The Middle Is a Trap

There is a persistent assumption among people who allocate to venture capital that fund performance improves, more or less monotonically, with the stature of the manager. Better managers raise larger funds; larger funds signal track record; track record predicts returns. The assumption is intuitive and mostly wrong. If you sort venture funds by size and ask where returns are most reliably disappointing, you do not find the answer at the small end, where first-time managers are supposed to flame out, and you do not find it at the very top, among the franchises everyone assumes are overcapitalized. You find it in the middle — in the hundred-to-three-hundred-million-dollar fund, the fund large enough to look institutional and small enough to be trapped by its own balance sheet.

The reason has little to do with the people running these funds, many of whom are excellent. It has to do with when and how a fund acquires its ownership of the companies that matter — and with a structural disadvantage the mid-size fund carries at exactly that task, one that no amount of skill at the task itself can overcome.

Start with the only equation that matters. A venture fund returns capital as the sum, across its portfolio, of each company's exit value multiplied by the fund's ownership of that company at exit. Everything else — the diligence, the board seats, the platform teams, the quarterly letters — is in service of those two numbers. Exit value is largely outside the fund's control; it is set by markets and acquirers years after the check clears. Ownership at exit is the variable a fund can actually engineer. So the entire game, stripped to its skeleton, is the engineering of exit ownership against outcomes you do not get to choose.

And there are really only two moments at which a fund can engineer it. A fund can buy ownership at entry, when a company is cheap and unproven. Or it can defend ownership later, round after round, paying steadily rising prices to keep its position from eroding as new shares are issued. Ownership bought early is cheap and certain. Ownership defended late is expensive and contingent. Almost everything about a fund's fate is decided by which of these two it depends on — and by whether its size lets it depend on that one well.

The small fund depends on the first. It buys eight to fifteen percent of a company at entry, sometimes more, when ownership is cheap, and it does not greatly matter that it may go on to hold thirty or forty names and lean heavily on the power law, because it acquired its ownership before that ownership became expensive. Even ordinary dilution leaves it holding real economics, and a single ordinary exit — a few hundred million dollars, an outcome an order of magnitude more common than a unicorn — can return the fund outright. The small fund is not better at picking companies. It simply bought its ownership at the moment when being right was cheap and certainty was not yet required.

The mega fund can do both, and crucially can afford the second. It takes a meaningful position at entry and then has the balance sheet to defend its winners through the later rounds. It does not defend everything — no one does; even the largest franchises ration reserves and let positions go. But it can defend far more of its winners, and delay the decision of which to defend far longer, than a smaller fund can. A thirty-million-dollar pro rata is one percent of a three-billion-dollar fund. It can carry many positions deep into their lives before committing, and it can lead the later rounds rather than beg into them. It still selects. It simply selects later — closer to the moment a winner is actually identifiable — and from a far larger menu, with the odds dramatically improved.

The mid-size fund is built to depend on the second moment and cannot afford it. And here the experienced allocator objects, rightly: ownership decay is not news; that is what reserves are for. Correct. Take two hundred million dollars — deploy roughly half across twenty companies at five-million-dollar checks for ten to twelve percent each, and hold the other half, a hundred million, in reserve to follow the winners. When one breaks out, you defend it round by round. Hold twelve percent of a winner through its Series B, C, and D and you might spend fifty million dollars of reserve capital on that single name; at a two-billion-dollar exit, twelve percent returns two hundred and forty million — more than the entire fund, from one company. On paper, the mid-size fund has solved the very problem I am accusing it of having.

It has solved it on paper. The portfolio is not paper, and the reserve model fails in practice for three independent reasons, each of which would be sufficient on its own.

The first is finiteness. Reserves are a fixed, lumpy resource, and the cost of defending a winner rises sharply round over round — a few million at the Series B, twelve at the Series C, thirty at the Series D. A single winner, fully defended to exit, can consume fifty million dollars or more — half the entire reserve. A hundred million does not defend twenty companies, or ten, or five. It defends two, perhaps three, and then it is gone. The fund cannot hold its ownership everywhere. It must choose.

The second reason is why that choice is a trap rather than a decision, and it is the heart of the matter. The fund has to commit its reserves before the portfolio has revealed which companies are the winners. At the Series B, when the reserve calls come due, the general partner is looking at two companies that appear to be breaking out and three that might be, and he points his finite capital at the ones that look like winners today. But the eventual breakout is, with uncomfortable frequency, not the company that looked like the breakout at the Series B. If the unicorn were obvious that early, it would be priced that early. The defining feature of the distribution venture lives inside is that the largest outcomes are surprises — which means even a fund that reserves perfectly and spends every dollar will systematically defend some of the wrong companies and under-defend the one that matters most. Ownership still decays. Not because the capital was absent, but because it was aimed at the wrong names. The scarce resource was never money. It was foresight.

