A Statistical View of the Investment Process in Venture Capital

A VC with a 90% accuracy distinguishing good from bad investments will end up with a portfolio of 30% winners and 70% losers*.

Guillem
Guillem
Jan 2, 2018 · 5 min read

Intuitively, one would expect that a 90% accuracy distinguishing good from bad investments leads to the same % of winners in the portfolio, however, because this accuracy is applied on a dealflow that intrinsically contains a very small number of winners, let’s say 5%, the result is 30% winners/70% losers (*32/68 to be precise). This math became clear to me while reading Pawel Chudzinski’s post “Why you get recruiting wrong in 80% of cases and what to do about it”. That was my aha moment, so first things first, thanks Pawel!

To better understand the math, let’s make a very simple model based on 3 assumptions (Please, note that this is an oversimplified view of the world!):

  • Assumption #1: The future success of a startup is predetermined. The VC acts solely as a startup picker.
  • Assumption #2: There are only two types of investments: Winners (green rocket) and Losers (red rocket)
  • Assumption #3: The unbiased inbound dealflow of a VC has a 5% concentration of winners and 95% of losers.

In this model, the unbiased inbound dealflow of a VC would look like this, with the green (red) rockets representing the winner (loser) investments:

Fig. 1 Schematic diagram of the unbiased inbound dealflow of a VC. The red rockets represent the loser and the green rockets the winner investments.

In this model, the dealflow of a VC can be described using one single parameter, the , which represents the % of winners in the total dealflow. The formula of would then be:

When analyzing and selecting which startups to back, the VC applies a selection process aimed at distinguishing the good from the bad investments, which in our model means distinguishing the winners (green) from the losers (red). The quality of this process can be described with one single parameter, the , which represents how many times in %, the VC is able to successfully distinguish a winner from a loser. The formula of would then be:

Note, that not investing is also a decision! ;-) The result of applying a selection process with Accuracy on the basis of a dealflow with Concentration is what determines the quality of the portfolio of a VC. This process can be schematically described as follows:

Fig. 2 Schematic description of the selection process of a VC that has an accuracy of 70% when analyzing a dealflow with concentration 5%. This creates a Portfolio of 10% winner and 90% loser investments.

If we define the portfolio quality as

Then is a function of the Accuracy and the Concentration such that:

Portfolio quality (P) as a function of the Accuracy A and the Concentration C

In table form:

Table 1. Portfolio quality (P) as a function of the Accuracy A (y-axis) and the Concentration C (x-axis)

Note, that in the process described in Fig. 2 the VC is able to successfully distinguish a good from a bad investment 70% of the times and yet, it builds a Portfolio of 10% winners and 90% losers. If the VC increases its accuracy to 90%, then the Portfolio will consist of 32% winners and 68% losers.

How to improve the odds part 1. Bias the Dealflow

Besides improving the Accuracy, the other lever to pull is, obviously, get better dealflow. However, improving the dealflow cannot just mean to “see more opportunities” because as long as the Concentration stays the same, the outcome will not improve. So what to do, to improve the dealflow Concentration? — Bias it.

Fig. 3 Schematic description of dealflow bias. In this example, the bias improves concentration from 5% to 9%, ultimately improving the Portfolio quality with the same Accuracy.

If a VC is able to bias the dealflow in such a way that the Concentration of winners increases, then its portfolio will likely have a much better quality. By the way, biasing the dealflow is something that many investors consciously or unconsciously already do by specializing on specific sectors, geographies, themes or investment theses.

VCs should be wary of implementing the wrong biases because this can dramatically reduce their odds of finding successful investments. The best way to go about it is to be data driven, i.e. to carefully analyze what are the biases imposed by the investment strategy and to simulate how these will affect the universe of portfolios that can be built.

How to improve the odds part 2. Do the right Outbound

Outbound strategies are a way to generate targeted dealflow that can be pre-selected to have certain characteristics. When the outbound strategies are carefully chosen, the Concentration of winners in the overall dealflow can be dramatically increased, leading to much better portfolios.

Fig. 4 Schematic description of outbound generation. Outbound generation improves the overall Concentration, ultimately improving the Portfolio quality with the same Accuracy.

Just as when biasing the dealflow, the outbound strategies must be carefully chosen using data. In the worst case, it can lead not only to a lower Concentration of winners but also to a larger operational complexity in the organization.

— — — — — — — — — — — — — — — — — — — — — — — — — — — — — — —

About myself: I’m VC @ Nauta Capital. I’d love to blog more often but I only do it when I’m not doing DD or helping our portfolio ;-) email: guillem.sague@nautacapital.com

Guillem

Written by

Guillem

Partner @ Nauta Capital, VC, PhD in Physics, love Tech and Math