Volatility Is A Poor Measure Of Risk

by Peter Pearce

This is part of a series written to explore the myths in popular investing as exposed in Michael Dever’s new book, “Jackass Investing.” In the book, Dever uses experience from three decades of hedge fund management to explore how the conventional wisdom in investing and portfolio management preaches little more than gambling. For an introduction to the series and the book, see our previous article looking at the return drivers for stocks.

Today’s article will examine the problems with using volatility to measure risk in Myth 9: Risk can be Measured Statistically. My goal is to highlight how irrational it is to gauge the riskiness of a trading strategy based on superficial statistical measures without a more fundamental understanding of the market. As has been the theme throughout this series, it’s essential to base investment decisions on analysis of the return drivers and baseline conditions.

To illustrate more clearly, let’s start with an unconventional example Dever provides in the book. Suppose I offered you the opportunity to invest in a mystery market where the market followed the pattern in the graph below, and supposed the 2009 price pattern is representative of the historical pattern for this market.

Click to enlarge

Our mystery market performance is:

Would you invest in this market? The annualized volatility of 32% is three times higher than the annualized return and about twice as high as the volatility of the S&P 500. With only a superficial statistical look at the numbers, this market appears to be very risky, and most investors wouldn’t consider investing here.

Because I understand what the return driver is that is causing the “price” to move, I’d gladly trade on it each year, buying on Feb 1st and selling on Sept 1st. Why? Because this chart shows the daily high temperatures for 2009 in Fahrenheit for Philadelphia, Pennsylvania. Betting on the temperature rising in Philadelphia from the end of winter to the end of the summer is as sure of a bet as you could ever make. Unfortunately, you’ll have a hard time getting anyone to take the other side of this trade.

Now that you understand what is driving the change in “price”, you can see that the volatility exhibited in the chart above doesn’t really represent risk at all. In fact, the “risk” here actually decreases as the temperature falls, because it’s more likely to increase in the next period. The point here is that many investors take this exact approach when evaluating the risk of a position. They will examine some statistical volatility measures and form a conclusion about the riskiness without really understanding the risk of the position at all.

Trading temperature is an example of a strategy with high short-term volatility but negligible long term risk of loss. There are also strategies that exhibit low volatility but a high risk of loss in the long term.

The volatility of returns is a good indicator of risk when the returns are random and the environment is static. Static market environments are certainly not the case in today’s market. As the baseline conditions change (e.g. Italy defaults), the historic volatility measures you originally based your investment decision on could be thrown out the window. This is why financial institutions calculate risk measures using hundreds of hypothetical scenarios; one for example could be the break-up of the European Union.

The “TED” spread is the difference between the interest rates on Treasury Bills and LIBOR (the average rate banks in London charge each other). The T-Bill is considered a risk-free investment while bank loans are not, and therefore the TED spread reflects the risk that the banks won’t repay their loans. In “normal” times, the TED spread hovers between 0.1% and 0.7%, and looks like this:

It’s possible to trade the TED spread by buying T-Bills and shorting LIBOR contracts when the spread is low, expecting it to revert up to the mean. When the spread is at the high end, you would then short T-Bills and buy LIBOR contracts, expecting the spread to revert back downwards.

Any trader who undertook this trading strategy between 2002 and 2006 would have yielded impressive returns, while statistical analysis would have indicated little risk. Seems like a great trade, right? But without fully understanding the baseline conditions necessary for this strategy to work, the statistical measures of risk are worthless. After years of bouncing around this range, the sub prime mortgage crisis hit in August 2007 and the spread blew out of the range to 2%, and by 2008 it had reached 4.5%.

The trader who had traded the high end of the range in August 2007-- expecting it to revert back to the mean, as it always had-- would have been absolutely devastated. Even the most sophisticated statistical analysis of the TED spread couldn’t have predicted that it would blow out of the range by more than 4%. The statistics that indicated a low risk in this strategy are an illusion.

I’m not looking to completely dismiss statistical measures of risk, but instead, advise you that they must be tempered with evaluations of baseline conditions and return drivers. The creation of a trading strategy cannot only be based on a mathematical model but must be based on a clear understanding of the risks inherent in the return driver. To highlight how to do this, here is an example that I really enjoyed from Jackass Investing.

In 2004, almost 9% of North American adults were on the Atkins diet, where participants must dramatically reduce the amount of carbohydrates they consume to lose weight. The effect of this diet craze was that sales of carbohydrate rich foods (like orange juice) declined and led to lower commodity prices as supply was greater than demand. Because of the lower demand, producers of these commodities shift their production to something more profitable and supply decreases to match demand. This creates a natural “support” level for prices, as at some point, prices will stop falling because producers are leaving the market. All of this is straight out of your Economics 101 textbook.

In 2004, the author found that orange juice prices had fallen so much because of the Atkins craze that they hit a 27 year low in May 2004. He believed that the price of 55 cents a pound of orange juice concentrate was near the support level and his hedge fund began buying orange juice futures. Ultimately, the price rebounded as expected, and reached 90 cents in November-- ultimately reaching $1.50 a pound.

I’ve abbreviated the example and the change in price wasn’t completely attributable to the Atkins diet, as Hurricanes battered the Florida coast that year, drastically reducing the supply. But the return of Dever’s trade was based on a sound return driver, even if the trade might have statistically looked very risky. If the price was anywhere above the support level, this strategy wouldn’t work, but at that support level, there was the necessary driver for the price to appreciate.

In conclusion, do not dismiss strategies that have a solid return driver simply because they have unappealing statistical qualities like high volatility. Many investors do not fully understand how to interpret statistical risk measures and weight them too heavily when making a decision. Please see our website for a complete list of articles in the series.

Disclosure: I have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.