This is a followup to my previous Instablog about the Ultra-Simple Slope technique.
The 3-day technique gave a CAGR of 37.1% from 2004 to 2013, with a standard deviation of 28%, a Sharpe of 1.32 and a max drawdown over almost 30%. I wondered if this could be improved upon.
The 3-day technique invests in the worst-performing ETF over the past three days. I wondered if we selected the ETF that was least correlated with the BEST performing ETF, whether the results would be the same.
Indeed, the inverse correlation technique provided good gains, with CAGR of 30% since 2004 when an 8-day correlation was used.
I thought to combine the two systems to reduce volatility. Initially, I tried the following in which I would buy two positions each trading day: (1) the worst-performing ETF over the last 3 days and (2) the ETF with the least correlation to the best performing ETF over the last three days. This system provided better returns with reduced volatility.
I optimized these parameters for the period 2004-2013 and found that the performance was further improved by using the ETF with the least correlation to the second-best ETF over the last three days.
Thus, the system is: (1) buy 1/2 position in best 3-day ETF and (2) buy 1/2 position in ETF that is least correlated to the second-best 3-day ETF (using an 8 day correlation).
The result is CAGR of 34.5% and a remarkable Sharpe ratio of 1.45, showing the substantial reduction in volatility (stdev now only 23%) with minimal reduction in returns. Max drawdown is about the same at 30%.
Below is performance vs MDY:
Disclosure: I am long SSO.