ETF Sector Strategy
The goal of this article is to demonstrate a relatively simple method of investing in ETFs. A trend following system was chosen to keep funds invested in the currently dominant sectors of the market. A monthly review cycle was chosen to minimize the time commitment of the strategy. This isn't hyper-day-trading or anything like that, but it is active management. In fact, as you will soon see, the system only switches funds about 4 times per year. Because we don't want all of our eggs in one basket we will always have our funds divided among two sectors of the market. More precisely, two sectors or styles.
This project began with the family of 9 Select Sector SPDR ETFs. A fund family created to divide the S&P 500 stocks into independent market sectors. Being sub-sets of the S&P 500 stocks and equity weighted, these ETF's all tend to be dominated by large cap stocks. But large cap is not always where you want to be so the style-based Russell iShares funds were added to expand the options. In hindsight we could probably have just added the small-cap Russell ETFs, but we included all 12.
That gives us 21 ETFs to screen each month, covering the major market sectors and styles. The next question to be answered was how to sort them. We could do some complex screen, but we only have a short history to backtest so the decision is made to keep it simple and use the trailing 6 month return. On this site that equates to the past 126 trading days. Each month we'll calculate the total return for these 21 funds over the prior 126 days (ending on the last day of the prior month) and sort the ETFs by this return in descending order. The first month we'll buy the top two positions, but in subsequent months we'll hold a position as long as it is in the top 6. This keeps us in the top quarter to third of our universe and reduces fund switches. If we do exit a position, or both of them, we'll buy the top funds we don't already own. If one posiiton is closed out we'll just roll that money into the new position. If both are closed we'll divide proceeds between the new funds. This is an imperfect rebalancing method but saves on transaction fees.
ETFs are relatively new to the investment scene with most being created after 2000. The Select Sector family in our test began trading in late 1998 while the Russell ETFs began in early 2000. Since we need 6 months of trading history we'll begin our short backtest on January 2, 2001. You can see in Table 1 how the funds ranked for the first 12 months of our backtest.
The first month we buy the top two positions, XLF and XLU, and we hold these until they drop below position 6 in our screen. Following the information in table 1 you will see that as of the first of March XLF has dropped to position 8 triggering a sell of that holding and a buy of IWN, which was the top position we did not already hold. Then by May XLU has dropped to the number 7 position and is replaced in our portfolio by XLB. IWN and XLB remain in the top 6 positions throughout the remainder of the period shown in Table 1 and well into 2002.
Table 2 shows all the trades in our backtest along with the dates and returns for those trades. The test began on 1/2/2001 and ended on 8/31/2005. You can see that 10 of the 16 trades were positive with 9 outperforming the SPY during the same time period. At least as important to me is that during periods of significant market movement our model seemed to outperform. When the market was up we were up more, and when the market was down we were either up or down less.
The CAGR of this system was 15.7% over this time period while the market returned an annualized 0.4%. Not to say that such performance will continue but it is nice to start knowing your system has the potential to significantly outperform the market. More stats are in Table 3.
Month by month performance data for the model and the SPY are in Tables 4 and 5. As you can see from this data our simple model has outperformed the broad market in 37 of 56 months and for every year of our test period.
As the graph below shows this model has nearly doubled in value during a time when the market was basically flat. That performance did not come evenly, however, as even this model was not immune to the bear market of 2002. During that year and again in 2003 this model experienced declines of over 20% from it's prior valuation peaks. But even then it's valuation stayed well above what the SPY would have provided.
This trend following system appears to perform well on sector funds and the system is simple and straight forward. The 6-month returns can be calculated in a spreadsheet or can be obtained from this website.
If you have any questions or comments feel free to post them using the form below, or email me using the link at the bottom of the page. I plan to post a system using other etf groups in the future. If you have a preference of family or other suggestion let me know.
Disclaimers: Note that these results should be considered hypothetical, that the closing prices used may not have been achievable in real trading, and that no transaction fees are included. Also recognize that past performance does not necessarily predict future performance and therefore you should perform your own due diligence before following this or any other investment strategy.
// This script is part of the Comment Mgmt System // and contains several functions for support thereof. //Please set the following variables for your MySQL database: //Please set the following variables for your MySQL database: //it is highly recommended that you restrict access for the user for security reasons //it only needs "INSERT" privileges //for security, html is not allowed, so bbcode is used for formatting //START 3rd PARTY CODE: I did not write this /************************************************/ /* BBCode v1.0a */ /* Date: 03/2003 */ /* */ /* A simple and effective script that */ /* allows you to implement bbcode type */ /* behaviour on your php website. */ /* */ /* Contact: firstname.lastname@example.org */ /* */ /* Usage: */ /* */ /* Put the following line at the top of */ /* the page you want to have the bbocde */ /* in...(assumes both pages are in the */ /* folder */ /* */ /* include("bbCode.php"); */ /* */ /* Pass the text to the function: */ /* */ /* $mytext = BBCode("This is my BBCODE"); */ /* or */ /* $mytext = "This is my text"; */ /* $mytext = BBCode($mytext); */ /* */ /* echo $mytext; */ /* */ /************************************************/ ?>
Current Comments284 comments so far (post your own)
© 2005 H.M Todd