14th October 2019

Orbis Investments: Man and machine: mapping risk better together

In Above-average companies, below-average prices, we walked through valuation and fundamental measures for the Global Equity Strategy, highlighting a few “clusters” of ideas. We believe that perspective is important. To assess a portfolio’s return prospects, looking at fundamental and valuation metrics is vital. But to assess a portfolio’s risk, another angle is also useful: how our holdings behave.

For this, we turn to our quantitative analysis team, who use proprietary technology and objective research to help us better understand the risks in the portfolios. The “behaviour map” below is just one of many outputs from the team’s analysis and self-developed tools. The map uses a machine learning technique to visualise relationships between stocks, letting us assess their historical behaviour at a glance.

Here’s how to read it. If two stocks appear close together, their prices move together, and if they appear far apart, their prices move less similarly. Don’t worry about whether a stock is high or low on the map, or towards the left or right. The absolute location doesn’t matter. All that matters is a stock’s distance from other stocks. The reddish stocks in the Japan cluster, for instance, broadly hang together, but behave completely differently from the green shares in the South Africa cluster.

Read more

Investments, Investment Commentary