Update: If you like World Beta, please consider nominating us for the Loeb Awards in blogging and book categories!
I few years ago I wrote and published a paper to avoid taking the CMT Level III Exam. The timing was fortuitous as it preceded the global meltdown by a few years and illustrated the benefit of diversification as well as using some simple risk management to sit out the simultaneous bear markets. The real time performance has been far better than the historical numbers, and the system hit new highs at the end of 2009 while the benchmark buy and hold and the S&P500 are still down 30% or so. This is one of the reasons the paper is now in the top 5 in all time downloads on the SSRN. (Had you told me 10 years ago that I would write a popular academic paper, after bursting out laughing, I would have assumed it was in gene therapy…funny where life takes you!)
We tried to expand on the paper a bit in the book, but there are still a lot of common misconceptions about the timing model (it works all the time, it is fail-safe, etc neither of which are true. Look at January a crappy month for timing model as an example.), but we think it was a simple example of proactive risk management. We don’t run it remotely like this in-house, but it is a useful description of “how to fish” rather than “where the fish are”.
Below is an updated equity curve and chart of the timing model since 1973. We are considering writing an update to the model with new additions including a section on global rotation strategies, as well as adding more asset classes (small caps, gold, etc).
The problem is I literally have 7 more papers to write on some really, really cool topics. So, I’m not sure exactly what will come first. I thought about posting the 7 topics and letting people vote, but I’m not so sure I want to disclose all of them yet….
Cambria is considering starting a newsletter or perhaps charging for some of the research to:
1. Give investors an early look at our research, with maybe a 3, 6, or 12 month time lag before we publish the paper publicly for free (it at all).
2. Update a number of our strategies in depth. I get dozens of emails every day about people wanting to know specifics, how we manage money (proprietary), and how to update the portfolios. I had hoped that providing links on the website and publishing a free monthly update would have accomplished this, but it only generates more emails. Not sure what the solution is here.
3. Circulate our investment research.
So, if you have any thoughts or suggestions on paper topics, publication methods, etc let me know in the comments.
1973-2009
And the real time results.
2006-2009
So these charts are for all of 2009? If so, what were the number for just 2009?
More data generates more questions LOL!
I think that circulating selected pre-publication research might play a role in increasing the influence of your firm's research arm (grin). Here is a very interesting book that explores the value of selling copyrighted material versus open distribution: “Against Intellectual Monopoly”. The book is full of interesting case studies. http://www.dklevine.com/general/intellectual/ag...
Getting more emails per month is a Good Thing. I am sure that you have some time saving systems in place, but perhaps there is more you could do to decrease your emailing time an increase the AUM, citations, blog mentions, what-have-you that you can glean from your “warm pool” of email correspondence. I have heard InfusionSoft recommended as one example tool, just don't be evil ;-.
OT: Collective2.com seems to be rocking http://www.collective2.com/cgi-perl/t200.mpl
Well, that is until it become unmanageable and becomes an opportunity cost for doing other things (research). I take pride in answering all my emails but it is already overwhelming…
Meb, seems there is alot of interest in more detail about how you combine the research from Ivy Portfolio and AlphaClone to manage money at Cambria, me included. I have been “reinventing the wheel” for months now trying to combine your QTAA paper, Sector Rotation, and 10-20 names from AlphaClone into a comprehensive equity approach.
I bet that over the last 20 years the 5-factor QTAA paper alone beat 95% of the “top” investment newsletters ranked by Hulbert Digest, with less risk than 100% of them. Last time I checked … even the “best” newsletters per Hulbert were mediocre and probably barely beat the SP500 net of fees and taxes. See http://money.cnn.com/2008/03/26/pf/funds/Ask_th....
In my 15-year career … before your QTAA paper.I have never come across a mechanical risk-managment strategy using technical analysis that a non-CMT could really understand and apply. Until then, I was a “buy, pray and rebalance” investor b/c I couldnt dispute the “miss the best 10 days, 20 days, etc. and you're screwed” argument.
Your QTAA paper caused me to dig deeper and find out that the “best 10 days and WORST 10 days” usually happen below the 200-SMA … and if you miss BOTH … you have better returns & lower risk that B&H. It doesnt seem that the mainstream has figured this out yet, and the big firms will never endorse that on an “official” basis b/c it would mean going to cash/bonds for months at a time.
My suggestion would be (1) check out the “best” newsletters per Hulbert however long he goes back, for performance and pricing, (2) develop a newsletter, subscription or pay-per-research download website to steer your blog followers and emailers to, (3) charge a fair/premium price based on what other so-called “top” newsletters charge, and (4) use the extra revenue to hire a top Stanford,Wharton, Harvard, etc. MBA or PhD as an intern to help you with your research; basic analysis, number-crunching, etc.
That way you can focus on writing up and “polishing” the research and turning it into a paper (e.g., $400/hr work), and delegate the “grunt” work (e.g., $40/hr work). You could hire them for a 1-year internship to leverage your time, avoid a long-term commitment, give yourself time to evaluate them, and keep the compensation reasonable. Lots of “rocket scientists” beating the bushes now I would think that would line up for that job.
