Emerging Manager Portfolio Construction: Building a Fund I That Works

Emerging manager portfolio construction is the set of decisions that determine what your fund actually is: how many companies you back, at what check size, at what ownership, with what reserves, over what period. Get these right and a modest fund can return multiples. Get them wrong and no amount of good picking saves you, because the math was broken before the first check went out. This guide covers the decisions in order, with the numbers that matter for a first fund.

The perspective behind it: VC Lab has accelerated 950+ VC firms across 20 cohorts, 85% of them investing at pre-seed or seed, and their portfolios have produced 390+ up rounds. We've seen a lot of Fund I models, and the ones that fail usually fail the same few ways.

Why portfolio construction decides the fund before picking does

Venture returns follow a power law: in a typical portfolio, one or two companies produce most of the return, a handful return something, and a large fraction return nothing. Portfolio construction is how you position the fund to catch the outlier and own enough of it to matter.

That's why LPs read the construction model before they read the deal stories. A manager who says "25 to 30 checks at $250K targeting 8% ownership, 30% reserves, deployed over three years" has described a machine an LP can stress test. A manager who says "great deals from my network" has described a hope. In our experience across LP conversations, the construction model is the second question after the thesis, and it's the question where first-time managers most often lose the room.

The five decisions, in order

1. Fund size, chosen from the market backward

Fund size is a portfolio construction decision, not an ambition. Our first fund fundraising statistics show roughly 90% of emerging manager commitments go to funds under $15MM, with the average LP check at $159K. A $10MM to $12MM Fund I is not the consolation prize; it's what the market funds, and the rest of the model should be built from it.

2. Number of positions

The power law argues for enough shots on goal that catching an outlier is probable. Most credible Fund I models land between 20 and 35 core positions. Below 20, the fund is closer to a concentrated bet on your picking than on the asset class. Above 40, checks get too small to buy meaningful ownership and your attention fragments past usefulness, especially for the 61% of new managers running solo.

3. Check size and ownership

Work it as arithmetic. A $12MM fund reserving 30% has roughly $8.4MM for initial checks. Thirty positions means ~$280K initial checks. At typical pre-seed rounds, that buys low single digit ownership as a collaborative check, or more if you lead smaller rounds. The question LPs will ask: at your entry ownership, what does one big winner return to the fund? If a fund-returner outcome at your ownership doesn't return the fund at least once, the model needs different checks, different rounds or a different fund size.

4. Reserves

First funds usually hold 20% to 40% for follow-ons. The honest tension: reserves protect ownership in winners, but at pre-seed your information advantage between rounds is small, and a Fund I that over-reserves ends up returning capital it never deployed. Many strong small-fund models run lean reserves and say so plainly, using the fund's size as the discipline. Whatever you choose, write down the rule for deploying reserves before the first follow-on decision, because deciding case by case under enthusiasm is how reserves evaporate into bridge rounds.

5. Pacing

Deploy initial checks over roughly three years. Faster concentrates you in one vintage's prices; slower starves the back half of the fund's life. Pacing is also the honest answer to a common LP worry, that a first-time manager will spray the fund in year one. A written pacing plan, tracked in your fund's own numbers, answers it.

The model is a promise: keep it checkable

A construction model only builds trust if the fund's operations can show performance against it. That means tracking deployed capital, ownership, reserves and pacing in real time, not reconstructing them for the annual letter. This is operational table stakes now: 1,000+ firms run on Decile Hub, where pipeline, checks and portfolio data live in one system, and Decile Partners keeps the capital accounts that make "we're on plan" a verifiable sentence instead of a claim. A first-time fund that reports against its own model every quarter is doing the single most trust-building thing available to it between closes.

How construction changes across fund situations

The pre-Fund I vehicle

If you're building evidence before a full fund, construction still applies at miniature scale: a Start Fund with a handful of positions is a track record generator, and the discipline of stating checks, ownership and pacing on a small vehicle is precisely what makes the eventual Fund I model credible.

Fund I proper

Everything above, sized to what closes. Our emerging manager performance data, from a sample of 1,000+ PACTs, 1,000+ LPAs and 900+ funds, showed February through May 2026 among the top five fundraising months in four years, each at 1.2x to 2.2x the prior year's month. The market is funding first vehicles; it's funding the ones whose models hold together.

Fund I to Fund II

Construction questions change shape at Fund II: LPs interrogate how Fund I's model performed against plan, and what changes. We've covered that transition specifically in portfolio construction from Fund I to Fund II, and the older mechanics of building the underlying spreadsheet live in how to build a VC fund model. This page is the strategy layer above both.

A worked model: $12MM, end to end

Numbers make the discipline concrete, so here's the full arithmetic on a realistic Fund I.

The frame. $12MM fund, inside the band where roughly 90% of emerging manager commitments land. Management fees at 2% over ten years with a step-down consume roughly $2MM, leaving about $10MM investable. Reserve 30%, roughly $3MM, for follow-ons. That leaves $7MM for initial checks.

