Backtesting 3 ETF Portfolio Strategies

Follow me on Twitter

Enjoy this free post or others? Write a review here or tell your friends, it is much appreciated!

======

A reader asked for comparison of different ETF strategies.  The primary limitation of backtesting ETFs is that many ETFs have limited trading history, so it is difficult to draw significant conclusions from test results.  Having said that, I created a modified portfolio of 5 ETFs based on the portfolio listed in Mebane Faber's The Ivy Portfolio.  The portfolio consists of AGG (Aggregate Bond), DBC (Commodity), EFA (International Developed), VNQ (REIT), and VTI (US Stock Market).  I substituted AGG for BND and EFA for VEU because they have longer trading histories.  The portfolio and strategies highlighted below are similar to concepts used in the "Basic Portfolio" I publicly track at Scott's Investments.

Strategy 1 is to buy the ETFs when they are above their respective 10 month simple moving average (SMA) at month end.  When the ETF is below the 10 month SMA, the position is sold and proceeds are held in cash.  It is an equal weight portfolio.  I tested the strategy from 2006 to present.  DBC started trading in February 2006 so for the first 11 months of the test, only 4 ETFs were available (10 months of trading history for DBC were required before a 10 month SMA could be generated). The results are below:

Total Changes	Total Return	CAGR	Volatility
23	64.10%	10.13%	10.20%
Total Days	Sharpe Ratio	Strategy Drawdown
1293	0.71	-9.80%

Next, I tested the results of buying and holding this portfolio since 2006, equal weighted. There is some discrepancy with test #1 because in order to generate a 10 month moving average, DBC needed 10 months of trading history.  Thus, DBC was not incorporated into test #1 until December 2006.  However, for the purpose of test #2 I started the test in January 2006 with DBC being added February 6th 2006, its first day of trading.

The results of test #2, buy and hold:

Total Return	CAGR	Volatility
22.00%	4.00%	18.90%
Sharpe Ratio	Strategy Drawdown
0.09	-46.60%

The third test is a relative strength test with 2 variations.  The first variation  is to buy the top 1 ETF based on a combination of 3 month returns (weighted 40%), 20 day returns (weighted 30%), and 20 day volatility (weighted 30%, the lower the volatility the higher the rank).  The second variation is to buy the top 1 ETF based on a combination of 6 month returns (weighted 40%), 3 month returns (weighted 30%), and 3 month volatility (weighted 30%, the lower the volatility the higher the rank).  As with the other strategies, there is some discrepancy with the date DBC is incorporated into the strategy.  In order to rank DBC on historical returns, it needs sufficient trading history before it can be used. For example, in the 3/20/20 strategy DBC was not available until June 2006.

The strategy is updated/rebalanced monthly at the end of the month.  The results for both variations are below:

3/20/20 Variation
Total Changes	Total Return	CAGR	Volatility
36	169.80%	21.30%	18.50%
Total Days	Sharpe Ratio	Strategy Drawdown
1293	1.25	-7.20%
6/3/3 Variation
Total Changes	Total Return	CAGR	Volatility
28	171.00%	21.40%	18.60%
Total Days	Sharpe Ratio	Strategy Drawdown
1293	1.13	-10.30%

I will let the reader draw his or her conclusions from these results.  While 5 years may seem like a long time period, in the world of investing it is difficult to draw long-term conclusions from a relatively short time period.  However, one fact is obvious - in recent years trend trading this particular ETF portfolio has outperformed buy and hold by a significant margin, with lower drawdowns and less volatility.  Also, we can use other, longer term test results (such as those in The Ivy Portfolio) which use indices (as opposed to ETF results) to backtest strategies similar to those highlighted above.  Those tests tend to support the effectiveness of momentum and tactical asset allocation as a portfolio strategy.  Of course, history may not repeat itself.


Notes: Test results include dividends and tests were performed using ETF Replay.  No positions in ETFs mentioned.  For additional studies on momentum, including studies that both dispute and support the effectiveness of the momentum effect, I recommend CXO Advisory.