How to Model VC Fund Returns: Portfolio Construction Math

If you can't build a fund model, you can't manage a fund. This isn't optional — it's the core skill that separates fund managers from people who call themselves fund managers.

Most first-time GPs present a portfolio construction slide that says something like "we expect a 3x fund return." That's not a model. A model tells you exactly how many companies you need to invest in, at what ownership, with what reserve allocation, to hit your target — and what happens to returns when your assumptions are wrong.

Let's build one from scratch.

The Power Law: Why Most of Your Returns Come From One or Two Bets

Every venture portfolio is governed by the power law. This isn't a theory — it's an empirical observation backed by decades of data.

In a typical early-stage portfolio:

• The top 1-2 investments return more than the entire fund
• The top 10% of investments return roughly 90% of total value
• The bottom 50% return approximately nothing — some zero, some 1x

This has a direct implication for portfolio construction: your job is to maximize your chance of owning a piece of the top outcome, not to minimize losses across the portfolio.

That's counterintuitive for investors trained in public markets or private equity. In PE, you avoid losers. In VC, you optimize for winners. A portfolio where you avoided all your losers and got a 2x fund is a bad VC fund. A portfolio where you had 8 zeros and one Stripe is a legendary fund.

The math follows from this. You need enough shots that the power-law winner lands in your portfolio. That means diversification — but not infinite diversification. There's a sweet spot.

Building the Portfolio Construction Model

Every fund model starts with five inputs:

1. Fund size — total capital raised
2. Management fee — typically 2% of committed capital for 10 years (though structures vary)
3. Number of investments — your initial check count
4. Average initial check size — the amount you deploy in the first round
5. Reserve ratio — how much you hold back for follow-ons

From those five inputs, everything else flows.

Investable Capital

First, subtract management fees from your fund size to get investable capital.

A $25M fund charging 2%/year for 10 years incurs $5M in management fees — leaving you $20M to invest. That's your budget.

Note: many LPs negotiate step-downs after year 5 (e.g., 2% years 1-5, then 1.5% years 6-10). This increases your investable capital. Model both scenarios.

Deployment Allocation

Split your investable capital between initial investments and reserves.

A typical seed fund structure:

• 50-60% in initial checks
• 40-50% in reserves for follow-ons

If you're deploying $20M with a 50/50 split:

• $10M for initial checks
• $10M for follow-ons

At $500K average initial check: 20 initial investments. At $750K average initial check: ~13 initial investments.

The reserve ratio is one of the most consequential decisions you'll make. Under-reserve and you get diluted in your winners. Over-reserve and you make fewer initial bets and miss out on power-law outcomes.

$25M Seed Fund: Full Worked Example

Let's run a complete model.

Fund Parameters:

• Fund size: $25M
• Management fees: 2%/yr for 5 years (2%), 1.5%/yr years 6-10 — total fees: $3.625M
• Investable capital: ~$21.4M
• Initial capital (55%): $11.75M
• Reserve capital (45%): $9.65M
• Average initial check: $600K
• Number of initial investments: ~19 companies
• Average follow-on per company: ~$500K (into ~half the portfolio, ~10 companies)

Ownership Targets:

• Initial ownership: ~8-10% (at $600K into a $6-7M round)
• Pro-rata target on winners: maintain 6-8% through Series B

The Portfolio Math at Exit:

Assume a 10-year fund life with the following outcome distribution (based on historical seed fund data):

Total portfolio value: ~$96.6M on $21.4M invested = 4.5x MOIC (gross)

After 20% carry and fees: roughly 3.1x net MOIC to LPs.

That's a top-quartile seed fund.

Base / Bull / Bear Scenarios

A real model runs three scenarios with different outcome distributions.

Bear Case (bottom quartile)

• No outlier outcome
• 12 zeros instead of 8
• Top company returns 15x instead of 80x
• Result: ~1.3x net MOIC. LPs get their money back plus a little. You don't raise Fund II.

Base Case

• Distribution as modeled above
• One outlier at 60-100x
• Result: 2.8-3.2x net MOIC. Solid fund. Raises Fund II at 1.3x the size.

Bull Case

• Two outlier outcomes (not uncommon — Sequoia's Fund III had both Google and PayPal)
• Top company returns 150x+
• Result: 5-7x net MOIC. Franchise-maker fund. Fund II closes oversubscribed.

