Using ETF Replay‘s relative strength backtest system an investor can backtest various user defined ETF portfolios. I decided to test a very basic 3 ETF system that looks at the relative strength return of GLD (Gold), SPY (S&P 500), and SHY (Barclays Low Duration Treasury ETF, a close substitute for cash). The strategy ranks the 3 ETFs based 40% on the 3 month return, 30% on the 20 day return, and 30% based on the 20 day volatility. The backtest started in 2005 (when GLD started trading) and rebalanced monthly. No commissions, slippage or taxes are assumed. The benchmark I used was buying and holding SPY.
The returns for a relative strength system that rotates between SPY, GLD, and SHY (cash) have been impressive over the past 5 years. The total return to date is 90.6% vs -1.1% for SPY. More impressive, the volatility for the system is 14.8% versus 23.7% for SPY.
The system made 37 changes or “trades” over the past 5+ years. GLD was held 34.8% of the time, SHY 37.3%, and SPY 27.9%. The current signal on April 30th was for GLD, switching from April’s position SPY.
There are a number of variations an investor could use with the system as well as different ETF portfolios. For those interested in reading about relative strength ETF investing, there is also a new book available by Leslie Masonson Buy–DON’T Hold: Investing with ETFs Using Relative Strength to Increase Returns with Less Risk
No positions
I just found your site from the seeking alpha link. I was wondering if you could elaborate on your buy and sell signals from the rsi.
Thank You!
Thanks for checking out the site. I would suggest visiting etfreplay.com for a more detailed explanation on the relative strength system. Click on the 'watch video' link at http://www.etfreplay.com/backtest_rs.aspx
The system I wrote about buys only the highest ranked ETF and holds it for the specified period (1 month). However, there are variations one can add.
I admit I'm dense when it comes to this stuff, but I don't see where the current months pick can be found in this relative strength system.
Can you explain?
thanks
The current pick using the example I ran is GLD but you can run your own screens and set your own parameters here: http://www.etfreplay.com/backtest_rs.aspx
I have the same problem as "Anonymous": the back testing tells me where I should have invested a month ago but not where I have to invest today.
Is there a way to find that out through "etfplay" or elsewhere?
Thanks
Scott – question – is there good background/paper that describes this strategy – taking a combination of returns, volatility, etc to come up with signal? i don't quite get it but maybe I'm just too new to RSI
i watched the video -but nothing there really other than how to do it using their tool
As far as investing today, I would just change the parameter to 'weekly' and that will get you close and then re-check it after the last day of the month if you want to do a monthly rebalance. I would suggest contacting ETFreplay directly with any follow-up questions. Regarding a paper on this specific strategy, none that I am aware of but there are similar strategies detailed in other areas which I will post now.
Hi guys,
RSI stands for Relative Strength Index — this is NOT what the ETFreplay.com backtest does.
RSI is an indicator you can run off any one ETF — our frameworks needs multiple ETFs to determine which of the chosen ETFs is showing the strongest strength.
You can see what it is signalling any day by using the ETF screener. The ETF screener shows the details of the model. You can choose multiple ETFs and save them to a portfolio and view this for any day you would like, including the current day.
The backtest application takes the methodology of the ETF screener and runs reports based on what the Screener said to buy at each month-end. It then conveniently runs a report using the subsequent period.
If you look at the very bottom of the Backtest report, whatever is listed last is the ETF that it is currently 'in'.
It is very important to understand the link between the ETF screener and the backtesting applications. Once you can understand what the ETF screener is doing then it will become obvious what the backtest applications are doing. If you don't understand the link between the two, then you won't understand how it works.
I will do a new video explanation this weekend to explain this link.
chris
ETFreplay.com
I think it is marvelous that questions asked on this subject are answered right away! I really appreciate that.
Just a comment and a suggestion: Although I now understand how to use the back testing as a recommendation of what to buy right now, it would be helpful if that was stated on the screen. Just a thought. And many thanks
Hi Anon,
We uploaded a new video that should help explain it. The answer lies in the link between the ETF Screener and the ETF Backtesting applications.
The ETF Screener for your own portfolio lists will update daily — as the backtest application runs off the model of the ETF Screener.
Watch the video for a full explanation:
http://tinyurl.com/35c8p2y
chris
ETFreplay.com
I notice you can put in two time periods for return and one time period for volatility. And you weight the three choices to 100.
I think I understand weighting over multiple periods on the basis of return. But how does throwing in a volatility factor work in the weighting.
If you weight returnA by say 33%, returnB by 33%, and volatility for period X by 34%, does it weight average returns and volatility in the same direction? ie ReturnA x .33, ReturnB x.33, and std dev period X by .34?
Hi Scott,
Thanks for your detailed description of your adventures with ETFreplay – I’ve been looking at them for a few weeks too.
What other books would you recommend for your ETF trading system?
Hi, I have a recommended book list at the top of the site, I’ll give it some thought and perhaps update that list in the near future. Thanks,