I do not believe you would cannibalize Cambria, HNW and institutional investors by and large will never be “do it yourself” investors. Most are still blindly overpaying managers who are lucky to beat their index by 100bps over 10-years and/or indexing with a B&H/rebalance strategy. They will be creamed again someday; just like in 2008. Charging for a newsletter or your research in my view would leverage your intellectual capital. You'd probably beat a fair number of hedge fund managers out there charging 2/20 for their time.
My 2 cents …
why the hell would you give it away for nothing? is it of no value? charge a nominal fee. those who recommend otherwise are too parsimomious to pay for it.
I've read and enjoyed your book, research paper, and the always educational and enlightening posts on this blog. Perhaps even more important, your collective wisdom and research has convinced me to ease out of active stock picking and embrace the lower stress method of TAA portfolio management you advocate so well. A reasonably priced newsletter or occasional reports updating the model would be welcome, and I would be happy to pay for those. Jay's suggestion of hiring a talented underling to do the basic work while you “polish and refine” it, makes sense. After all, none of us has more than 24 hours in each day, so using each hour effectively is essential.
Whatever decision you make, keep the good work coming. And keep in mind, most blog control panels do have an option for simply turning off comments if it becomes too overwhelming !
The book (and earlier paper) were quick and enjoyable reads. Didn't put me to sleep like many I've read. I have some critical comments about following the strategy, however, for any but very long time horizons (much greater than 10 years).
As more people start following momentum strategies and using metrics such as the 10mo or 200-day SMA as triggers for buying and selling, it is inevitable that the strategies will lose value. One of the issues with the timing methods apparent in the paper and in the book is the occurrence of “whipsaws,” in which one sells one month only to buy back the next, or vice versa. As more people follow these strategies, I predict we will see an increase in whipsaws and more rapid transitions near the trigger points (attempts at “front-running” could make it worse.) Some market time periods, by their “flattish” and volatile nature, produce inferior returns to buy and hold.
I have a Excel spreadsheet which agrees well Mr. Faber's results over the time period he used in the book. It is also set up to test any time period for which data exists (presently back to 1950 for S&P close data and 1970 for S&P “total return” data.) Some periods are very good. Others are not so good. For example, during the period from the beginning of 1970 through the end of 1981 (a period in which the strategy generated many whipsaws), the CAGR results using S&P total return data and 3-month T-bills at the appropriate rate for each month are: B&H: 7.47% 10-mo Timing: 7.91%. That's still positive, and the lower volatility of the timing method is an added bonus. But a later period with many whipsaws is much worse. For example, the decade from 1984 through 1994: B&H: 15.37% 10-mo Timing: 9.78%. [Please post if you find errors.]
My opinion is that you should charge a nominal fee for the information. This should dramatically reduce email load, and possibly also the popular use of any enhancements made to the model, hopefully extending their usefulness. However, since most of the methods described in the paper and book are fairly straightforward to duplicate on one's own, I do not think there would be many subscribers to a newsletter which charged on the same order as the newsletters which employ a more active approach ($150-200 per year is far too high, in my opinion.)
Correction to the time period stated above. “decade from 1984 through 1994″ should be “decade from 1984 to 1994″, i.e. 1/1/1984 through 12/31/1993.
How about an ETF/ETN? That'd be great
I prefer to compare the TAA returns to investor returns, not benchmark returns. Investors who “buy and hold” do not get buy and hold returns. Because they don't really buy and hold. They listen to mutual fund and annuity salesmen who always offer the “latest and greatest”. Or they attempt to hold but bail at the worst possible times and then wait to get back in after missing most of a rally. Compare the TAA real time returns to data from the QAIB for a more realistic comparison.
Wow, good stuff here by everybody.
I would be interested in Meb's response to Dave D's thoughts. My own two cents is that I am significantly less skeptical than Dave D. I believe the basic TAA model will continue to achieve superior risk adjusted returns going forward. Why? I believe the folks that could benefit most from adding all or a portion of Meb's approach will be the least likely to adopt it. The two worst performing groups in my opinion that would benefit most are the retail do-it-yourself investor and the closet asset gatherer, pseudo-investment advisor group that do nothing but swap clients back and forth and sell crap. Neither has any interest in becoming a better investor. Even some of the more sophisticated, large AUM, independent advisory firms, have woeful performance numbers. Those who are in a position to promote wide spread adoption of the approach aren't going to do it. That would be a radical shift from bitching about politics, deficits, and other economic non-sense. Those folks, many Meb link to, are far more concerned about being heard and promoting an agenda than making their readers money. Those perhaps are the most arrogant of the bunch. Change or assimilating new information will not happen for them.
Also, we are not going to see Buffet, Berkowitz, and the other AlphaClone tracked guys suddenly adopt a meaningful TAA based approach. Likewise with the wire house brokerage firms. Not going to happen. I am sure we may see some funny stuff in certain asset classes around the 200 day MA at the end of the month, but on the margin it will prove to be uneventful.
