Risk Mitigation Obtained By Investing In Multiple Tactical Strategies

by: Cliff Smith


Investing in multiple tactical strategies at the same time reduces risk compared to investing in just a single tactical strategy.

In this article, the allocation between ten diversified tactical strategies is optimized for maximum Sharpe Ratio. Monthly correlations between strategies mainly range between 0.3 and 0.7.

In extensive backtesting (1988 to present), the optimized allocation between strategies produces an annual return of 11.7%, a maximum drawdown (MaxDD) of -3.6%, and a risk-adjusted return (MAR) of 3.3.

Although some enhancement is seen with optimization, most of the risk reduction can be captured just by using an equal weight distribution between strategies.

The equal weight distribution portfolio produces an annual return of 11.5%, a MaxDD of -3.9%, and a MAR of 3.0. The MAR is significantly higher than the best individual strategy.

Diversification is a well-known method of reducing risk in investment. But little has been reported on diversification of tactical asset allocation strategies. It makes sense that investing in a number of diversified strategies should reduce overall portfolio risk in terms of standard deviation and Maximum DrawDown, or MaxDD.

As many of my readers know, I currently invest in ten different tactical strategies that I developed using Portfolio Visualizer, PV. I've often wanted to know how the combined package of strategies performed in backtesting. Two questions I've often wanted answered are: 1) Are there any benefits in investing in a diversified set of strategies? and 2) Does optimization of the allocation between strategies enhance performance and/or risk?

Well, to my surprise (and maybe yours), the capability to answer these questions is already available in the PV software. In a recent SA message, Michael Morgenstern told me how to perform this type of calculation in PV (thank you, Michael).

Here's how you do it. Each strategy is run separately, and from the PV results, csv files can be generated for each strategy. Each csv file has columns of date and monthly returns. The csv files are then input into PV using the "input benchmark" feature under your user tab (you need to login first). You can identify the benchmark files with ticker names, and then use the ticker names like any other ETF or mutual fund.

This opens up a lot of opportunities in PV that I had thought were impossible. The optimum allocation of strategies can be determined. Momentum models can be tested not only on individual strategies, but also on multiple strategies. Defensive momentum strategies can be assessed as a substitute for a single risk-off asset. I'm sure there are more ways to utilize this capability, but these few examples will suffice for now.

In this article, I will just examine the optimization feature. I loaded the results of ten different strategies into the Mean Variance Portfolio Optimization tool in PV. PV explains the optimization tool in this way,

"This online portfolio optimizer tool performs Markowitz mean variance optimization on the provided portfolio to find the optimal risk adjusted portfolio that lies on the efficient frontier. Supported optimization goals include:

  • Minimize risk for a given expected return
  • Maximize the Sharpe Ratio of the portfolio.

Mean variance optimization is based on the monthly return statistics of the selected portfolio assets for the given time period. The optimization result does not predict what allocation would perform best outside the given time period, and the actual performance of portfolios constructed using the optimized asset weights may vary from the given performance goal."

I will not go into specifics about each strategy I included in the analysis. Many of the strategies feature Vanguard mutual funds and ETFs. Some of the funds and ETFs in these strategies include:

1. Vanguard Convertible Securities Fund (MUTF:VCVSX)

2. Vanguard Wellesley Income Fund (MUTF:VWINX)

3. Vanguard High Yield Corporate Fund (MUTF:VWEHX)

4. Vanguard High Yield Tax-Exempt Fund (MUTF:VWAHX)

5. Vanguard GNMA Fund (MUTF:VFIIX)

6. Vanguard Long-Term Treasury Fund (MUTF:VUSTX)

7. Vanguard Intermediate Treasury Fund (MUTF:VFITX)

8. Vanguard Short-Term Treasury Fund (MUTF:VFISX)

8. Nuveen High Yield Municipal Bond Class A Fund (MUTF:NHMAX)

9. Nuveen High Income Bond Class A Fund (MUTF:FJSIX)

10. Guggenheim S&P 500 Equal Weight Healthcare ETF (NYSEARCA:RYH)

11. Guggenheim S&P 500 Equal Weight Consumer Staples ETF (NYSEARCA:RHS)

12. Schwab US Aggregate Bond ETF (NYSEARCA:SCHZ)

13. Vanguard Long-Term Government Bond ETF (NASDAQ:VGLT)

14. Vanguard Health Care ETF (NYSEARCA:VHT)

15. Vanguard Information Technology ETF (NYSEARCA:VGT)

16. Vanguard Consumer Discretionary ETF (NYSEARCA:VCR)



The first eight strategies are based on articles I have written in SA (and actually invest in). The last two strategies are versions of strategies proposed by dfrtr (thanks, drftr). I included two of drftr's strategies to enhance the diversification of strategies. Shown in the table below is the performance and risk of each strategy. Each strategy has a ticker name: TACTICAL01, TACTICAL02, . . . TACTICAL08, TACTICAL09, and TACTICAL13.

