Venture Math for Legal Tech Angel Investors - Part I

My previous post containing observations for legal tech angel investors met with a lot of interest, and - in particular - I received DMs asking for the explanation about why I cautioned angels against making only two or three investors. This article (I hope) will shed some light on how venture math works.

The thing to remember is that a lot of angel investors in legal tech are lawyers. Lawyers, as a profession, are extraordinarily accomplished at many things. Math, as a rule, is not among them. The real danger is not ignorance, but rather the partial knowledge that comes from familiarity with finance and deal structures, which creates the illusion of understanding venture math. That illusion is the road to perdition.

The 90% problem — and why it doesn't mean what you think

Most lawyers investing in startups know the famous statistic: approximately 90% of startups fail. They apply it the way they would apply any other fact in a legal argument — linearly. Three investments means, in expectation, 2.7 failures.

This is wrong, and a statistics professor I had in college illustrated why with a bet he offered at the start of every semester. To every class he taught - about 50 students - he offered $20 that at least two people in the room shared a birthday. The class almost always took the bet. The class almost always lost.

Here's why. The probability that any specific pair of people shares a birthday is 1/365 — less than 0.3%. Intuitively, the room feels too small for a coincidence. But the number of possible pairs in a room of 50 people is C(50,2) = 1,225. With that many shots at the target, the probability of at least one match exceeds 97%. The professor collected his $20 virtually every time.

The same counterintuitive logic applies to a venture portfolio. The fundamental question is not "will this startup fail?" Slightly oversimplifying (the math is actually a lot more complex), the real question is "will all of my startups fail?" And that probability collapses dramatically as portfolio size grows.

The math is straightforward. If each startup has a 90% chance of failure, then:

P(all fail across N companies) = 0.90^N

For N=3 (a typical angel): 0.90³ = 72.9% chance of total wipeout. For N=20 (a modestly sized fund): 0.90²⁰ = 12.2% — already a different world.

The input sensitivity problem — why this isn't something angels can easily replicate

A sophisticated lawyer reading the above might reasonably object: fine, but I can just make 20 angel investments myself. In principle, yes. In practice, the calculation is acutely sensitive to one variable where individual angels are at a structural disadvantage: the underlying probability of failure for each individual company.

Retail investors in venture - angels writing cheques based on personal relationships, a compelling founder story, or adjacency to the sector - are typically selecting from whatever opportunities come to them. They are, in effect, sampling from the general population of early-stage startups, where the 90% failure rate is roughly accurate.

A VC fund (especially a specialist fund), with structured deal flow and a very different view of the sector, operates from a very different distribution. Selecting more carefully from a curated pipeline, conducting real diligence, and applying a thesis-driven filter can reduce the average probability of failure per company to something like 60%, give or take. That doesn't sound transformative. The math says otherwise.

At a 60% failure rate:

P(all fail across 20 companies) = 0.60²⁰ = 0.000037% — effectively zero.

Compare that to an angel making 5 investments at 90% failure rate:

P(all fail) = 0.90⁵ = 59% — better than a coin flip that an angel investor loses everything.

The difference between a 90% and a 60% individual failure rate — a modest-sounding 30 percentage points — produces an outcome difference that spans several orders of magnitude at the portfolio level. This is the exponential math that experienced VC managers understand and most angels may struggle with.

What this means in practice

None of this is an argument against lawyers having views on legal tech, or against the value of domain expertise. Quite the opposite — as I have written numerous times, domain expertise is one of the most durable edges in early-stage investing. But domain expertise deployed through a couple of angel investments is a bit like knowing the law and choosing to represent yourself: the knowledge is real, the application is sub-optimal.

A specialist fund combines that domain edge with the portfolio construction discipline that the math demands. The combination is more powerful than either alone.

If you are an actual or a prospective angel investor in legal tech and you have any questions about the space, feel free to reach out to me or to Oleg Loukhton.