Post 15 discussed an academic article, “Volatility-Managed Portfolios,” linked below, that’s been mentioned in the financial press and has interest from hedge funds. Instead of buy-and-hold it proposes reducing your stock portfolio when volatility rises. Here's its thesis. Sharpe ratio, which equals return divided by volatility, is the most widely used measure of risk-adjusted return. Since volatility is predictable, when it’s high it will more often than naught stay high; Returns, however, are unpredictable, as detailed by the 2013 Nobel prize winner. The arithmetic is: do more investing when the denominator, ie,volatility, is low (and less when it’s high) because this yields higher Sharpe ratios, ie better risk-adjusted return.
By happenstance while reading this article I was battling volatility in my buy SPY-at-noon model and noticed that when volatility is high my model weakens. Specifically, when volatility exceeds 28 SPY-at-noon had negative returns, and so I wrote that post supporting the article.
But recently I revisited SPY from the buy-and-hold strategy suggested by my physicist friend and surprisingly the article’s thesis doesn’t hold with ETFs representing the world’s largest centers. They have increased dramatically more in high VIX periods than in low VIX periods. The table lists returns of ETFs, representing China, Europe and the US, grouped by quintiles of the VIX over nine years. Notice that China, return of 91% in the top two quintiles versus -33% in the other three, and Europe, return of 114 % in the top two versus -14% in the lower three, do even better that the US (93% in the top two versus 45% in the lower three. So all centers have performed better in high VIX periods. I haven't computed annual volatility but it seems clear that these world proxies would not have yielded better risk-adjusted return by lowering investment in high-volatility periods.
To check influence from the end of the financial crisis I dropped July 2009 until June 2010 and results are similar.
As always if you have comments please send them. Meanwhile I’m sending this to the authors for possible comment