Tactical Asset Allocation: Are Uber-Simple Systems Robust?

Tactical Asset Allocation: Are Uber-Simple Systems Robust?

September 20, 2014 Tactical Asset Allocation Research
Print Friendly
(Last Updated On: January 18, 2017)

Simple and Effective Market Timing with Tactical Asset Allocation

Abstract: 

A simple market timing algorithm is examined that switches from an exchange traded fund representing U. S. equities to one holding treasury long bonds every month on the last day, the switch being made to whichever ETF has the greatest ratio of current adjusted closing price to adjusted closing price μ months earlier. The parameter μ is determined so as to maximize total return and minimize the total number of trades, however the results are relatively insensitive to μ over a fairly wide range. The performance of this scheme is compared to that of an Ivy 5 portfolio consisting initially of equal dollar amounts of ETFs in U. S. equities, foreign large blend, 7-10 year treasuries, real estate, and commodities. As with the paired switching approach, each ETF is purchased only once a month, on the last day, in this case only if its adjusted closing price exceeds the 10-month simple moving average (SMA). Otherwise that portion of the portfolio is invested in a cash surrogate.

Comparison is made over the 10-year period ending on 12/31/13. It is shown that the average annual return of the paired switching algorithm exceeds 30% in this period, which is 3 times greater than that of the Ivy 5. Moreover, only 45 trades were required for the paired switching approach, whereas the Ivy 5 required 70 in the same period. The maximum draw down was 14.6% for the Ivy 5 and 18.8% for paired switching.

Alpha Highlight: 

The IVY 5 portfolio, described by Faber(2007) and then further elaborated by Faber and Richardson (2009), is a portfolio of 5 asset classes, including the S&P 500 index, the MSCI EAFE index, the US 10-year government bonds, a commercial real estate index, and a commodity index (GSCI). The “simple and effective market timing trading rule” is to buy each index when the monthly price exceeds the 10-month simple moving average (SMA), and to invest in cash (or the risk free asset) when the moving average rule is broken. This strategy is simple and easy, which is why we offer a low-cost version of the concept.

In this paper, the author extends the IVY5 with MA concept, and points out that the simple moving average (SMA) method on the IVY 5 portfolio may not work as well as a 2-asset class paired switching method.   The paired switching method concept requires one to invest in a pair of negatively correlated assets and periodically switch the position based on relative performance of the 2 assets in the pair (eg. SPY and TLT)

The results in the paper show that a simple switching system (Simple TAA) between SPY and TLT outperforms the IVY 5 portfolio from 2004/01 to 2013/12. We replicate the backtest presented in the paper and comment accordingly.

Key Results: 

  1. The Simple TAA strategy works well between 2004/01 and 2013/12.
  2. Out of sample results from 1978 to 2003 suggests that the IVY 5 outperforms the Simple TAA strategy.
  3. Out of sample results from 1927 to 1977 suggests that the Simple TAA strategy is not robust. A fixed allocation can achieve similar performance.

Our Replication Process: 

Same Period in this Paper: 2004/01/02 to 2013/12/31

Here are the acronyms in our report:

  • SPY-TLT_3_MOM: when SP500’s total return over the last 3 months is bigger or equal to TLT’s total return over last 3 months, invest SPY; otherwise, invest in risk-free.
  • IVY5: Equal-weighted portfolio of SP500, LTR (10-year), GSCI (commodity), REIT (commercial real estate), EAFE (international)
  • IVY5_MA_1_10: IVY5_MA is IVY5 with 10-month MA rule trigger: when the last month’s price is bigger than the average of the last 10 month price, risk on; otherwise, risk-off and go to risk-free asset.
  • TLT: iShares Barclays 20+ Year Treasury Bond.
  • SP500: S&P500 total return index.
  • EAFE: MSCI EAFE Total Return Index
  • LTR: The Merrill Lynch 10-year U.S. Treasury Futures Total Return Index
  • REIT: FTSE NAREIT All Equity REITS Total Return Index
  • GSCI: S&P GSCI Total Return CME

Summary Statistics:

The simple TAA (simple switching method) outperforms during 2004/1 to 2013/12.

2014-09-02 10_02_34-Simple and effective TAA_v2.pptx - Microsoft PowerPoint (Product Activation Fail
The results are hypothetical results and are NOT an indicator of future results and do NOT represent returns that any investor actually attained. Indexes are unmanaged, do not reflect management or trading fees, and one cannot invest directly in an index. Additional information regarding the construction of these results is available upon request.

Dollar Growth:

Our replication matches the paper’s results. The chart below is from the paper, and the second chart is our replication results.

