Using Leveraged ETFs With A Market Timing System: SPY, SSO, SDS, SPXL, SPXS

by: ETF ProTrader

The last decade has seen an explosion in Exchange Traded Fund (ETF) products. Among the most popular are ETFs that mimic the performance of major indexes; SPY for the S&P500, QQQ for the NASDAQ-100, and IWM for the Russell 2000 among others. Recently there has been considerable expansion in the ETF world with the addition of both leveraged and inverse leveraged ETFs, that purport to deliver the daily price movement of 2X, 3X, -2X, or -3X the underlying index. This article will focus on some of these products that are built around popular indexes like the S&P500, NASDAQ-100, and Russell 2000. These leveraged, and in particular these inverse leveraged ETF products can be described as highly speculative and risky, but used efficiently and effectively can actually yield some surprising results. First let's look at what's currently available and their inception dates:

1x 2x 3x -1x -2X -3X
1/1993 6/2006 11/2008 6/2006 7/2006 11/2008
3/1999 6/2006 2/2010 6/2006 7/2006 2/2010
5/2000 1/2007 11/2008 1/2007 1/2007 11/2008
2/2010 2/2010

Now let's look at the performance of several of these products over their limited lifespans. The graphs below detail the performance of the S&P500 series of leveraged ETFs where the y-axis is the Adjusted Closing Price gain/loss and cumulative percent variation between SPY and the leveraged product. So, for instance, let's suppose that a given day is an up day for the S&P 500 and the 1X ETF SPY gains 0.15% for the day. We would expect, then, that the 2X leveraged ETF SSO would gain 0.30% for the day and the 3X leveraged ETF SPXL would gain 0.45% for the day. Likewise, we would also expect the 2X inverse leveraged ETF SDS to lose 0.30% and the 3X inverse leveraged ETF SPXS to lose 0.45%. Plotted below is the cumulative daily percentage variation between a given leveraged or inverse leveraged ETF and its expected result. To make this a little easier and to also account for dividends and the occasionally split or reverse split, we'll use the Adjusted Closing Prices as tabulated by Yahoo! Finance. The blue line is SSO AdjClose, the light grey line is SPY AdjClose and the red line is the cumulative percent daily variation. Note that a theoretically "perfect" ETF would have the red line be zero throughout.

Graph 1 SSO (2X S&P500 ETF)

All of the charts were created by the author.

Overall, this leveraged ETF has a cumulative percent daily variation of -25.87% for the period Jun 2006 - Feb 2014. Owning SSO for the life of the ETF would have resulted in a gain of 121.54% vs 73.70% for SPY, significantly less than the 2X gain one might expect. However, if you bought SSO at the bottom of the last bear market on Mar 9, 2009 it was trading at $14.06 (adjusted) and closed February 14th, 2014 at $101.33 while SPY traded at $61.32 (adjusted) on Mar 9, 2009 and closed February 14th, 2014 at $184.02.

Next, how did the 2X inverse SPY ETF ((NYSEARCA:SDS)) perform over roughly the same period?

Graph 2 SDS (-2X S&P500 ETF)

Interestingly, the -2X Inverse SPY ETF SDS actually does quite well with a positive cumulative percent daily variation of +18.24% for the Jul 2006-Feb 2014 period. However, owning this ETF for the entire period would be a disaster.

Next let's look at the newest 3X S&P500 ETFs.

Graph 3 SPXL (3X S&P500 ETF)

Graph 4 SPXS (-3X S&P500 ETF)

The cumulative percent daily variation for SPXL for the period Nov 2008-Feb 2014 was -13.56% and for SPXS it was -25.08%.

As one can quickly see, owning an inverse leveraged ETF in a bull market can be a very painful trade, and your losses could be a great deal more than 2X or 3X depending on your entry point. On the other hand, SDS was actually performing better than -2X up until the bottom in March, 2009.

So what have we learned? If you get the direction correct and you use leveraged ETFs you can actually expect to see better than expected gains; however you better be right most of the time because you will pay a heavy price when the trade goes against you.

Over the last year we've developed a series of ETF timing models for a number of index ETF Pairs including the S&P500, NASDAQ 100, and Russell 2000. Each has been backtested to 1990 (24 years) and passed a number of important tests, including independent timeframe tests. While they are quite effective the current state of the art is a daily, closing price timing model that gets the direction correct about 54% of the time and that coupled with about a 0.10% positive difference between the average daily gain and the average daily loss generates a very nice multiple vs the index modeled. One thing we clearly wanted to determine was how well will the models work in conjunction with leveraged and inverse leveraged ETFs?

The following graphs show the performance of the underlying SPX index (green line), the model's theoretical performance using a "perfect" +1 and -1 index ETF (blue line) and the simulated closing price performance using a leveraged/inverse leveraged ETF pair (red line).

Graph 5 SPX vs SPX Model vs SSO/SDS

Graph 6 SPX vs SPX Model vs SPXL/SPXS

Performance Summary

SPX Gain/Loss

SPX Model Long ETF Short ETF Leveraged Model
2X (Jul 2006 - Feb 2014) +58.2% +266.8%



4.58X 7.21X
3X (Nov 2008 - Feb 2014) +93.2% +135.6% SPXL SPXS +337.8%
1.45X 3.62X

Disclosure: I am short SPY, QQQ. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.

Additional disclosure: I am long IWM