Please Note: Blog posts are not selected, edited or screened by Seeking Alpha editors.

Backtesting the Ivy Portfolio Using Relative Strength

|Includes: Vanguard Total Bond Market ETF (BND), DBC, GTAA, SHY, VEU, VNQ, VTI
Perhaps no book is more cited on my site than Mebane Faber's The Ivy Portfolio: How to Invest Like the Top Endowments and Avoid Bear Markets.  Faber does a lot of great work on his site and also has a new tactical asset allocation ETF coming out under the ticker GTAA.

I decided to do some backtests of his original tactical asset allocation portfolios using relative strength instead of moving averages.  The free tool I used for the tests was ETF Replay

The primary drawback of my test is that some of the ETFs available for testing  have limited trading history. Faber had access and time to test underlying indices in lieu of ETFs.  Thus, I am hesitant to draw too many conclusions on the results below (Ihave concluded that Faber's model works).  

When backtesting a portfolio consisting of BND, DBC, VEU, VNQ, and VTI and buying the top 3 ETF at the beginning of each month based a combination of 3 month returns, 20 day returns, and 20 day volatility (each weighted 40%/30%/30% and lower volatility receives a higher weighting), the results are below:

  • Since the start of 2007 the strategy returned 14.8% total, 3.7% CAGR. SPY returned -10.9% total, -3% CAGR.
  • Strategy volatility was 20.8% versus 28% for SPY.
  • Strategy drawdown was -36.8% versus -50.8% for SPY.
  • Note 2 ETFs - BND and VEU - did not begin trading until mid-2007.
When testing the same portfolio and buying the top 3 ETF at the beginning of each month based a combination of 6 month returns, 3 month returns, and 3 month volatility (each weighted 40%/30%/30% and lower volatility receives a higher weighting), the results are below:
  
  • Since the start of 2007 the strategy returned 17.1% total, 4.3% CAGR. SPY returned -10.9% total, -3% CAGR.
  • Strategy volatility was 20.6% versus 28% for SPY.
  • Strategy drawdown was -38.1%.
Note that these test do not take into account moving averages, as Faber's original model suggested.  Thus, the relative strength test was "forced" to purchase 3 ETFs during some significant drawdowns in 2008 even if the ETFs were below their long-term moving average.  Thus, I made a small tweak and added SHY, a short-term treasury ETF as a proxy for cash. 

When performing the same test on an Ivy Portfolio that includes SHY and buying the top 3 ETF at the beginning of each month based a combination of 3 month returns, 20 day returns, and 20 day volatility the results are better:

  • Since the start of 2007 the strategy returned 30.5% total, 7.3% CAGR, and strategy volatility was 15.6%
  • Strategy drawdown was -21.2%
 When testing the same portfolio and buying the top 3 ETF at the beginning of each month based a combination of 6 month returns, 3 month returns, and 3 month volatility the results are the best of the 4 tests:
  • Since the start of 2007 the strategy returned 41.3% total, 9.6% CAGR, and strategy volatility was 15.4%
  • Strategy drawdown was -20.3%
This relative strength test experiences higher drawdowns and volatility when compared to a system that simply buys the 5 ETFs when they are above/below a long-term moving average such as the 10 month moving average.  The 10 month moving average system of BND, DBC, VEU, VNQ, and VTI had a 11.1% drawdown and 10.9% volatility while still returning 23.1% (5.66% CAGR) over the same time period according to ETF Replay. 

Stay tuned for more test results over the next couple of days including one on Faber's 10 ETF Ivy Portfolio as well as a few other test results that may surprise you...(follow me on Twitter for real-time updates)

Get 100 Commission-Free Trades at OptionsHouse.com!
 

Disclosure: no positions