World Beta Blog

Asset Class Rotation

February 5th, 2009 by Mebane Faber

Comments

I was at dinner the other nite and someone mentioned Decision Moose, so I thought I would do an update on a very simple system. I have written a lot on World Beta about cross market momentum in the past (ie asset class rotation). Some previous posts are listed at the end of this blog.

Anyways, below are some log equity curves for a simple rotation system. The same five asset classes as my paper back to 1973. Ranked on the average of the trailing 3, 6, and 12 month total returns. Rebalanced monthly. Top 1 is all-in the top asset class, Top 2 is all-in split 50/50 between the top two asset classes, Top 3 is all-in split between the top three asset classes, etc. Results are frictionless. Top1 would have done around +15% last year, Top 2 -9%, and Top 3 -24%. I think you could easily expand this to more asset classes but you would have to increase the number of holdings in line with the %ages above (ie Top 2 40% of the investable universe).

Again, I think this just captures the alpha of momentum and increases your Sharpe to the .80 area. (Asset classes cluster around 0.20 and buy and hold around 0.40.) There are some tweaks and improvements but this is a nice example of parameter stability – and they beat buy and hold roughly 75% of the time (and have continued to work in recent years).

Some rotation funds of various flavors.

FUNDX (Newsletter here), (Fact Sheet here)
Rydex International Rotation Fund (RYFHX), (Fact Sheet here)
Rydex Sector Rotation Fund (RYSRX), (Fact Sheet here)
DWA Technical Leaders (PDP), (Fact Sheet here)
DWA Balanced (DWAFX, DWAFX), (Fact Sheet here)
VL Industry Rotation (PYH) (Fact Sheet here)
VL Timeliness (PIV) (Fact Sheet here)
Claymore/Zack’s Country Rotation (CRO) (Fact Sheet here)
Claymore/Zack’s Sector Rotation (XRO) (Fact Sheet here)
BlackStar (study here)

Previous Posts:

You Spin Me Round

Alpha Persistence

Cross Market Mo (and here)

More on this topic (What's this?)

Dividend Stocks Role In The Future Recovery (Dividends Value, 1/14/09)

Estimating Beta-Based Expected Returns (Dividend Tree, 2/18/09)

Read more on Assets, Beta at Wikinvest

Damian

I use a similar approach in my MOMO1 strategy - takes top two assets from a group of about 80 ETFs. Worked well until this December - so I'm in the process of making modifications. DecisionMoose is great.

http://skillanalytics.wordpress.com/strategies/...

WorldBeta

Yeah, even taking the top 15 out of 80 is just like taking the Top 1 out of 5. Expect some fireworks!

Nice returns though...

ttocs

nice work.

oyz79

Hey Mebane, I'm just wondering if you have the data available for the returns of buy & hold vs. momo from 1932-1962. I read a fairly elementary article on seeking alpha from indexuniverse saying that 'now is the time for buy&hold' and their only real evidence was that from 1932-1962 buy & hold returned around 12.9% CAG. I'd be curious to see how the timing model compared during that time period. All in all I was fairly disappointed with their misrepresentation of Tom Lydon.

WorldBeta

You need to be more specific by what you mean regarding momo. Do you mean the timing model? I imagine B&H outperformed.

oyz79

sorry, looking for the timing model (s&p) returns/deviations in each year post 1931. Not sure if you have this information available?

J Sam

Has anyone looked at a variation using say the Top 2 with the caveat that both must be above 200 Day moving average or 10 Month SMA (If only one is above use that one). I would love to see numbers for this. Probably reduces returns slightly but reduces volatility by alot

Red Hue

I think it would be interesting to see how a model of 70% Top2(3) and 30% Long term Bonds
would fair?

WorldBeta

Taking the top 2 with 30% in 10 yr bonds (i dont have 30 yr data) results in stats very similar to taking the top 3, with lower drawdown of course.

Red Hue

Hi Mebane,

Interesting...do you have on hand how much lower drawdown? Probably not nearly as low
as your GTAA strategy eh?

WorldBeta

-20% but the Sharpe is similar.

Red Hue

Thanks Mebane....
...I like the new site by the way!! Looking forward to what you'll have
on your "Timing Model" tab.

Red

Patrick

Comparing anything to buy and hold using frictionless data is suspect. Unless you can account for trading costs, the above chart is worthless.

WorldBeta

Sounds like a response from someone who bought and held last year ;) Just kidding.

Well, I don't include them because it varies depending on your situation. But, plenty of people could be trading this with zero friction and tax free.

Brokerage Cost - Varies from zero (FolioFn, lots of mutual funds) to $5-$20 a trade....If I remember correctly the Top 3 has about 130% turnover per year. Obviously the Top 1 would be much higher, I think around 300%. (All that means is you are making three trades per year which isn't that much when you think about it.)

Bid/Ask Spread - Zero for mutual funds, varies for ETFs.

Taxes - Zero for tax exempt, varies for taxable.

Steveal

I took a quick look at the funds you mentioned. They are all very much the same over the past year. They all lost ~50%.
How did they get such bad results when, apparently, using a winning system?

WorldBeta

I have no idea exactly how most of these are run - but most are confined to a single asset class so they aren't really that similar to the above system. . .no diversification benefits.

Kostas

Hey Meb, the MaxDD for the SP-500 in your table looks like it does not include the 2007-2008 span... are those updated numbers with 2008?
Kostas

WorldBeta

I think they're right...monthly resolution data though. . .not daily.

ashpak001

It would be interesting to see result for Long-Short strategy where you would every month go long the top asset class and siumultaneously go short the bottom one...

MarkM

Mebane-

Could we classify the inverse of your 5 as separate asset classes and expand the strategy to include those as possible longs? What would that do? Anybody running anything similar?

MarkM

FredCDobbs

It would be interesting to see the results based on weekly rebalancing ala Decision Moose using say 26 weeks as your evaluation period. Moose prefers to hold only the top asset but it would be helpful to see if this is really ideal from a risk adjusted point of view.

oyz79

For those interested in seeing how momentum strategies can be combined with moving averages, see a recent article on CXOAG: http://www.cxoadvisory.com/blog/internal/blog5-...

Their conclusion: "In summary, a very limited test suggests that adding simple moving average signals to asset class momentum investing may enhance returns."

blog comments powered by Disqus

Get the Latest from World Beta

Subscribe to Newsletter

Find Me On…

My Book

Ivy Portfolio Book by Mebane Faber

Search

Archives

My Web Sites

LinkFests

Data Resources

Blogroll

Institutional Research

Disclaimer

The opinions expressed on this site are those solely of Mebane Faber and do not necessarily represent those of Cambria Investment Management, Inc. (“CIMI”). This website is made available for educational and entertainment purposes only. Mr. Faber is an Investment Adviser Representative of CIMI, an investment adviser registered with the states of California and Utah. This website is for informational purposes only and does not constitute a complete description of the investment services or performance of CIMI. Nothing on this website should be interpreted to state or imply that past results are an indication of future performance. A copy of CIMI’s Part II of Form ADV and privacy policy is available upon request. This website is in no way a solicitation or an offer to sell securities or investment advisory services. Mr. Faber and CIMI disclaim responsibility for updating information. In addition, Mr. Faber and CIMI disclaim responsibility for third-party content, including information accessed through hyperlinks. Faber receives affiliate commissions through Amazon and AlphaClone.

© 2009 Mebane Faber. All Rights Reserved. | About | Contact