Module 04 established the power law as a fact about venture returns: a handful of investments produce nearly all the gains. This module is about what a fund manager actually does with that fact. How many investments should a fund make? How much should it reserve for follow-ons? Should it spread bets widely or concentrate on high conviction? The answers are not matters of taste — they follow from the mathematics of the power law, and getting them wrong is one of the most common ways venture funds fail.
If a venture investor could reliably identify the one company in a hundred that becomes a fund-returner, they would invest only in that one company and skip the other ninety-nine. They cannot. No one can. The single most important fact about early-stage investing is that the winners are not identifiable in advance — not by the best investors, not by the founders themselves, not by anyone. The power law describes the outcome distribution, but it gives no method for predicting which specific company will land in the tail.
This is why portfolios exist. Since you cannot pick the one winner, you must make enough bets that you are statistically likely to capture a winner when one occurs. The portfolio is not a hedge against risk in the ordinary sense — it's a mechanism for ensuring exposure to an outcome distribution where the rare event drives everything.
You know that roughly 1 in 100 early-stage companies will become a fund-returner (a 50×+ outcome). You cannot identify which one in advance. Therefore you must construct a portfolio large enough and structured well enough that you are likely to hold the winner when it appears — and large enough ownership in it that the win actually returns your fund. Portfolio construction is the discipline of making those two things true simultaneously.
Everything in this module follows from that problem. The portfolio must be big enough to capture a rare winner (which argues for many investments), but each position must be big enough that the winner actually moves the fund (which argues for fewer, larger investments). These two pressures pull in opposite directions, and resolving the tension — for a specific fund size, stage, and strategy — is the essence of portfolio construction.
Why do venture funds typically make 20-40 investments rather than 5 or 100? The answer comes from the probability arithmetic of capturing a rare winner.
Suppose roughly 1 in 50 early-stage investments becomes a fund-returner — call it a 2% hit rate (the rate varies by stage and strategy; this is illustrative). If a fund makes only a handful of investments, the chance of capturing even one fund-returner is low:
| Investments in portfolio | Chance of zero fund-returners | Chance of at least one |
|---|---|---|
| 5 investments | 90.4% | 9.6% |
| 10 investments | 81.7% | 18.3% |
| 25 investments | 60.3% | 39.7% |
| 40 investments | 44.6% | 55.4% |
| 100 investments | 13.3% | 86.7% |
(These are computed as the binomial probability of zero successes given a 2% per-investment hit rate: chance of zero = 0.98n.) The pattern is stark. A 5-investment fund has only a ~10% chance of capturing even one fund-returner — meaning a 90% chance of failure regardless of how good the individual picks are. A 40-investment fund crosses the 50% line. A 100-investment fund is almost certain to capture at least one.
So why don't funds just make 100+ investments to maximize the chance of catching a winner? Because of the other pressure: ownership. The more investments a fund makes from a fixed pool of capital, the smaller each check, and the smaller the fund's ownership in any single company — including the eventual winner. Section 04 works through this, but the intuition is: a 100-investment seed fund might own only 2-3% of each company, so even capturing a fund-returner produces a smaller return on the fund than a more concentrated fund would have gotten from the same winner.
More investments → higher probability of capturing a winner, but smaller ownership in each (so each winner returns less to the fund).
Fewer investments → larger ownership in each (so each winner returns more), but lower probability of capturing one at all.
The optimal portfolio size balances these. For most early-stage funds, the balance lands at roughly 20-40 initial investments, though seed funds run larger (more, smaller bets) and concentrated Series A funds run smaller (fewer, larger bets).
One of the most important disciplines in venture investing follows directly from the power law: every initial investment a fund makes should be capable, on its own, of returning the entire fund. This is the "fund-returner test," and it's a more demanding filter than it first appears.
The logic: since most investments will fail or return little, the fund's overall return depends on its few winners being big enough to carry the rest. If no single investment in the portfolio could plausibly return the whole fund, then even the fund's best outcome won't generate venture-level returns. So a disciplined GP asks, before every initial investment: "If this company becomes everything we hope, could this single investment return our entire fund?" If the answer is no, the investment fails the test, regardless of how attractive it otherwise looks.
