In my last post I looked at using the change in trend of the unemployment rate as a market timing indicator. The results were impressive to say the least – almost a doubling of risk adjusted returns over buy and hold. In this post I want to make the analysis I did a little more real world by adding safe assets to the mix.
Let’s take the same analysis from the last post and add US gov’t bonds as the safe asset to switch into when the market timing indicator triggers an exit from US stocks. I’ll use real world ETFs. SHY for the the short term safe asset. IEF for the intermediate term safe asset and TLT for the long term safe asset. The only negative in doing this is that these safe asset ETFs don’t have a long history. They only go back to about August 2002. But there is still a good deal of insight to be gained in how these systems work in several quite distinct market environments. We can also do one more thing. We can combine the unemployment indicator with the traditional market timing indicators, SMA and absolute momentum. Below is the table with the results.
As in the last post, where the time period looked back to 1999, the SPY-UI indicator strategy outperforms both of the traditional timing strategies. By adding any of the safe asset ETFs we get even better performance with the best performance coming from switching into intermediate government bonds. This is consistent with other strategies – intermediate term bonds are the best safe asset to add to the mix.
Of course the obvious next question, is what happens when you combine a traditional timing systems with the new UI system. This is exactly what PhiloEcon did in the GTT system. The combination system works like this. As long as the UI indicator has not triggered, the system is long the risky asset. Once the UI indicator triggers then the system uses the traditional timing indicator (200 day SMA or 12 month abs momentum) to determine when to exit the risky asset. Once the system is out of the risky asset it is invested in the safe asset. This is what the last two lines of the above table show. The results are mixed, at least in this short timeframe analyzed. Combining the UI system with the 200 day SMA yields slightly better return but slightly lower risk adjusted returns. Using the UI system with the 12 mo abs return produces lower results.
In summary, adding a safe asset to the UI system produces even better results over the standalone system. And combining the UI system with traditional timing systems also produces impressive results. Way better than buy and hold to say the least. Now, you can make further extensions to the system. How about foreign risk assets? How about using momentum across risk assets combined with the UI system? That’s what we’ll look into in a future post. Stay tuned.
10 Comments
David McNamara · May 28, 2016 at 4:09 pm
I just love how you analyze these variables, backrest them, and present your findings in a way that the average person can put together a thoughtful portfolio. Fantastic! No doubt we need to be out of risk assets when both the UI indicator and either of the other tactical indicators flash red. However, haven’t we been in a period of declining interest rates for most of the period tested. Would a rising period of interest rates render the so called “safe” assets less so? Love your thoughtful posts! Thanks.
paul.novell@gmail.com · May 29, 2016 at 9:29 am
Probably not. Switching to the safe asset has worked historically very well in rising and falling rate environments. The switch to the same asset is more about avoiding the loss on the risk asset than the return you get from being in the safe asset.
Paul
Lynne · May 28, 2016 at 9:41 pm
Great stuff Paul! I’ve been doing some similar backtesting lately using IEF rather than SHY as the “cash” (safe) asset. I wasn’t able to do UI backtesting, but did some various ratios of SPY/IEF and best risk/returns seemed to be a timing model of 100% SPY with IEF as the the cash signal. Curiously though, for SPY during my testing period of 2004-16, an 8-mo SMA seemed to do better than the typical SMA200 for the timing. Have you found the same result? I also found that a mix of Bond ETFs (similar to your Bond Portfolio) did well with a 6-mo SMA timing model, and again using IEF as the cash asset. But with bond funds, I was not able to get a long enough backtest to include the ’07-’08 market. Were you ever able to simulate your Bond Portfolio for a longer backtest? Is a 6 mo timing model the sweeter spot for bonds (than SMA200)?
Greatly appreciate all your research. It has given me quite an education (and courage) to begin a few disciplined investing strategies. Much thanks!
paul.novell@gmail.com · May 29, 2016 at 9:32 am
Lynne, the optimal SMA period varies a lot across time. See this post. There is no one number that works across all time periods. And its important to use numbers that have been tested over long periods of time not just those periods we can test we these modern tools – which are quite limited in scope.
Paul
Zach · May 30, 2016 at 10:50 am
Paul,
Interesting post as usual! I am curious if you plan on tracking these signals and any strategies based on this system?
paul.novell@gmail.com · May 30, 2016 at 11:16 am
Hey Zach, currently thinking on that. I’m probably going to track the UI indicator on its own, and the SPY-IEF system based on the UI indicator. Only problem is that its a bit manual and I’m always hesitant about adding manual stuff to my maintenance list.
Paul
Tony · June 1, 2016 at 3:05 pm
Hi Paul,
One of the best applications I can think of is to use the simple system you outlined in an account like a 401k that doesn’t have a wide range of investment options (no factor or sector ETFs, individual stocks, etc.). Sometimes you only have a few basic choices in these accounts and this system would allow you to simply alternate between a stock fund and a bond fund based on the UI and net a good return while limiting drawdowns.
Looking forward to more info from you on how to best use the UI.
paul.novell@gmail.com · June 2, 2016 at 9:09 am
Good point Tony.
David · June 2, 2016 at 2:35 am
Hi Paul !
How do you manage to track 200 sma for the UI when the numbers comes monthly?
Is the UI data available only from FED ? I have searched but cant find any site who publish it in a way possible to follow easy.
David
paul.novell@gmail.com · June 2, 2016 at 9:17 am
David, its the 12 period SMA. Since the data is out monthly it ends up being a 12 month SMA. You can get the data from FRED but you need to run your own SMA. Stockcharts.com has an unemployment rate chart that you can modify the default SMAs and use. The ticker is $$UNEMPRATE.
Comments are closed.