Research Department

A Well-Diversified Stock Portfolio

May 28, 2014 - 1 comment

In the circles of traditional finance, the widely-held view is that it takes a few stocks - maybe as few as twenty – to build a well-diversified portfolio. At the other end of the spectrum, most market-based investors – such as the professionals at PWL – will assert that diversification requires thousands of stocks. Who has it right? A new academic paper brings light to this debate.

Diversification under the Microscope: The Classic Way

The usual way to measure investment risk is called standard deviation. This statistical measure is used to gauge the variability of short-term (typically monthly) returns. The more variable the returns on an investment, the more uncertain they are. The classic study of diversification starts with a one-stock portfolio and then adds randomly-selected stocks one by one too see at what point, on average, most of the potential reduction in standard deviation is achieved. This is how several academics and practitioners have come to conclude that 15 to 20 are enough to eliminate diversifiable risk.

Serious Flaws in the Classic View of Diversification

Standard deviation became popular in finance in the 1950s. Back then, it was a great new way to measure risk. However, following events such as the 1987 stock market crash, the collapse of the Long-Term Capital hedge fund in 1998 and the great financial crisis of 2008 exposed its main flaw: it grossly underestimates the frequency of capital market crises. In addition, the standard deviation of monthly returns says nothing about the risk regarding cumulative returns for periods of ten, twenty or thirty years. Nonetheless, many market professionals rely solely on it to gauge investment risk: the simplicity and convenience of standard deviation is just irresistible.

An Enlightening New Paper

This is where two finance professors come into play. In their new paper, Francis Tapon (University of Guelph) and Vitali Alexeev (University of Tasmania) studied the benefits of increasing the number of securities in a portfolio of Canadian stocks. They concluded that in order to achieve most of the risk reduction (therein risk is defined as the dispersion of short-term returns, long-term cumulative returns and the losses resulting from severe market crises) from adding securities to a portfolio of Canadian stocks, it requires up to 70 equally-weighted securities.


In my view, the methodology proposed by Tapon and Alexeev is the right way to look at diversification. And if it takes 70 securities to build a solid Canadian equity portfolio, then it stands to reason that a global portfolio would require thousands of stocks to achieve its diversification potential, considering there are about 50 developed and emerging countries in the investable world.

By: Raymond Kerzérho with 1 comments.
  30/05/2014 5:04:27 PM
Don Attwood
Understood. For "crisis" you might also say "covariance." In a panic, nearly everyone takes a bath together.

 Security code