We use cookies to personalize content and ads and to analyze our
traffic.
We also share information about your use of our site
with our advertising and analytics partners.
See details.

We use cookies to personalize content, ads, analyze traffic
and share information about your use of our site.
See details.

Meanwhile in China: Which Strategies Protect Your Portfolio Against the Bear Market?

By Bryan Chik, Portfolio Analytics Specialist - Asia Pacific

Jul 23, 2015

While the world is watching closely what will happen to Greece, here in Asia Pacific investors have witnessed another epic story. After a sustained rally, Chinese Equity markets have suddenly retracted and moved into an official bear market. The Shanghai Stock Exchange Composite index has taken a dive of 47.68% from its highest point on July 8. Government attempts to stabilize markets so far have not been successful, with single day drops of nearly 6% in both the Hang Seng and Shanghai index and suspended trading of more than 50% of Chinese stocks as a consequence.

Such a dramatic index performance quite naturally leads to the question: could one have done better? With the emerging discussion on risk-based portfolios, I set to explore whether these types of strategies would have performed better than the index.

Ways to Hide

The four construction techniques I explored were Minimum Variance, 1 over N (equal weighted), Maximum Diversification, and Risk Weighted.

I chose the Shanghai SSE Composite index and ruled out any stocks that were not trading in the last 30-trading days as my trading universe. I then formed the following portfolios as-of May 29, 2015 and observed the performance in the subsequent period to July 6, 2015. Risk statistics were calculated using one of the China Short-Term horizon models available on FactSet.

Despite their risk based nature, the 1 over N, Risk Weighted, and Maximum Diversification strategies provided little protection in the most recent market decline. The only strategy outperforming the index over this time period is the Minimum Variance approach. The results make sense, because although these techniques all promise risk-aware portfolios, only the minimum variance approach minimizes market exposure.

In predicted beta, the Risk Weighted and 1 over N strategies produced portfolios with a predicted beta close to one; maximum diversification and minimum variance give us a lower beta. The moment market further declines and volatility spikes, the beta of the Risk and Equal Weighted strategies increase to 1.1 even though we are not rebalancing the portfolio, giving us more, rather than less, exposure.

Looking at the predicted risk, the absolute risk of the strategies was initially rather stable. But when the market started to decline, total risk increased significantly. By definition, it is not surprising to see not only that Minimum Variance has the lowest absolute risk, but also that it doesn't escape the spike in volatility.

Quantifying Diversification

"Don't put all your eggs in one basket" is one of my favorite investment philosophies. Since diversification is an important aspect in my portfolio construction methods, I wanted to see how well-diversified these portfolios are. In Toward Maximum Diversification, Choueifaty and Coignard [2008] propose a simple ratio quantifying diversification by dividing the weighted sum of security volatility by the total volatility of the portfolio. The higher the ratio, the more diversified the portfolio.

Where:

w_{i} : Security weight

σ_{i}: Security volatility

σ_{p}: Portfolio volatility

Instead of using realized volatility measures, I used the ex-ante volatility calculations, using the same risk model as above to calculate the ratio.

By construction, I should expect my Maximum Diversification strategy to be the most diversified, which is confirmed as it as the highest Diversification Ratio. Despite being a much more concentrated portfolio, Minimum Variance came in second. For the 1 over N and Risk-Weighted strategies, even though they have the largest number of stocks, they have less diversification. Although initially this might seem counterintuitive, what they lack of is a concept of correlation. In other words, they score well on the numerator but not on the denominator of the above formula. What might be more interesting is that for all strategies diversification decreases over time and Minimum Variance seems most resilient.

Zooming Out

The above analysis gives us some interesting perspectives on how the different strategies would have performed when the Chinese market moved into bear territory. There is however one slight caveat: that we would have timed the market perfectly. I constructed the portfolios just before the market went south, which is much harder to do in practice than in hindsight. Therefore I also constructed the portfolios at the beginning at the year and examined how they fared in both the up and down market.

This performance scenario paints a rather different picture. What we observe is that all the strategies outperformed the benchmark over this period. Even though Minimum Variance was the worst performing strategy, it still outperformed the benchmark by 25% (portfolio return 58.37% vs benchmark 32.97%). During the period of up-rally, all strategies outperformed the benchmark, with Maximum Diversification faring best. The 1 over N and Risk-Weighted strategies lagged slightly in capturing the upside but essentially picked up in late May, at which point the market was at peak. As seen before, with the exception of Minimum Variance, the three other strategies were less resistant to the down market. Overall, all of them outperformed.

Conclusion

What can we conclude? First, from the four strategies, only Minimum Variance offers us some protection in a down market. While the other strategies didn't provide protection during the down market, their risk aware construction should give us is a better risk adjusted returns profile, and their returns on a year-to-date basis clearly seemed to point in that direction.

Receive stories like this to your inbox as they are published. Subscribe by e-mail and follow @FactSet on Twitter. If you are looking to source FactSet data or analytics in your publication, email media_request@factset.com.