I am more convinced of this now that I am re-reading Reminiscence of a Stock Operator that human's simply have not changed and will not.
To the larger point of the thread. I agree with Jay's comments. A little bit more insight into how you apply your research work at Cambria. I respect your desire to keep whatever you think needs to be kept on a proprietary level, but if you let the cat out of the bag a little up to your comfort level I believe you could monetize it and it could add value for us serious money managers. Even with that, each of us is going to use whatever info is actionable in different ways. A rotation system may work for some and for others it may not be a fit. As Richard Dennis said, he believed he could print his trading rules in the NYT and nobody would be able to follow them on their own. I believe this holds true for anybody's system.
I much more interested in what the research shows, for instance by adding various sub-asset classes. Something like, how would the model work if I break the commodity sector up or if I add UUP or if I add emerging market bonds etc. How about individual stocks etc.
I hope that helps a bit.
Good point.
Good points here, too. I will concede that, because the timing method requires less active management by “professionals”, greed might actually prevent the widespread use of such a system. I theorized that a 10 month timing interval, because it is slightly longer than the widely-followed 200-day SMA, might produce inferior returns than say, a 9-month timing interval. I tested this in my spreadsheet, but have not found the results to be consistent enough to validate this theory (perhaps I should test 8 months).
However, the lack of validation might lend support to your assertion: that the system is not currently, and may never be, widely followed. After all, as Mr. Faber has acknowledged, the timing system is not new (though his interpretation and excellent description of it is). Surely there has been more than enough time for adoption and any possible market adaptation. Then again, prospective adopters may have back-tested the system over different intervals, as I did, and come to the same conclusion: given a short time period (less than 10-15 years) the timing system may actually produce far worse returns than buy and hold (notwithstanding Tommy Sikes' good point about “investor returns” vs. buy and hold.) I came into this exercise, expecting to give up some of the return of buy and hold, in exchange for timing's lower volatility. Given a long time horizon, the timing system seems to do just that. But over my shorter horizon (15 years), the trade-off seems far less certain.
I will also say that my findings for the decade beginning in 1984 are so far from what I expected that I plan to go back to my spreadsheet to see if there could be some error, or at least find out what in particular about that decade caused such under performance of the timing system.
I second that. I would love to be able to invest in a leveraged GTAA model done through ETN/CEF/OEF or some other model. I don't think an ETF would work because the daily portfolio disclosure would erase some of the protection against copycats.
This would especially help out those of us who do not meet the Accredited Investor threshold for whom Cambria is not even an option. Consider myself: I'm three years out of school as an engineer and made ~$90K last year pre-tax. If I managed to save 1/3 of my pre-tax gross each year and compound at 10% it would take me another decade to make the $1 million net worth minimum.
t's ironic that the Accredited investor limit basically requires you to learn how to DIY for many years until you have gained sufficient assets to turn it over to someone else. If you are good enough to DIY for that long, why would you then pay someone else to do it?
Have you tried adding in a strength weighting and see if that helps the returns without increasing the risk?
THe portfolio I built using similar stats as the original but adding in a strength rotation… It seems that if I decrease the 10month MA to 6months MA, my returns increase and my std deviation decrease, and drawdown decreases… This seems to begin at 7 months in order for all 3 to change(returns above 10month, std deviaiton below 10month, and drawdown below 10mont)…
But I wonder at 6month MA if I am chasing??
I like the idea of being able to pay a nominal fee to gain access to Cambria research and strategies, but not if the same information is going to be available later on in a publication. I understand the desire to publish, but it seems to me there should be a line between publication and fee-based investment ideas.
You might consider, as a value-add to the fee-based service, to allow access to databases you use in your research, such that your subscribers could expand on your research to fit their specific needs or interests. Of course if those databases are themselves fee-based or copyrighted, that wouldn't work.
does anyone know what assumed interest rate is for the published results for leveraged timing?
It is in the paper and book – the broker call rate.
one further point as regards prospective performance. from the lows of 1974 until sometime in 1998, the model UNDERPERFORMED the SP500. that's about 25 yrs of underperformance. that said, the SP was up about 16% acr during that period. the point is that this model will likely do fine over a long period because it will likely underperform a BH for sustained periods of time and people will tire of it.
Wow.
Ad shorts via futures and proper position sizing depending on vol – et voila, you are running a proper trend-following hedge fund.
Meb, I would echo Jay's comments above almost verbatim.
Jay, I would echo your comments completely.
I was also wondering what the 2009 numbers were, standalone.
Working backwards from the data in the paper and the 2006-2009 table above, I used Excel's XIRR function to approximate the standalone 2009 numbers:
2009 B&H: 18.83%
2009 Timing: 13.99%
Working backwards from the data in the paper and the 2006-2009 table above, I used Excel's XIRR function to approximate the standalone 2009 numbers:
2009 B&H: 18.83%
2009 Timing: 13.99%