Quant Concepts: Using Beta to Build Low Volatility

Discipline around market deviations in your portfolio can bring stability and downside defence, Emily Halverson-Duncan finds 

Emily Halverson-Duncan 19 June, 2020 | 1:49AM
Facebook Twitter LinkedIn



See more episodes of Quant Concepts here

Emily Halverson-Duncan: Welcome to Quant Concepts' virtual office edition. While there are a lot of different ways to measure stock volatility, beta remains one of my personal favourites. To summarize, beta is a measure of how sensitive a company is relative to changes in a defined benchmark. For example, in Canada, a common benchmark would be the S&P/TSX, while in the U.S. you would likely use the S&P 500. A company with a beta of less than 1 has historically moved less in the benchmark. So, from a risk perspective, lower values of beta are considered preferable. Today, I've created a strategy which looks for Canadian stocks with a low measure of beta across the past one, three and five-year time periods. The purpose of the strategy is to produce a portfolio that is less volatile than the market. So, let's take a look at how to build that.

First off, as always, we are going to go through and rank our universe of stocks. Here, our universe is all the Canadian stocks in the CPMS database and today that holds 698 names. In that ranking step, the factors we're going to look at as mentioned are one, three and five-year beta. So, the different timeframes there just refer to what we're looking at in terms of price data of a stock relative to the TSX. So, on the last one year, we are looking at how sensitive the stock's price has been in the last year compared to the TSX, three years would be across three years and then lastly, five years, of course, across the longer five-year timeframe. In all three metrics, lower values are preferred.

From there, we move on to our screening step. So, some of the screens we're going to apply here – that five-year beta, we want to see a value of less than or equal to 1. So, again, forcing that to be as sensitive or less sensitive than the market in the long-term. Market cap, we're setting that to be above – you can see the C minus here, that's roughly 50% of peers, which today has a value of about 156 million or higher. The reason for that is just to remove the really small-cap stocks in the universe. And lastly, earnings per share variability. This is another risk metric. We want that to be in the lower half of peers, so about 16.8% or below. What this metric is looking at is how volatile a company's earnings per share are. So, if they are constantly reporting earnings that vary significantly and swing up and then swing down, they would have a higher value here. If they are more stable quarter-over-quarter, you'd see a lower value.

On the sell side, once we've applied all those filters, we're going to sell the stocks once their beta rises above a 1.2 and again, that's the five-year beta. So, at that point, we're thinking it's too sensitive to the market, too overly sensitive and we're going to get rid of it and then, that earnings per share variability, if it goes into the higher one-third of peers, so that's 19.8% or higher.

Once all of that's applied, we're going to go ahead and run our back test and see how this model performed. So, let's take a look at that. Back test here holds 15 different Canadian names and we're running it from October 1990 until the end of May 2020. Across this timeframe the model returned 13.2% annualized and that's an outperformance over the TSX of 5%. So, pretty good outperformance there. Turnover is about 48%. Turnover, as mentioned before, is looking at how often you're trading in your stock model. So, about 48% means of 15 stocks you're trading probably about half of them on average in a year or that translates to probably about 7 or 8 trades.

Some of the metrics we want to look at, again, the focus of this model and the point of the model was to try and build something that's less volatile compared to the market. So, a couple of metrics that we can look at – standard deviation and downside deviation. Standard deviation looks at the volatility of all returns. So, for the strategy, I can see it's 11.9%. Versus the benchmark, it's 14.1%. So, again, we want a lower value there. So, the strategy is doing better.

My preferred metric to look at though is downside deviation. And the reason I like this one better is it's a similar metric but we're only looking at the volatility of negative returns. So, the strategy here I can see the value is 6.7% whereas the benchmark is 9.9%. So, that 6.7% shows that the strategy is managing those negative returns, the volatility there, a lot better than the benchmark is.

Lastly, my favourite green and blue chart I always like to look at looks at how the model did in both up and down markets. In up markets, the model outperformed 41% of the time, while in down markets, it outperformed 76% of the time. So, again, all in line with what we would be trying to do here and what the variables typically look for and that's to protect on the downside. So, again, beta, one of my personal favourites. It might be worthwhile incorporating into a model of your own.

For Morningstar, I'm Emily Halverson-Duncan.

Are you getting the right returns?

Learn what to expect from your equity management here

Facebook Twitter LinkedIn

About Author

Emily Halverson-Duncan  Emily is Director, CPMS Sales at Morningstar

© Copyright 2022 Morningstar, Inc. All rights reserved.

Terms of Use        Privacy Policy