Risk Asset Allocation Based On Momentum Relative To T-Bills

The timing model produced a 19.67% return over the period 1997-August 2018 versus 8.42% for the equal weight portfolio, which in this case is 100% S&P 500. Note that these results are based on investable assets, not indexes.

While results such as this could have been available to you if you did this on your own, they would not have been effective for a “hedge” fund type investment with their 2% management fees and 20% of gains above a preferred return, because the return difference of 2.22% would have been wiped out by their fees.

The volatility (standard deviation) of the method was lower than buy-and-hold. The best year was better, and the worst year was dramatically better (negative 7.67% versus negative 37.02%). The superior maximum drawdown was critical (down 16.31% versus down 50.97% for buy-and-hold).

The average trade lasted 29.9 months, which tended to create capital gains as opposed to ordinary income when trades were executed.

The Sharpe Ratio (return in excess of T-Bill return, divided by the standard deviation of return – how “bumpy” the ride was) was better for the method (0.79 versus 0.48), as was the Sortino Ratio (like the Sharpe Ratio but only divided by the size of downside volatility) at 1.29 versus 0.70.

Last, as a small form of diversification, the correlation to total US stocks was only 0.74 versus 0.99 for buy-and- hold

The complete trade history from 1997 is in this table:

However, not all backtests are inspiring. The specific time period you test and whether or not a Bear intervened are critical to how attractive the method appears.

It can be years of effort with little if any outperformance to show (and taxes in regular accounts) if a Bear is not in the period.

It takes a Bear to make relative momentum versus T-Bills look good.

Best 2 of 5 Risk Assets vs T-Bills

As with the single risk asset (S&P 500) versus T-Bills, the best 2 of 5 risk assets was superior (excluding any tax considerations) to the equal weight, buy-and-hold approach.

Compound return was better by 4.73%. Standard deviation was almost the same. The best year was better by 16.73%. The worst year was better by 13.68%. The maximum drawdown was better (less severe) by 26.15%. The Sharpe Ratio was higher by 0.37. The Sortino Ratio was better by 0.79. The correlation to total US stocks was lower by 0.30, and the Beta to total US stocks was lower by 0.22. All good.

There were many trades for an average trade holding time of 2.37 months. That created a lot of ordinary income for someone doing this in a regular taxable account. The trades over the past year are in this table.

Selecting the best 2 of 5 risk assets might be expected to be more effective during periods without a Bear market, but the evidence in our test suggests that is not the case. Looking at the period 2010 through August 2018 (a period without a Bear), we were able to find an evaluation period that produced a 2+% higher return (before any tax costs), but it produced a 1+% more severe maximum drawdown — it did not provide the downside protection the method is intended to provide.

View single page >> |

Disclaimer: "QVM Invest”, “QVM Research” are service marks of QVM Group LLC. QVM Group LLC is a registered investment advisor.

Important Note: This report is for ...

more
How did you like this article? Let us know so we can better customize your reading experience. Users' ratings are only visible to themselves.

Comments

Leave a comment to automatically be entered into our contest to win a free Echo Show.
Harry Goldstein 2 years ago Member's comment

Good read, than you.