This post is the first in a series of posts documenting a momentum-based approach to trading asset class Exchange Traded Funds such as gold, real estate, equities, currencies, etc.
I am in the process of developing a momentum-based ranking system for asset class Exchange Traded Funds (ETFs). I now feel it is getting necessary to start documenting the testing and analysis in order to maintain some discipline in this activity. Otherwise I will be jumping around like a rabbit with no clear direction. I expect this activity to span multiple posts and I hope readers will bear with me for this exercise. I will be using tools provided by Portfolio123.
For my initial attempts, I have chosen gold (ETF Symbol: GLD), real estate (ETF Symbol: IYR), and bullish US Dollar (UUP). The ideal ranking system will ideally identify the best ETF from the three, with "best" meaning optimal profits post-buy. There are several reasons for choosing this combination of ETFs. First, all three have had significant up/down swings over the past 15 years, they have a low cross-correlation, and there is minimum bias for where the markets have been or where they may go.
By this I mean that interest rates have been trending down for the last 35 years. I don't want to develop a system that is biased towards falling interest rates when they will likely rise in the future. UUP will likely rise with interest rates, IYR will likely fall. GLD will possibly fall as the US Dollar rises but GLD has somewhat a mind of its own.
The downside for these initial tests is that UUP did not start trading until Late February 2007. IYR started trading in mid 2000. I am going to make use of the full data range from 2000- present even though results that include UUP won't kick in until sometime in 2007. The combined asset performance from late 2007 to present is shown below.
Of note is that there is an upward bias in the combined performance of the 3 ETFs since 2008. Also the combined drawdown in the 2008-2009 is quite minimal, on the order of 25%. Ultimately, this combined performance will be the measuring stick, or benchmark, for comparison with a 1-of-3 select system. Performance must be significantly greater than the benchmark before deploying a design.
Tomorrow I'll start getting my feet wet with some exploratory tests.