The moving day average (MDA) has long been recognized as a simple tactical growth and risk mitigation strategy for retirement portfolios. The basic idea is to be in a stock/ETF when it is above the MDA, and to be out of a stock/ETF when it is below the MDA. Research has shown that large drawdowns can generally be avoided and overall total returns improved by using the MDA strategy.
The most common MDA used by strategists is the 200 MDA. It is very obvious that many investors utilize a 200 MDA strategy because of daily stock market action when the stock/ETF approaches and/or goes below the 200 MDA. Other strategists, such as Mebane Faber, prefer a monthly moving average and monthly updating strategy to avoid excessive trading caused by daily volatility of the stock/ETF. Based on extensive research, Faber recommends using a 10-month moving average (MMA) strategy with monthly updating in his paper entitled A Quantitative Approach to Tactical Asset Allocation. A 10-month MMA roughly corresponds to a 220 MDA although the two do not give identical results because of different statistics.
The figures below show the backtested ETFreplay results using moving average strategies on the S&P 500 Index (SPY) from 2000-2013. Portfolio money was either in SPY if SPY was above its moving average, or in cash if it was not. The first figure shows results using a 200 MDA with trades/updates performed at the end of the crossover day. The second figure shows the results of a nine-month MMA with trades/updates performed monthly (end of month). Please note that the nine-month MMA roughly corresponds to a 198 MDA, but statistically nine values are averaged in the former while 198 values are averaged in the latter. The green curve called Equity Curve in the figures represents the results of using the moving average strategy. It can be seen that the nine-month MMA strategy with monthly updating produced much better total return (217% to 68%). The maximum drawdown is 12.1% for the nine-month MMA strategy and 27.3% for the 200 MDA strategy updating at the end of crossover day. To show the effect of updating schedule, the 200 MDA strategy was also run with monthly updating rather than updating on crossover day. The total returns of the 200 MDA strategy with monthly updating are close to the nine-month MMA strategy with monthly updating (238% to 217%).
However, as is well-known, there are many challenges using the MDA/MMA strategy. These challenges consist of:
- A long duration MDA/MMA (e.g. 100-200 days/five-nine months) must generally be used if the ETF is somewhat volatile, like most equity ETFs are.
- Multiple crossover points of the ETF with the MDA/MMA, especially when there is no overall trend of the ETF, result in poor performance (total return and drawdown) of the strategy.
- The response time of long duration MDA/MMA strategies is excessive. Risk mitigation against sudden drawdowns is sometimes difficult to obtain.
- The optimum MDA/MMA for any given ETF is a variable. For instance, the 200 MDA strategy with monthly updates works well for SPY, but it does not work well for some more volatile equity ETFs, such as the iShares Core S&P Small Cap 600 Index ETF (IJR).
Some researchers have proposed alternatives to the basic MDA strategy by introducing strategies that use crossover points of two MDAs. Many strategists use the crossover point of the 50 MDA and the 200 MDA to signal selling or buying of any given stock/ETF. However, this type of strategy results in a slow response time to market conditions, especially when sudden changes occur. By the time the crossover signal has occurred most of the drawdown has already happened.
In this article, I am proposing a unique application of the MDA strategy. In order to improve response time, avoid large drawdowns, improve performance and reduce the number of trades, I am proposing the use of very short duration (less than 20 days) MDA strategies on low volatility, high yield bond ETFs. I know this may sound counterintuitive, but please hear me out. I think it makes sense.
It is a well-known fact that bond ETFs generally experience lower volatility than equities. For instance, a low duration bond ETF such as the AdvisorShares Peritus High-Yield Bond ETF (HYLD) had a volatility of 7.3% between 2011-2013. In contrast, SPY had a volatility of 16.5%. Bond ETFs still experience many drawdowns, as many as equities, but the drawdowns are not as large or as volatile. Because of lower volatility of a bond ETF, shorter duration MDA strategies (20 days or less) can be applied more successfully than if they were applied to equity ETFs.
In order to test out my hypothesis, I attempted to find bond ETFs with very low volatility but still having good potential for a decent total return. In the current financial environment with the threat of higher long-term rates, the best bond ETFs for total return seem to be the short duration, high yield bond ETFs such as the PIMCO BofA ML 0-5 Year High Yield ETF (HYS), and bank loan ETFs such as the PowerShares S&P-LSTA Senior Loan ETF (BKLN). There is some risk in this class of ETFs because they invest in so-called "junk" bonds/loans. These ETFs might experience default of some holdings if the economy heads south. But the economy seems to be doing well right now and default rates are extremely low, so this risk seems minimal at the present time. Also, short-term interest rates probably will not be rising anytime in the foreseeable future (good for HYS), and the adjustable interest rates of holdings in BKLN seem to negate interest rate concerns in this ETF.
HYS and BKLN have not been around very long; both have been in existence for about 2 ½ years. But in this time, these ETFs have experienced three minor drawdowns while maintaining consistent yearly total return. Although my desire is always to have long periods of backtesting (the longer, the better), HYS and BKLN can only be backtested to 2011. The overall volatility of HYS is 3.3% over the last two years, while the volatility of BKLN is 3.0%. Both volatility levels are extremely low compared to equity ETFs and even other bond ETFs.
The backtested results of HYS using the ETFreplay software are shown below using a 20 MDA strategy. In order to reduce short-term trading, I have also introduced a second MDA (4 days). The end-of-day crossover between the 20 and 4 MDAs is used as the signal point to exit HYS and go into cash or vice-versa. The total return of this MDA strategy from June 17, 2011 - end of year 2013 is 26.2% compared to 20.3% for HYS by itself. The maximum drawdown is only 1.8% for the strategy compared to 8.3% for HYS by itself. The MDA strategy resulted in only 11 trades (one still open) from June 2011 - end of 2013. The cost on the Schwab platform was just $198 for the entire 2 ½ years.
The backtested results are even more impressive for the MDA strategy used on BKLN. In this case, because of BKLN's lower volatility, MDAs of 12 days and 3 days were used. Their crossover signaled the time to move from BKLN to cash and the time to move from cash into BKLN. The total return of the strategy from March 3, 2011 to the end of 2013 was 24.6%, while BKLN by itself only returned 13.0%. The maximum drawdown of the strategy was 1.5% compared to 8.7% for BKLN by itself. The MDA strategy resulted in 17 trades (one still open) in this time frame. The total expense in this case was $0 on the Schwab platform since buying/selling BKLN is commission free.
If the backtested results hold true for the future, then this unique short duration MDA strategy on low volatility, high yield bond ETFs is a viable way to diversify into bond assets under current conditions. The total returns may not be as good as equities during current market conditions, but the lower volatility of bonds and the fast response/risk mitigation of this strategy using low duration MDAs might appeal to some conservative investors. And the compounded annual growth rate, CAGR, is still above 9% for both the HYS and BKLN MDA strategies for 2011-2013.