Portfolio Construction

The arithmetic of how a venture fund sizes its bets (number of checks, ownership, and reserves) to survive a power law where one outlier returns the whole fund.

A good business gets passed on by a venture fund for a reason unrelated to the business. A fund is a portfolio built to a return target under a brutally skewed distribution of outcomes, and it isn’t trying to back companies that will merely succeed. It constructs a set of bets in which a few enormous winners pay for everything else and still clear the target. A company that can’t plausibly be one of those winners doesn’t fit, no matter how sound it is.

What It Is

Portfolio construction is how a fund decides to deploy its capital: how many companies to back, how large each check should be, how much ownership to target, and how much to reserve for follow-on rounds. It sits between the individual deal and the fund’s structure: its size, its fee-and-carry economics, its ten-year clock. The structure sets the return the fund owes its limited partners; construction is the plan for getting there.

The plan answers to one empirical fact: venture returns follow a power law, where the extremes are the distribution rather than rare outliers around an average. The fund-of-funds Horsley Bridge found that about 6% of investments produced roughly 60% of the returns across the venture portfolios it backed. Peter Thiel, in Zero to One, reports that Founders Fund’s best investment in a given fund tends to return more than all the others combined, and the second-best more than all the rest after that. So a fund does not optimize for the average bet, dominated by a tail event it cannot predict; it optimizes for access to the tail, underwriting every position to one question: could this one return the entire fund?

That question sets the math. A $100M fund that owes its LPs a 3x gross return must return roughly $300M. If it makes 30 investments and expects one to be the outlier, that position has to return a large fraction of $300M on its own, which takes meaningful ownership of 10–20% at exit. The fund works backward from that target exit, the ownership, and the dilution across later rounds to a check size and a number of companies. Three levers do the work:

• Number of investments. More companies means more chances at the tail but smaller checks and lower ownership; fewer means higher conviction but thinner coverage. Seed funds spread wider (“shots on goal”); concentrated funds make fewer, larger bets.
• Ownership target. The percentage the fund holds at exit. Below a floor, even a huge exit moves the return too little, which is why funds fight for ownership and why “a small piece of a lot of companies” is coherent only for the smallest funds.
• Reserves. Capital held back to follow on into the companies that are working. A common discipline reserves roughly half the fund to defend ownership in the winners through later rounds; a fund that deploys everything in first checks can’t double down on its own breakout.

flowchart TD
    A[Fund size and LP return target] --> B[Power-law outcome distribution]
    B --> C[Underwrite every bet to: could this return the fund?]
    C --> D[Set ownership target at exit]
    D --> E[Work backward to check size and number of bets]
    E --> F[Reserve capital to follow on into the winners]

The construction is the fund’s answer to genuine uncertainty: when you cannot know which bet is the outlier, you build a portfolio whose shape survives being wrong about almost all of it.

Why It Matters

Portfolio construction explains the most disorienting thing investors do, and three readers can act on the explanation.

The founder gets the answer to “why did a good business get rejected?” A company that grows to $40M in revenue and sells for $200M is not missing the merit: for a $500M fund that needs multiple fund-returning outcomes, that exit is a rounding error; for a $30M seed fund, it returns the fund several times over. So pitch the fund whose construction your realistic outcome fits. It also reframes dilution: an investor pushing for ownership is defending the floor below which the position can’t matter, not being greedy.

The aspiring investor (the angel scaling into a syndicate, the operator raising a first micro-fund, the LP evaluating an emerging manager) reads construction as the plan they must build or judge. The common first-fund mistakes are construction errors: too few positions to cover the distribution, ownership too low for any win to register, or no reserves, so the manager watches a breakout raise its Series B and can’t follow. An LP doing diligence on a new manager interrogates exactly this; a thesis without a construction is a hobby.

