Monte Carlo simulation of a venture portfolio

Portfolios of venture-based companies are inherently fraught with risk. As part of a project to build a strategy for managing a portfolio of high-risk biotech ventures, I created a Monte Carlo simulation of the portfolio. My simulation accepted basic assumptions about the expected life-span of ventures, their chances of success, and their expected costs. Based on a desired number of successful ventures over 10 years, we used the simulation to estimate:

• The number of deals the portfolio managers would need to do
• The approximate number of ventures they would be managing at a given time
• The expected amount of investment needed
• The approximate headcount in the portfolio at a given time
• My simulation helped to inform the portfolio managers in creating their strategy and setting expectations around the output of the portfolio.

The whole simulation is run from model.py. You can change assumptions in that file. It generates a Looper object, which simulates thousands of Accelerator objects. Accelerators contain Portfolios, which keep track of projects in different stages (i.e. screen, evaluate, develop, accelerate) of a venture portfolio. Accelerators loop over Years, and each year projects grow, succeeding or failing. Each year and project has an expected budget and headcount.