The positions. Twenty-eight initial checks at $250K each. At current pre-seed entry prices, a $250K collaborative check typically buys low single digit ownership; call it 3% to 5% depending on round size and whether you lead. Deployed over three years, that's roughly one new position every five to six weeks, a pace a solo GP can actually diligence.

The return test. Now the question every LP will run: what has to happen for this fund to return 3x, or $36MM? At 4% average entry ownership diluted to roughly 2% at exit, a single $500MM outcome returns about $10MM, most of the fund's capital back from one company. Two such outcomes plus a modest middle, a few 5x to 10x positions among the 28, clears 3x. That's the power law doing what it does: the model doesn't need ten winners, it needs the position count and ownership to make one or two catchable.

The reserve rule, written in advance. The $3MM follows on only into positions marked up by a lead we didn't need to convince, at a maximum of $500K per company, top six positions only. Any reserve unspent by year five recycles or returns. One paragraph, decided while calm.

The sanity checks. The raise this model needs is plausible: at a $159K average check, $12MM is roughly 75 commitments, or fewer with a couple of anchors, which matches how funds in our cohorts actually close. The ownership assumptions survive contact with real round sizes. And the whole thing fits on one slide, which is itself a signal: models that need ten slides are usually hiding an assumption that doesn't arithmetic.

Change any input and the others move: a $20MM version of this fund needs either bigger checks, more ownership, or a bigger outlier to clear the same multiple. That's the entire discipline, made visible.

The failure modes we actually see

The recurring ones, from hundreds of Fund I models reviewed across our cohorts:

The fund-size mismatch. A $25MM target with a network that writes $159K checks. The model was fine; the fundraise it depended on wasn't real. Size to the market data, not the deck.

The ownership fiction. Models assuming 10% ownership from $200K collaborative checks at today's entry prices. LPs catch it in minutes, and it poisons the rest of the diligence.

Reserve drift. Reserves deployed into defensive bridges for the middle of the portfolio instead of doubling into the top. The fix is the pre-written rule, enforced by someone other than your enthusiasm.

The unfollowable pace. Fifteen checks in year one, then a two-year quiet stretch that LPs read as either lost conviction or lost discipline. Pacing plans exist to be boring.

The invisible model. A fine plan that lives in a spreadsheet nobody updates, so no one, including the GP, knows whether the fund is on it. If the model isn't tracked in the fund's operating system, it isn't really the model.

Recycling, fees and the fine print that changes the math

Two mechanics quietly move every number above, and first-time models routinely forget both.

Fees compress investable capital. A 2% management fee over a ten-year life, even with the customary step-down after the investment period, consumes something in the range of 15% to 20% of committed capital. A $12MM fund is really a roughly $10MM investing machine. Models that deploy 100% of commitments are advertising that nobody checked the arithmetic, and it's the first thing a fund of funds analyst recomputes.

Recycling gives some of it back. Most LPAs permit reinvesting early exit proceeds within limits, which can push invested capital back toward, or even past, committed capital. A model that states its recycling assumption, even if the assumption is zero, signals the manager has read their own LPA. The provision itself is settled in the fund documents during formation, which is one more reason standard, well-understood documents matter: the construction model and the legal machinery have to describe the same fund.

Neither mechanic changes the strategy. Both change the numbers on the slide, and LPs notice which managers noticed.

Frequently asked questions

How many investments should a first-time VC fund make?

Most defensible Fund I models hold 20 to 35 core positions. Fewer than 20 concentrates outcome risk beyond what the power law forgives at pre-seed; more than 40 dilutes checks and attention past the point where ownership or support matter, particularly for solo GPs, who are now 61% of new managers.

What percentage should a Fund I hold in reserves?

The common band is 20% to 40%, and lean is increasingly defensible at pre-seed, where information advantage between rounds is small. What matters more than the number is a written deployment rule set before the first follow-on decision, and reporting reserves against it.

What ownership should an emerging manager target?

Whatever entry ownership makes one outlier outcome return the fund at least once, at your real check sizes and real entry prices. For collaborative pre-seed checks that's often low single digits, which the model must survive honestly rather than assume away.

How fast should a first fund deploy?

Roughly three years for initial checks is the standard the market trusts. Materially faster buys one vintage's pricing with the whole fund; materially slower suggests the pipeline isn't real. State the pace and report against it.

What do LPs look for in a portfolio construction model?

Internal consistency, mostly: fund size, check count, check size, ownership, reserves and pacing that arithmetic together, sized to a raise that's plausible for your network, with tracking that will show plan versus actual. A model like that says the manager understands the machine they're operating, which is most of what a first-fund diligence can establish.

Where to go from here

Portfolio construction is where a first fund is won or lost before the first deal memo. If you want the model pressure-tested alongside the thesis, the documents and the LP strategy, VC Lab runs a free 14-week accelerator for new and emerging managers, with fund formation built in. The data behind this guide is at the VC Research hub.