The talent reader is furthest from this math but still inside its gravity. The investors on a company’s cap table are running a portfolio, and its place there shapes its fate: a marked outlier gets follow-on capital, attention, and patience; a company written down to “the living dead” (alive, but not a fund-returner) gets none of it. An employee weighing an offer is betting, without naming it, on which side of the portfolio it lands on.

How to Recognize It

A fund’s construction is readable from its public behavior, and it predicts how the fund will treat a company:

• Check size against fund size tells you the model. A fund writing $250K checks out of a $200M fund runs a wide, exploratory seed model; one writing $15M checks out of the same fund is concentrated, high-ownership, and fights hard for the few it chooses.
• Ownership demands reveal the floor. A fund that insists on leading and taking 15–20% can’t satisfy its math below that number. One comfortable with 5% and no board seat values coverage over concentration.
• Follow-on behavior signals reserves. A fund that consistently shows up in its companies’ later rounds is defending its winners; one that writes a first check and disappears is out of reserves or running an index-style construction.
• The “fund-returner” question in the room. A partner asking how big this could get, or how you’d deploy $100M, is testing whether the company can occupy the tail the construction depends on. A modest answer tells a venture fund the company isn’t for them.

How It Plays Out

A founder raising a seed round holds two term sheets: one from a $40M seed fund, one from the seed arm of a $900M multi-stage firm. The brand wins. Eighteen months on, the company is on a clear path to a $150M acquisition. But the multi-stage firm builds its portfolio around outcomes an order of magnitude larger, and the partner who led the seed has quietly stopped engaging, because the company will never be a fund-returner at that scale. The smaller fund that lost the deal would have treated the same trajectory as a top outcome. The company didn’t underperform; it was the wrong size of bet for its portfolio.

The reserve lever is just as concrete. Two micro-funds of identical size back the same breakout at seed; the first deployed nearly all its capital in first checks, the second reserved half its fund. When the breakout raises a competitive Series A, the second follows on at its pro rata and protects a stake that, three years later, returns its entire fund at exit. The first, unable to follow, dilutes round after round, and the same exit returns a fraction of its capital. Both picked the winner; only one was built to keep it.

Consequences

Treating a fund as a portfolio with a return target, rather than a collection of bets on good companies, changes how a founder raises and how everyone reads investor behavior.

Benefits. A founder who understands portfolio construction pitches the right-sized funds and reads ownership and dilution demands as structural rather than personal. An aspiring manager sizes positions, ownership, and reserves deliberately instead of by feel. For everyone, the model converts baffling behavior (passing on a sound business, swinging for implausible scale, fighting over a few percent of ownership) into a readable consequence of arithmetic.

Liabilities. The power-law logic, taken as gospel, distorts. It pushes founders to dress modest, durable businesses in fund-returner clothing for investors whose math their company doesn’t fit, when a smaller raise or a non-dilutive path would serve them better. It pushes some investors toward a “swing for the fences or write it off” posture that starves a genuinely good company the moment it stops looking like the outlier. And it is a generalization from historical returns: not every market produces a power-law winner, and concentration without judgment is just a more expensive way to be wrong. The math explains the fund’s incentives, not whether a given partner is a good board member, so a founder still owes themselves the human diligence the model can’t supply.

Related Articles

Sources

• Peter Thiel’s Zero to One (2014) articulates the power-law logic of venture returns for a founder audience, including the observation that a fund’s single best investment tends to return more than all the others combined.
• The Horsley Bridge return data — that a small fraction of investments produces the majority of a venture portfolio’s returns — is among the most-cited empirical demonstrations of the venture power law and underlies the “back the outlier” construction logic.
• The reserve discipline — holding roughly half a fund for follow-on to defend ownership in the winners — reflects the long-standing practice of established early-stage managers and is treated here as recognized field practice rather than the contribution of a single author.
• The reverse-engineering of check size and number of positions from a fund’s return target and ownership floor is standard fund-construction modeling, documented across emerging-manager education and institutional limited-partner guidance.