Financial models should never give us a point estimate of stock price

When we use financial models to forecast future cash flows and the stock price of a company, we need to keep in mind that there is always an uncertainty around those values. Each variable that we are forecasting, from sales volume to production cost, actually has a probability distribution behind it, given the uncertainty involved. It is not, therefore, sufficient for us to use our models as a valuation tool to generate a point-estimate stock price; we should be using them as question-asking tools where we can test our hypotheses and understand the drivers behind our price target.

A simple scenario analysis is a good starting point

In scenario analysis, we consider the key drivers behind our investment thesis and we distill the uncertainty behind these key drivers into several likely scenarios, usually a base case, upside case and a downside case. This provides a good starting point for us to analyse the risk/reward for our investment, as we are likely to react differently if one investment has high upside and low downside vs another with low upside and high downside, even if the base scenarios for each investment are identical.

Formulate the uncertainty surrounding key drivers into probability distributions

When we think about probability distributions, the thought of bell-shaped curve comes into our mind. However, not every variable is normally distributed. Probability distributions can be discrete, continuous, symmetric, asymmetric, etc. In addition, key drivers for our investment thesis can be correlated, which will affect their distributions. In this note, we collaborate with our UBS New Analytical Approaches team provide you with a simple flowchart to find the appropriate distribution for the variable that you are forecasting and how to account for correlation.