Venture Capital: The Maths Only Works With Outliers

Venture capital is often described as investing in startups, which is accurate and unhelpful. Plenty of investors put money into new companies without behaving anything like a venture fund.

What defines the model is the distribution of outcomes it operates under, and understanding that single fact explains almost every behaviour in the industry that looks strange from outside.

Returns follow a power law. Most investments return little or nothing. A handful return something modest. And the performance of an entire fund typically depends on one or two investments that return many times the whole amount invested.

That is not a description of bad luck or poor selection. It is the shape of the outcome distribution, and every practice in the field is a rational response to it.

What the maths forces

Once returns are dominated by rare enormous outcomes, several things follow that seem irrational under any other framing.

Missing a winner costs more than backing a loser. A failed investment loses one unit. Declining to invest in something that becomes an outlier costs many multiples of the entire fund. This asymmetry makes investors far more afraid of passing on the right company than of backing the wrong one — the opposite of most investing.

Modest success is close to failure. A company returning two or three times the money is a good business and does nothing for a fund whose maths depends on outliers. This is the source of most tension between founders and investors, and it is structural rather than personal.

Diversification works differently. Since the winners are unpredictable and few, funds need enough positions to have a reasonable chance of holding one. Concentration in a small number of carefully chosen bets is a worse strategy here than it would be in ordinary investing.

Pattern-matching becomes standard practice. With little hard information at the earliest stages, investors lean on resemblance to past successes. This works badly and there is not much better available — and it is a textbook route to survivorship bias, since the resemblance is to winners whose non-winning lookalikes are invisible.

Follow-on funding is the real decision. Deciding which existing investments to back further, with much better information, matters more to fund performance than the initial picks.

Why founders and investors talk past each other

The most common source of friction is not dishonesty on either side. It is a genuine difference in objective that both parties often fail to make explicit.

A founder generally wants a company that succeeds, provides a good living, and does not fail. A fund needs a portfolio containing outliers, and is comparatively indifferent between a company that fails and one that becomes solidly profitable at moderate scale — because neither changes the fund's outcome much.

That indifference is uncomfortable to state plainly and it is real. It means that when a company is doing reasonably but not spectacularly, the investor's rational preference is often to take more risk in pursuit of a much larger outcome, while the founder's rational preference is to consolidate. Both are correct given their positions.

This is a straightforward principal-agent divergence, and it is far better understood before signing than discovered afterwards.

The practical consequence is that venture money suits a specific shape of business: one that needs capital before revenue and could plausibly become very large. Where both are true, it is often the only realistic option. Where either is false, the mismatch surfaces later as pressure toward risks the business never needed to take.

Plenty of excellent businesses do not have that shape, and funding them from revenue keeps the decisions with the person living with the consequences.

What the industry is genuinely good and bad at

Good at: funding things with no collateral and no revenue, where a bank cannot lend and the failure rate is high. This is a real gap in the financial system, and filling it has funded a great deal that would not otherwise exist.

Good at: tolerating failure. An industry that expects most investments to fail does not punish founders for a failed company the way most institutions would, which lowers the personal cost of attempting something uncertain.

Bad at: evaluating things without precedent. Pattern-matching against past successes systematically underweights the genuinely unfamiliar, which is awkward for an industry that describes itself as funding the new.

Bad at: the middle. Businesses that will be profitable but not enormous are poorly served, and the funding landscape between a bank loan and venture capital remains thin in most places.

Contested: whether the model produces good aggregate outcomes. It has funded significant useful technology, and it also directs capital toward whatever currently looks like the pattern, which produces crowding, correlated bets, and periodic overvaluation. Both readings are supported by the record.

The honest summary is that venture capital is a specialised instrument that works well for a narrow category and is applied far more widely than that category extends — largely because it is the most visible form of startup funding, not because it is usually the most appropriate.