Chart from the original paper:

2014-09-02 10_07_07-Simple and effective TAA_v2.pptx - Microsoft PowerPoint (Product Activation Fail
The results are hypothetical results and are NOT an indicator of future results and do NOT represent returns that any investor actually attained. Indexes are unmanaged, do not reflect management or trading fees, and one cannot invest directly in an index. Additional information regarding the construction of these results is available upon request.

Our replication result:

2014-09-02 14_07_31-Simple and effective TAA_v2 - Microsoft PowerPoint
The results are hypothetical results and are NOT an indicator of future results and do NOT represent returns that any investor actually attained. Indexes are unmanaged, do not reflect management or trading fees, and one cannot invest directly in an index. Additional information regarding the construction of these results is available upon request.

Annual Returns:

The simple TAA (simple switching method) works well during the 2008 financial crisis, driven by the performance of the long-bond.

2014-09-02 10_08_29-Simple and effective TAA_v2.pptx - Microsoft PowerPoint (Product Activation Fail
The results are hypothetical results and are NOT an indicator of future results and do NOT represent returns that any investor actually attained. Indexes are unmanaged, do not reflect management or trading fees, and one cannot invest directly in an index. Additional information regarding the construction of these results is available upon request.

DrawDown Analysis:

The simple TAA (simple switching method) has smaller drawdowns.

2014-09-02 10_11_18-Simple and effective TAA_v2.pptx - Microsoft PowerPoint (Product Activation Fail
The results are hypothetical results and are NOT an indicator of future results and do NOT represent returns that any investor actually attained. Indexes are unmanaged, do not reflect management or trading fees, and one cannot invest directly in an index. Additional information regarding the construction of these results is available upon request.

Robustness Analysis:

Market Cycle: Strong recent bull market performance from the simple TAA (simple switching method).

2014-09-02 10_17_56-Simple and effective TAA_v2.pptx - Microsoft PowerPoint (Product Activation Fail
The results are hypothetical results and are NOT an indicator of future results and do NOT represent returns that any investor actually attained. Indexes are unmanaged, do not reflect management or trading fees, and one cannot invest directly in an index. Additional information regarding the construction of these results is available upon request.

Rolling CAGRs: Simple TAA (simple switching method) performance is driven by the past 5 years.

2014-09-02 10_19_55-Simple and effective TAA_v2.pptx - Microsoft PowerPoint (Product Activation Fail
The results are hypothetical results and are NOT an indicator of future results and do NOT represent returns that any investor actually attained. Indexes are unmanaged, do not reflect management or trading fees, and one cannot invest directly in an index. Additional information regarding the construction of these results is available upon request.

Rolling Drawdowns: Simple TAA (simple switching method) has strong drawdown protection.

2014-09-02 10_21_31-Simple and effective TAA_v2.pptx - Microsoft PowerPoint (Product Activation Fail
The results are hypothetical results and are NOT an indicator of future results and do NOT represent returns that any investor actually attained. Indexes are unmanaged, do not reflect management or trading fees, and one cannot invest directly in an index. Additional information regarding the construction of these results is available upon request.

Out-of-Sample Summary Statistics:

1978/1/1 to 2003/12/31 and 2004/1/1 to 2013/12/31

The replication results look good, but the 2004 to 2013 time period is short and unique.

How about some out-of-sample tests?

To do our out-of-sample test, we use the 10 years long-term bond (LTR) to replace the iShares Barclays 20+ Year Treasury Bond (TLT) due to data availability.

  • SPY-LTR_3_MOM: when SP500’s total return over last 3 months is bigger or equal to LTR’s total return over last 3 months, risk on; otherwise, invest in risk-free.
  • IVY5: Equal-weighted portfolio of SP500, LTR, GSCI, REIT, and EAFE.
  • IVY5_MA_1_10: IVY5_MA is IVY5 with MA rule trigger: when the last month’s price is bigger than the average of the last 10 month price, risk on; otherwise, risk-off and go to the risk-free asset.
  • LTR: 10 year long-term bond.
  • SP500: S&P500 total return index.
Summary Statistics (1978/1/1 to 2003/12/31):

Simple TAA underperforms IVY5_MA out of sample. IVY5_MA has much higher Sharpe Ratio than other strategies.

2014-09-02 14_39_50-Microsoft Excel - TAA_analysistool_v29
The results are hypothetical results and are NOT an indicator of future results and do NOT represent returns that any investor actually attained. Indexes are unmanaged, do not reflect management or trading fees, and one cannot invest directly in an index. Additional information regarding the construction of these results is available upon request.

Summary Statistics (2004/1/1 to 2013/12/31):

Simple TAA with 10-year bonds outperforms IVY5_MA, but the results aren’t as strong as the results with TLT. Owning longer duration treasuries through the 2008 crisis is a key driver of returns.