Click to enlarge

Each strategy features moderate growth (Compounded Annual Growth Rate, CAGR, of 9.7% to 14.2%) and low MaxDD (-4.3% to -7.9%). For reference, the risk adjusted return ratio, or MAR (CAGR/MaxDD) of each strategy is:

Also note how each strategy correlates to the US market. We obviously want our strategies to correlate well with the US market when the market is in an upturn (near 1.0), and to not correlate well when the US market is in a downturn (near zero or negative if possible). Thus, we desire correlations with the US market to be somewhere between 0.25 and 0.75.

The correlations between strategies is shown in the table below. The missing headers for the columns going left-to-right are TACTICAL01 through TACTICAL13. The correlations are less than 0.81 and are generally in the 0.3 to 0.7 range. The lowest correlation is 0.29 (strategy 5 versus 6). The one anomaly is the correlation of 0.97 between strategies 2 and 8. This means these strategies are essentially redundant, and one can be eliminated. But for this exercise, I have kept both strategy 2 and strategy 8.

Click to enlarge

The allocation between these strategies was optimized to maximize Sharpe Ratio of the portfolio. Sharpe Ratio is a risk adjusted return ratio that takes the CAGR and divides it by the Standard Deviation, SD. I personally like to use MAR (CAGR/MaxDD) as my risk adjusted return ratio, but PV cannot optimize on MAR.

In the results, three allocations are noted. These are defined as: 1) Provided Portfolio that is an equal weighted portfolio, 2) Volatility Weight Portfolio that adjusts the weighting based on volatility, SD, and 3) Optimized Portfolio that has the optimized allocation for maximum Sharpe Ratio of the portfolio. The allocations of each portfolio are shown below.

Click to enlarge

Click to enlarge

Please note the optimized allocation: TACTICAL02 (15.8%), TACTICAL03 (9.6%), TACTICAL04 (15.8%), TACTICAL05 (18.1%), TACTICAL06 (33.4%), and TACTICAL13 (7.3%). It is interesting that TACTICAL01, TACTICAL07, TACTICAL08, and TACTICAL09 do not have any allocation.

A summary of the PV backtest results (1988 - Sept 2016) is presented below. In addition to the results of the three portfolios mentioned above, the Vanguard Total Bond Market Index Fund (MUTF:VBMFX) is included for reference. It is interesting to note that all three portfolios give similar results; relatively minor differences are seen. For the optimized portfolio, the CAGR is 11.7%, the SD is 4.4%, the worst year is 3.6%, and the MaxDD is -3.6%. The adjusted risk return ratio MAR is 3.3, and the monthly win rate is 77.1%. For the equal weight portfolio, CAGR = 11.5%, SD = 4.7%, worst year = 1.8%, and MaxDD = -3.9%. The MAR = 3.0, and the monthly win rate = 74.5%.

Click to enlargeAn interesting observation is the reduced level of MaxDD of both the optimized and equal weight portfolios compared to the individual strategies. The best MaxDD for the individual strategies is -4.3% for TACTICAL01, while for the combined strategies the MaxDD is -3.6% (optimized) and -3.9% (equal weight). Also note the MARs are better for the combined strategies (3.3 for the optimized portfolio and 3.0 for the equal weight portfolio) compared to the MARs of the individual strategies (best = 2.5 for TACTICAL13). And finally, the SDs of the combined strategies (4.4% for optimized portfolio and 4.7% for equal weight portfolio) are lower than the lowest SD of the individual strategies (best = 5.0% for TACTICAL01)

Other results from PV are shown below, including Portfolio Growth, Drawdowns, Efficient Frontier, Annual Returns, and 3-Year and 5-Year Rolling Returns.

Click to enlarge

Click to enlarge

Click to enlarge

Click to enlarge

Click to enlarge

Click to enlarge

Click to enlarge

In conclusion, the allocation to ten different tactical strategies has been optimized to give the highest Sharpe Ratio (CAGR/SD). The results are very similar to an equal weight distribution between strategies. By combining the strategies it can be seen that the MaxDD and SD are reduced and the adjusted risk return MAR is increased from the best of the individual strategies. It is seen that an equal weight distribution obtains most of the gains that can be achieved by investing in multiple, diversified tactical strategies in one's portfolio.

Disclosure: I am/we are long VCVSX, VWEHX, FJSIX.

I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it. I have no business relationship with any company whose stock is mentioned in this article.