The spread between bear and bull scenarios shows why fund construction matters so much: your ownership percentage at exit is worth exponentially more if you're in the right company.

MOIC vs. IRR: Why You Need Both

MOIC (Multiple on Invested Capital) tells you how much money you returned. A 3x MOIC means $1 in became $3 out.

IRR (Internal Rate of Return) tells you how fast you returned it. A 3x in 5 years is a 25% IRR. A 3x in 10 years is a 12% IRR.

LPs care about both — but for different reasons:

• Endowments and foundations with annual spending requirements care more about IRR
• Family offices with long time horizons often care more about MOIC
• Fund of funds typically target 2.5x+ net MOIC and 20%+ net IRR

The relationship between MOIC and IRR depends on when returns come. A fund that distributes early (because companies IPO'd quickly, like 2019-2021 funds) shows higher IRR at the same MOIC than a fund that distributes late.

Build an IRR model that maps distribution timing to year, not just total return. It changes the story significantly.

The J-Curve: What Your LPs Will See for Years 1-5

Every VC fund shows negative IRR for the first 3-5 years. This is the J-curve. Fees are being paid, initial investments are marked at cost or below, and no distributions have gone out yet.

Your LPs know this intellectually. They still don't love looking at -15% IRR on their quarterly statement.

Prepare them in advance:

• Explain the J-curve mechanics at your first LP meeting
• Show historical data on when well-managed seed funds typically turn positive
• Set expectations that unrealized value will show up in TVPI (Total Value to Paid-In) before it shows up in DPI

DPI (Distributed to Paid-In) is the hardest metric to fake. It measures actual cash returned. Until your fund starts distributing, DPI is zero. That's normal — but LPs will ask about it at year 5 if you haven't started returning capital.

Management Fees + Carry Impact on Net Returns

The gap between gross and net returns is larger than most first-time GPs expect. Walk through the math:

Gross return: $75M on $25M fund = 3x gross MOIC.

Net return calculation:

1. Subtract management fees ($3.625M total) — already reflected in investable capital
2. Carry on profits over the hurdle rate (typically 8% preferred return, though many seed funds waive the hurdle)
3. If no hurdle: carry is 20% of all profits
4. Profits = $75M - $25M LP capital = $50M. 20% carry = $10M to GPs.
5. Net to LPs = $75M - $10M = $65M on $25M invested = 2.6x net MOIC

With a full 8% hurdle and catch-up provision, carry can be structured differently, but the outcome is similar.

The point: a 3x gross fund is a 2.5-2.6x net fund. Model both and present both to LPs. Presenting only gross returns without explaining the net is a credibility killer.

Sensitivity Analysis: What Actually Moves the Needle

Run sensitivity analysis on these five variables to understand what drives your fund outcome:

1. Ownership at exit Model 5% vs. 8% vs. 12% ownership in your top company. The difference on a $2B exit is $40M vs. $64M vs. $96M — at 8x fund size, 10x fund size, and 15x fund size respectively. Pro-rata rights matter enormously.

2. Number of investments 15 vs. 20 vs. 25 bets. More bets = more chances at the outlier but smaller checks = lower initial ownership. There's a real tradeoff. Model it.

3. Reserve ratio A 40/60 split (40% initial, 60% reserves) vs. 60/40. The latter gives you more ownership going in; the former protects you from dilution in winners. Your answer depends on your conviction about which companies you'll choose to double down on.

4. Fund multiple of the outlier What if your best company returns 20x instead of 80x? Run that case. It dramatically changes your net MOIC.

5. Exit timing Companies exiting in year 6 vs. year 10. The IRR difference is significant. Note this especially for LPs who have liquidity concerns.

Building the Model in Practice

Use a spreadsheet. Seriously — don't overcomplicate this with code or specialized software. A well-built Excel or Google Sheets model with:

• An input tab (fund size, fees, check sizes, reserve ratio)
• A portfolio tab (20-25 rows, one per company, with outcome columns)
• A returns tab (aggregated MOIC, IRR, DPI over time by year)
• A scenario tab (bear/base/bull with different outcome distributions)

...will serve you better than any tool that obscures the underlying assumptions.

The point of the model is to understand the inputs, not to get a specific output. When an LP asks "how do you think about reserve allocation?" — you should be able to answer from intuition built by running the model 50 times.

That fluency is what separates GPs who are in control of their portfolio from GPs who are managing reactively.