For a $100M fund that will own 10% of a company at exit, the company must be able to exit at $100M ÷ 0.10 = $1B for that single investment to return the fund. If the company's realistic ceiling is a $200M acquisition, the investment can return at most $20M — a fifth of the fund — and therefore fails the test. The investment might still make money, but it cannot be a fund-returner, and a portfolio built entirely of such investments cannot produce venture returns.
The test explains several otherwise-puzzling venture behaviors:
The fund-returner test is the single most useful lens for understanding venture investment decisions. When a VC's behavior seems irrational from a founder's perspective — passing on a profitable company, pushing for a riskier bigger outcome, insisting on a certain ownership level — it usually makes sense once you apply the test.
A common mistake of first-time fund managers is to deploy all their capital into initial checks, leaving nothing for follow-on investments. Experienced funds reserve a substantial fraction of their capital — often 40-60% — to invest again in their best-performing portfolio companies in later rounds. The reserve strategy is one of the highest-leverage decisions in portfolio construction.
Recall the power law: most investments fail, a few succeed enormously. The problem is that you don't know which is which at the time of the initial check — but you learn over the following 1-3 years. The companies that are clearly working start to separate from those that aren't. Reserves let a fund double down on the apparent winners once the signal is clearer, concentrating more capital into the companies most likely to be fund-returners.
Initial checks are made under maximum uncertainty — you're spreading bets because you can't tell winners from losers yet. Follow-on checks are made under reduced uncertainty — the winners have started to reveal themselves. By reserving capital for follow-ons, a fund effectively gets to invest more in its winners after gaining information, partially solving the "can't pick the winner in advance" problem. A fund that reserves 50% for follow-ons is saying: "We'll make our initial bets broadly, then concentrate our remaining capital into whichever bets are working."
Reserves also serve a defensive purpose: maintaining ownership. Recall from Module 07 that every new round dilutes existing holders. A fund that wants to maintain its ownership percentage in a winning company must exercise its pro-rata rights (Module 05) and invest in subsequent rounds. Without reserves, the fund's ownership in its winner erodes round after round — exactly the wrong outcome, since the winner is where the fund's returns come from.
The reserve decision is genuinely hard because it requires the fund to predict, mid-life, which of its companies deserve follow-on capital — and to resist the temptation to "average down" into struggling companies (throwing good money after bad) or to over-reserve and under-deploy. The best funds are disciplined about following their winners and ruthless about not following their losers.
One of the genuine strategic debates in venture is how concentrated a portfolio should be. The probability math of Section 02 argues for many bets; the ownership math argues for few. Different respected firms have landed in very different places on this spectrum, and both extremes have produced great returns. The debate is real, not settled.
The two strategies are coherent in different ways. The concentration strategy bets that the firm can pick winners better than chance — that its judgment, diligence, and involvement justify making fewer, bigger bets. If the firm really does have superior judgment, concentration amplifies it. The diversification strategy bets that winners can't be reliably picked, so the best approach is broad exposure plus the ability to follow on into whoever emerges. If picking really is mostly luck, diversification captures the upside without requiring a skill that may not exist.
The empirical evidence is genuinely mixed, which is why the debate persists. Benchmark's concentrated model produced some of the best fund returns in history (the famous $6.7M investment in eBay that returned $5B; the early Uber stake). Y Combinator's hyper-diversified model also produced extraordinary returns (funding Airbnb, Stripe, Coinbase, Doordash, and dozens of other major outcomes across thousands of small bets). Both work. The question for any given fund is which model fits its actual capabilities — and the most common error is a fund believing it has winner-picking skill (justifying concentration) when it actually doesn't (and should diversify).
Two numbers drive most portfolio-construction decisions: the expected loss ratio (what fraction of investments return less than the capital invested) and the target ownership (what percentage of each company the fund aims to hold). These two numbers, combined with fund size, largely determine how a fund must be constructed.