The third reason is that the dollars the fund does deploy correctly are worth less than they look. A reserve dollar put into a Series D buys ownership at the Series D valuation; it earns a growth-equity multiple, not a venture one. Only the original entry check earned the fifty-fold return; the fifty million spent defending the position earns something closer to two or three times. The mid-size fund thus spends a quarter of its capital manufacturing a stake whose marginal dollars behave like late-stage crossover checks — growth-equity returns carrying venture-stage risk. And even that assumes the fund can buy what it wants, which in the hot later rounds it frequently cannot: allocation is constrained, the lead and the deepest balance sheets are filled first, and the mid-size fund's pro rata is cut back or honored only at the clearing price.

And that is the reserve model with the paper stripped away: a bet that the fund can name the winner before the winner is knowable. A bet is not a strategy. It is a hope with a management fee.

So the two ends of the size spectrum escape, and the middle does not, for reasons that have nothing to do with the quality of anyone's judgment. The small fund leaned on the cheap moment and never had to defend. The mega fund leaned on the expensive moment and can afford to defend far more, far longer, and far later than the decision is forced. The mid-size fund leaned on the expensive moment with reserves deep enough to compress its own multiple and too shallow to defend its winners — and with no way to know, at the moment the choice is forced, which winners those will turn out to be. Its exit ownership is therefore only partly engineered, and the part left to chance is the part that matters most — its position in whichever company turns out to be the winner. That position is a residual: whatever survives a rationing it could not avoid and a selection it could not reliably get right. The mid-size fund is a price-taker on its own cap table.

This is the trap, and it is worth stating in its general form, because the lesson is larger than any single fund size. At the level of fund architecture — the choices a manager makes before a single investment is selected — venture returns are governed less by skill at picking companies than by when and how the fund is built to acquire its ownership. The structures that win either buy ownership early, when it is cheap, or hold enough capital to defend it later from a position of strength. The structure that does neither — that buys a modest position early and then attempts to defend it with reserves too small to cover its winners and too blind to find them in time — is not unlucky, and it is not badly run. It is positioned against an arithmetic that does not care how good it is.

There is, finally, a purer version of the cheap-entry strategy than any fund can manage, and it is worth naming not as a recommendation but as the logical endpoint of the analysis. A vehicle that holds a single company has no reserve pool to ration and no portfolio across which to misallocate it, so the defense-selection problem — which of my winners do I protect — cannot arise, because there is only one. But this is not a free escape, and the honest form of the argument insists on saying so. The single-asset vehicle does not abolish the selection problem; it relocates it entirely onto the entry decision, where there is no portfolio to bury a bad one and no later round in which to quietly correct it. That is a harder discipline, not an easier one. What it is not is the mid-size fund's discipline — the one that asks an investor to know which bet will pay before the bet has paid.

None of this says mid-size funds cannot succeed. Some do — almost always by behaving against type, buying like a small fund or defending like a large one. But they succeed by escaping the middle, not by occupying it well. The middle is a trap because its central task, choosing which winner to defend, demands the one thing the asset class is built to withhold until it is too late to act on: the knowledge of which bet will pay. Nearly everything else in venture can be learned, hired, or bought. That cannot — and a fund whose survival depends on having it has already lost its argument with the arithmetic.

Sources & Methods. This letter argues from the arithmetic of fund returns rather than from a proprietary dataset; the figures in the text — fund sizes, checks, ownership stakes, exit values — are illustrative, chosen to expose the mechanics at a representative scale rather than to report measured averages. The empirical record the analysis is set against:

■ Fund-return dispersion by size cohort — Cambridge Associates U.S. Venture Capital Index: net-to-LP distributions, loss ratios, and TVPI/DPI trajectories by fund-size cohort (sub-$50M, $100–300M mid-tier, $1B+ vehicles).

■ Ownership decay through successive rounds — PitchBook venture and cap-table data: the erosion of entry ownership from Series A through late-stage rounds and liquidity events.

■ Portfolio breadth and follow-on capacity — Carta State of Private Markets: the relationship between fund size, portfolio diversification, graduation rates, and pro-rata defense capacity across multi-stage financings.