The results are hypothetical results and are NOT an indicator of future results and do NOT represent returns that any investor actually attained. Indexes are unmanaged, do not reflect management or trading fees, and one cannot invest directly in an index. Additional information regarding the construction of these results is available upon request.

Out-of-Sample Summary Statistics:

1927/4/1 to 1977/12/31 and 1978/1/31 to 2013/12/31.

In this out-of-sample test we examine 2 fixed allocation methods.

  • SPY-LTR_3_MOM: when SP500’s total return over last 3 months is bigger or equal to LTR’s total return over last 3 months, risk on; otherwise, risk off.
  • 60_40 SPY-LTR: 60% in SP500, 40% in LTR.
  • 70_30 SPY-LTR: 70% in SP500, 30% in LTR.
  • LTR: 10 years long-term bond.
  • SP500: S&P500 total return index.
Summary Statistics (1927/4/1 to 1977/12/31):

Simple TAA (simple switching method) performs similar on a risk-adjusted basis to fixed allocations. There is no evidence for value-add.

2014-09-02 10_34_18-Simple and effective TAA_v2.pptx - Microsoft PowerPoint (Product Activation Fail
The results are hypothetical results and are NOT an indicator of future results and do NOT represent returns that any investor actually attained. Indexes are unmanaged, do not reflect management or trading fees, and one cannot invest directly in an index. Additional information regarding the construction of these results is available upon request.

Summary Statistics (1978/1/31 to 2013/12/31):

Simple TAA (simple switching method) performs similar on a risk-adjusted basis to fixed allocations. Again, no strong evidence for value-add.

2014-09-02 10_40_14-Simple and effective TAA_v2.pptx - Microsoft PowerPoint (Product Activation Fail
The results are hypothetical results and are NOT an indicator of future results and do NOT represent returns that any investor actually attained. Indexes are unmanaged, do not reflect management or trading fees, and one cannot invest directly in an index. Additional information regarding the construction of these results is available upon request.

 Conclusion

The out-performance of the Simple TAA strategy is mainly driven by long duration bonds. In out-of-sample tests, Simple TAA performs similar to fixed 60/40 and 70/30and underperforms the IVY5 with an MA asset allocation trading rule. Simple TAA is an interesting idea, but robustness tests suggests that we should be suspect of the extreme outperformance during the 2004 to 2013 time period analyzed in the paper.


Note: This site provides no information on our value investing ETFs or our momentum investing ETFs. Please refer to this site.


Join thousands of other readers and subscribe to our blog.


Please remember that past performance is not an indicator of future results. Please read our full disclaimer. The views and opinions expressed herein are those of the author and do not necessarily reflect the views of Alpha Architect, its affiliates or its employees. This material has been provided to you solely for information and educational purposes and does not constitute an offer or solicitation of an offer or any advice or recommendation to purchase any securities or other financial instruments and may not be construed as such. The factual information set forth herein has been obtained or derived from sources believed by the author and Alpha Architect to be reliable but it is not necessarily all-inclusive and is not guaranteed as to its accuracy and is not to be regarded as a representation or warranty, express or implied, as to the information’s accuracy or completeness, nor should the attached information serve as the basis of any investment decision. No part of this material may be reproduced in any form, or referred to in any other publication, without express written permission from Alpha Architect.


Definitions of common statistics used in our analysis are available here (towards the bottom)




About the Author

Yang Xu

Mr. Xu is currently a managing member of Alpha Architect, where he leads the capital markets group and assists in quantitative research. Mr. Xu has unique skills related to "big data" analysis. His recent research investigates various proprietary trading algorithms, tactical asset allocation models, and longer-term security selection models. Prior to joining Alpha Architect, Mr. Xu was a Principal Data Analyst at Capital One, where he was a member of the Basel II data analysis team. Mr. Xu graduated from Drexel University with a M.S. in Finance, and from the University of International Business and Economics in Beijing, China, where he earned a BA in Economics.


  • Tom Rinaldi

    The outperformance of TLT in 2008 and the SPY vs everything since 2008 makes it kind of tough to swallow. Would you really recommend this? I’ve been playing with a “top 3” momentum system with a small number of basic assets. SPY, BND,GLD,PCY and VNQ is what I am toying with. Own a third in each of the top three on one year trailing momentum. It doesn’t seem to work better if I add in EFA or EEM.

  • As the posts highlights, the system is highly optimized and fragile. So no, I wouldn’t recommend it.

    “Top 3” type systems, or any system that boots out a large portion of the asset classes, always freak me out. When you compare the performance of those models to “keep random top 3” simulations, they usually aren’t much different and well within the bounds of random luck.

  • gregorsam

    I beg to differ that the system is not robust.

    During 1991-2014

    VFINX/VUSTX

    1. Swtching every 1 month based on prior three months’ returns:

    CAGR 13.3% Only one year of annual loss.

