Unigestion says human input required to fully understand minimum variance
Statistical analysis may form the basis of minimum variance strategies, but it requires the human touch to fully understand the context of volatility scores, argues Unigestion.
It may be tempting to judge minimum variance funds purely on their volatility and returns figures, but this is misguided, argues Unigestion, manager of Europe’s largest such strategy.
Alexei Jourovski (pictured), head of the Swiss firm’s €5.4bn minimum variance equities programme says the strategy draws on much more than just statistical sampling of equities’ variance. He says statistical analysis forms only a small part of the total work his 16-member investment team does.
Minimum variance investing stems from academic findings as early as 1991, that stocks with low variance can deliver superior risk-adjusted returns to more volatile shares over time. This is not least because they lose less value in falling markets than more volatile shares.
Unigestion offers various minimum variance strategies with Switzerland, Europe, Japan, the US and emerging markets as their focus, plus a global offering. The funds have achieved about 40% reductions on broad market volatility, while outperforming between 2% and 8% per annum since inception for each regional portfolio.
To construct portfolios takes far more than just using statistical models to identify the ‘lowest variance’ stocks, and those showing low cross-correlation to the market and other holdings.
To illustrate why this would be insufficient, Jourovski picks out carmaker Volkswagen.
Figures show VW shares have been little correlated to European shares since 2008. But answering ‘why?’ needs human insight. It is largely because VW shares jumped by nearly 300% to €945 in one week in October 2008, while all other Dax index constituents fell, as rival Porsche revealed the full extent (74%) of a stake it held in VW. As short sellers rushed to exit VW, they found just 6% of its shares still widely available for purchase. The demand/supply imbalance briefly made VW the world’s largest company by value that week. Absent this abnormal, probably one-off ‘rally’, correlation of VW shares would not be unusual.
Jourovski also highlights managers needing to acknowledge and account for a company becoming more volatile when earnings seasons near. Other events, such as management reshuffles, low trading volumes and M&A activity, each influence volatility.
“Companies are not static time series, they undergo changes and you have to understand variance [in their share prices] is a consequence of a living company,” Jourovski says.