How many startups should an angel back? The maths of finding a winner

Most angels start with a handful of cheques into companies they know well. That is a sensible way to learn. It is also, on the maths alone, a portfolio that will most likely not contain the one company that makes the whole effort worthwhile.

This article sets out that maths: why one company tends to decide the result, how the chance of backing it grows with each cheque, and what a portfolio large enough to give you a fair chance costs.

One company usually decides the result

Early-stage returns are uneven. Many companies return nothing or less than they raised. Some return a modest multiple. A very few return many times the money invested, and those few usually account for most of the portfolio's gain.

That shape changes the question. It is not "how good is my average company?" It is "how likely am I to hold at least one of the few companies that matter?"

The maths of backing at least one winner

Call a large winner a company that returns, say, 20 times your cheque. Suppose each company you back has a chance p of becoming one, and that the outcomes are independent. The chance that none of n companies is a winner is (1 − p) multiplied by itself n times. So the chance of at least one winner is:

At an illustrative 5% chance per company:

Your assumption matters more than the formula

Nobody knows the true chance that a given startup becomes a large winner. It depends on stage, sector, geography, the market for the next rounds, and how strictly you define a winner. The 5% above is an illustration, not a measured rate.

So run the plan at more than one rate:

If your plan only works at the optimistic rate, it is a hope, not a plan.

Why the formula flatters a concentrated portfolio

The formula treats each company as an independent draw. Real outcomes are linked. Companies in the same sector, or funded in the same year, tend to do well or badly together: a funding market that tightens hurts all of them at once.

That makes concentration riskier than the table shows. Spreading cheques across sectors and across several years of vintages gets you closer to the independent draws the formula assumes.

What a portfolio of that size costs

The count is half the plan. The other half is the money. A portfolio needs first cheques, reserves for follow-ons in the companies that earn them, and time to deploy both.

A useful check on the economics: one 20x outcome on a single $25,000 cheque returns $500,000. Across 25 equal first cheques, that repays 80% of the money invested on its own. The rest of the result has to come from other companies, or from having put follow-on money into the winner. That is what reserves are for.

Building towards the number

• Pick a cheque size you can repeat. Twenty-five cheques you can afford is a better plan than five you cannot.
• Deploy over several years. Vintages spread the funding-market risk.
• Keep reserves for evidence. Follow on where the company has earned it, not everywhere by default.
• Track every position. The formula only helps if you know how many companies you really hold, at what cost, and how each is doing.

The portfolio size calculator runs this on your own assumptions: the chance per company, the confidence you want, your cheque size, reserves and years.

This article is general information, not investment, tax or legal advice. Early-stage investing is high risk; you could lose all the money you invest.