    2. Switching every 2 months based on prior three months’ returns:

    CAGR 13.5% Only two years of annual loss.

    3.. Switching every 3 months based on prior three months’ returns:

    CAGR 11.6% Only two years of annual loss.

    Similar results are obtained with switching between mid-cap value and VUSTX (IJJ/TLT)

    1991-2014 FDVLX/VUSTX

    1. Swtching every 1 month based on prior three months’ returns:

    CAGR 14.4% Only one year of annual loss.

    2. Switching every 2 months based on prior three months’ returns:

    CAGR 13.7% Only one year of annual loss.

    3.. Switching every 3 months based on prior three months’ returns:

    CAGR 13.6% Only two years of annual loss.

    All the annual losses are quite low except for one case.

    Definitely, the results for the last 24 years suggest that the system is quite robust.

    The fact that you had to dig up the data from 19927-1977 to prove that the strategy does not add value speaks for itself.

    The strategy is based on the sound (mathematical) principle that if two assets are negatively correlated, there will be non-overlapping periods when one performs better than the other. The empirical data suggests that the sign of the cross-correlation is persistent for sufficiently long periods.

    I agree that behaviorally it is hard to completely switch between two assets. The Naive Graham approach (http://seekingalpha.com/instablog/709762-varan/2990923-naive-graham-passive-investing-according-to-the-master ) provides a simple solution to this dilemma.

  • Tom Rinaldi

    Wes, I guess what I’m trying to combat is the situation like now where only one asset class is really working and you only have 15 or 20 percent in it and commodities are under all moving averages so you have another 15 or 20 percent in cash. I’m speaking to like an ivy five with a moving average. I guess I am looking for something in between ivy five with moving average and and a SPY/TLT system. So thinking aloud maybe a five or six asset system that hold the top three assets on one year momentum….

  • Tom Rinaldi

    Something tells me if you added GLD to SPY and TLT and did a top two on one year momentum it would be pretty interest. Like the three basic elements!

  • Tom,
    Who knows what will work best in the future. As long as you stay discipline to a model that seems reasonable and has elements of diversification, I’m sure you’ll do a-okay. Keep fees low and defer taxes and you’ll really be playing ball…

  • eber terandst

    Two points:
    The simple approach proposed ( SPY vs TLT based on 3months ) is easily duplicated on etfReplay. The results are quite different : for the period 2003 – 2013, the CAGR is only 13.6%, not 16.94 % as reported in the article. The Maximum drawdown is -17.1%, not -14.4%. Some years are extremely different, example: 2006 is only 4 %, not the 12.87% reported in the article.
    I have found that the data and methodology from etfReplay are extremely accurate, so I have serious doubts about the methodology used in the article.
    Question: where did you get the bond data going back to 1927 ? I have been simply unable to get them even from pay source.
    Thanks
    eber terandst

  • http://www.hec.unil.ch/agoyal/ has data on 10year. We splice that data series with bloomberg ML 7-10 total return index.

    Yes, ETFreplay is a great tool.

  • Smruti

    The comparison between SPY and UST switching with a 5 asset IVY portfolio is not really apples to apples because the investment universe is different. Maybe taking the same 5 assets and picking the best performing over the past ‘n’ months would have been a better comparison.

    On a slightly different note, these simple ETF switching models (which can be so easily tested in ETFreplay) are very sensitive to the assets in your universe. Which of these models will work going forward will probably depend on what assets you have put in there rather than the algorithm itself.

  • Григорий Исаев

    the last 30 years or so have seen the mother of all bear markets in USD interest rates, which has ofcourse supported both us stocks and goverment bonds. to get anything near the same result you have to bet this bear market will continue going forward, which does not look like a wise assumption for the NEXT 30 years, given that short term rates are 0.

  • yep, outstanding point and highlights why this result is probably an artifact of data-mining and good luck, and not very robust.

  • Григорий Исаев

    Also, the negative correlation between stocks and bonds is simply not there in the larger datasample.

    https://media.pimco.com/Documents/PIMCO_Quantitative_Research_Stock_Bond_Correlation_Oct2013.pdf

    Wesley, thanks for the site, it is really really interesting.

  • All true.

    The one thing interesting about treasury bonds is their “flight to quality” aspect. One can argue that the long bond might lose that quality going forward, but as long as the US/USD are king, the “insurance” aspect of long bonds make for an interesting diversification asset.

  • dph

    Wes,

    Would you advocate a leveraged IVY5_MA type system given the historically low drawdowns and high sharpes across many different decade/ market conditions?

  • Jack Vogel, PhD

    Not a bad idea, but that would increase drawdowns. Also, you would want to make sure the margin rate you are paying is not too high.