Core Idea - Outperform the market by systematically switching between the underlying market (say SP500) and bonds (say US Long term treasuries) following objective rules. Note: This is not pairs trading where you buy one and short the other. Here we are just doing switching/rotation.
Pair Combo's:
This out-performance is not specific to S&P500 i.e., noticed similar out-performance for other markets like Russell 2000 and MSCI EAFE. Results for all three indexes are available at the end of the post.
Caveats - Results are friction less (i.e., no commission, no slippage). I am not sure if it makes that much difference though as the system is long term in nature. Another caveat is the duration of test data. The test covers last 10 years i.e., 2002-Current.
I would like to do a test for longer duration though. I have multi-decade data for S&P 500 but don't have data beyond 2003 for bonds though. If you know/have the data and would like to share then can you please drop me an email? Now to the system...
- SPY (S&P 500) <-------------------> TLT (Barclay 20+ year US treasuries)
- IWM (Russel 2000) <--------------------> TLT (Barclay 20+ year US treasuries)
- EFA (MSCI EAFE) <--------------------> TLT (Barclay 20+ year US treasuries)
Rule - If index performance over last 13 weeks is better than bonds then close out any open bond position and buy the index etf (i.e., SPY) @ tomorrow open. On other hand, if bond performance is better then index then close out any open index position and buy the bond etf (i.e., TLT) @ tomorrow open.
Following images provide the performance summary of applying this rotation for the above mentioned 3 markets. Further details are on the chart itself.
Question to readers:
- It seems to me this technique may be new to retail traders but probably not for professionals in finance industry. So I wonder why then financial planners or professionals don't recommend this approach for long term investments. I am curious to hear your thoughts on what problems do you see with this technique?
- I am interested in hearing your feedback on the studies published so far - like do you find them useful? Are there any specific studies you would like to see? etc. For any new study requests I will give a shot to do them in subsequent weeks if my skill and time permits.
5 comments:
This is a sensible strategy but in case there's a real crisis like in Spain currently, there's nowhere to hide:
http://timelyportfolio.blogspot.fr/2012/07/the-failure-of-asset-allocation-bonds.html
Hi EasyTrader - Thanks. Good point. I am not sure about failure of asset allocation though. It seems the problem both with this study and the spain example is reliance on static correlation for diversification. It might be interesting to see how the results would be if there is a 3 way choice i.e., portfolio switches to bonds if low correlation but otherwise switches to cash.
nice post, I was thinking about similar strategies. Are the backtest done in R and are the data total return? I've got quite a bit of data going back to mid 70s for boths, monthly thou.
m
Hi Michael,
Thanks. Nice posts and charts. I have added your blog to my reader.
Currently I do backtests in C# (NinjaTrader), then export as CSV file and do Performance Analytics in R. I started looking into R just recently. I don't know if data is total return. My data feed is Kinetic. I have to check.
I have posted a follow up study on this topic that takes dynamic correlations into account. I suspect that study might be more robust than this study as in 1990s bonds and stocks were more correlated.
Is is possible to get higher resolution copies of the graphs?
Post a Comment