this, and FoF structures can proxy diversification if you accept double fees. Otherwise, save your money and invest in a high volume and disciplined early stage fund. AngelList
- Access/selectivity: the best rounds are rationed. Pure “invest‑in‑everything” can face adverse selection or signaling issues. Solutions we see: pre‑agreed small allocations, neutral platforms, and index‑plus‑follow‑on to concentrate in emergent winners. rebelfund.vc
Linking back to public‑market indexing skepticism
Some worry that too much passive capital distorts price discovery in public markets. There’s real literature on passive flows changing co‑movement and “index effects.” The analogy for venture: a tidal wave of indiscriminate seed capital could inflate weak rounds and reduce dispersion—invite active alpha back in. Today’s reality, though, is the opposite: seed remains under‑indexed and selection‑constrained. For LPs, right‑sizing N at seed is closer to adding a missing risk‑management tool than to creating public‑markets‑style distortions. ScienceDirect
Practical design for LPs who want seed “beta”
Objective: approximate early‑stage index IRR with lower left‑tail risk.
Design choices:
- Portfolio size (initial positions): 50–150 per fund if direct; or exposure to ~500+ startups across managers/vintages if programmatic (FoF/platform). This pushes your odds of owning at least one 10× to >95%, and meaningfully increases the chance of a 20–50× outcome over a program’s life. Institutional Investor
- Follow‑on policy: reserve capital to up‑weight winners post‑signal, not to smooth losses. Index at entry; concentrate on traction. (This is how accelerators and high‑volume seed funds preserve upside.) rebelfund.vc
- Fees/admin: prefer platforms that compress per‑deal admin and aggregate K‑1s. FoF approaches trade operational ease for a second fee layer; model net IRR accordingly. AngelList
- Stage discipline: keep this playbook to pre‑seed/seed. At Series A/B, use a smaller N, tighter underwriting, and underwriting to lower base‑rate IRRs. American Century Investments
Appendix: quick math LPs ask for
A. How many companies to capture at least one 10×?
Using Correlation’s p(≥10×) ≈ 3–4%:
- N = 74 (p=4%) → 95% chance of ≥1 ten‑bagger
- N = 99 (p=3%) → 95% chance of ≥1 ten‑bagger
- N = 50 → ~87%; N = 70 → ~94%
B. What IRR corresponds to “8–10× in 10 years”?
It doesn’t. 10‑year compounding is unforgiving: 20% IRR ≈ 6.2×, 25% ≈ 9.3×, 30% ≈ 13.8×. An 8–10× gross outcome typically implies ~24–27% IRR, and lower net after fees/drag. Cambridge Associates
Where this lands
- Seed “beta” looks attractive: mid‑ to high‑teens net IRR across cycles, with higher ceilings in strong vintages. Early‑stage has structurally outperformed later‑stage over multi‑year windows. American Century Investments
- Most seed funds are under‑diversified to capture that beta reliably. Scaling N is the simplest, most evidence‑based way to raise the floor without capping the ceiling—if you pair it with disciplined follow‑ons.
Sources
- Cambridge Associates benchmark materials (US VC Index; methodology; recent horizon IRRs). Cambridge Associates
- Early‑stage vs. all‑VC performance (CA indices summarized by American Century). American Century Investments
- Outcome distributions and power‑law skew (Correlation Ventures updated analyses). Medium
- Portfolio size → performance (AngelList LP cohort analysis; Access Fund). AngelList
- Simulation/portfolio breadth (Institutional Investor analysis advocating ~500‑company exposure). Institutional Investor
- Stage multiple compression by entry timing (AngelList unicorn multiple by stage). AngelList
- Public‑market indexing critiques (index effects, price discovery). ScienceDirect
- Market context on cyclicality (WSJ citing CA; venture vs. publics in recent windows). The Wall Street Journal
- Additional reading (Ignite Insights on entry valuations and portfolio sizing). insights.teamignite.ventures