The loss ratio is the fraction of a fund's investments that lose money or return roughly nothing. For early-stage venture it's high — often 50-70% of investments return less than 1×. The loss ratio directly determines how much the winners must return to compensate. If 60% of investments return zero and 30% return roughly capital, then the remaining 10% must generate the fund's entire return — they have to be enormous to carry the dead weight of the rest.
The ownership target follows from the fund-returner test (Section 03). A fund needs enough ownership in its winners that capturing one returns the fund. Working backward: for a winner to return a fund at a realistic exit valuation, the fund typically needs to own somewhere in the range of 10-20% of the company by exit (for early-stage funds). This ownership target then drives the check size, which combined with fund size drives the number of investments.
Fund size + ownership target → check size (initial + reserves). Fund size ÷ check size → number of investments. Number of investments + hit rate → probability of capturing a winner. The whole construction is a chain: you can't independently choose fund size, ownership target, and number of investments — fixing any two determines the third. Portfolio construction is largely the discipline of making this chain internally consistent.
This chaining explains why fund size is such a consequential decision. A fund that raises too much capital relative to its strategy is forced into either too many investments (diluting the winner's contribution) or too-large checks (chasing fewer, more expensive deals). A fund that raises too little can't maintain ownership through follow-on rounds. The "right" fund size is the one whose construction chain is internally consistent for the firm's stage, strategy, and capabilities — which is why experienced LPs are skeptical of funds that grow their size dramatically between vintages without a corresponding change in strategy.
Portfolio construction looks very different across the venture landscape. A $30M seed fund, a $500M multi-stage fund, and a 500-company-per-year accelerator are all "venture" but construct their portfolios according to completely different math.
| Fund type | Investments | Ownership target | Reserve strategy |
|---|---|---|---|
| Micro seed fund ($30M) | 30-50 | 5-10% | Light reserves; rely on initial checks |
| Classic Series A fund ($200M) | 20-30 | 15-20% | Heavy reserves (50%+) for follow-on |
| Multi-stage fund ($1B+) | 30-50 | varies by stage | Follow winners across stages aggressively |
| Accelerator (YC-style) | 100s-1000s | ~7% | Minimal initial; selective follow-on funds |
| Growth fund ($2B+) | 15-25 | 5-15% | Large checks, lower loss ratio, fewer bets |
A few patterns worth noticing:
Portfolio construction norms vary across ecosystems. European funds have historically been more concentrated (fewer bets, higher conviction), partly reflecting smaller fund sizes and a more conservative LP base. Indian and Southeast Asian funds often construct around a smaller number of category leaders per sector, reflecting markets where a few winners dominate each category. Chinese funds in their growth era ran unusually concentrated and large, fueled by abundant local capital and corporate-VC participation. The construction math is universal, but the inputs — fund size, hit rates, exit-value ceilings — differ enough by market that the resulting portfolios look quite different.
To bring the pieces together, construct a portfolio model for a hypothetical $100M seed fund from scratch. This is the kind of model a GP builds before raising a fund, to show LPs how the fund is expected to behave.
Project the 28 investments through a realistic power-law distribution. Assume the fund follows on into its winners, maintaining roughly 10% ownership in the best outcomes:
| Outcome | Companies | Avg exit value | Fund's ~10% stake | Returns |
|---|---|---|---|---|
| Total loss | 15 | $0 | — | $0 |
| Acqui-hire / small | 7 | $15M | 10% | $10.5M |
| Solid exit | 4 | $150M | 10% | $60M |
| Strong exit | 1 | $600M | 10% | $60M |
| Fund-returner | 1 | $2B | 10% | $200M |
| Total | 28 | — | — | $330.5M |
Portfolio construction is the discipline of making enough bets to be likely to capture a power-law winner (Section 02), while holding enough ownership in each (Section 06) that the winner actually returns the fund (Section 03), and reserving enough capital (Section 04) to defend that ownership through the winner's later rounds. Every number in the model — fund size, check size, investment count, reserve ratio, ownership target — is chained to every other, and the whole construction succeeds or fails on whether it captures and holds a fund-returner.
Six questions on portfolio construction, the fund-returner test, reserve strategy, and the concentration-vs-diversification debate.