Lessons from Mebane Faber and his Tactical Asset Allocation
Long-time readers of this blog will know what I think about technical analysis. The short answer is: not very much. There are thousands of ‘techniques’ for technical analysis and no widely agreed upon rules that could be properly studied by scholars to find out if indeed they add value or not. Therefore, in my opinion, technical analysis (TA) is little more than shamanism. As of this date, there does not seem to be any scientific evidence of its benefit, except for data showing that the use of moving averages and relative strength does add value. These two particular tools are, in my opinion, part of the well documented momentum effect, and therefore not part of the shamanistic TA pattern-seekers.
So trend-following, also called momentum, does have scientific support. In his wonderfully rigorous book on What Works on Wall Street, O’Shaughnessy describes a strategy combining both value and momentum called “Trending Value”. This strategy, combined with the Piotroski score as a quality filter for stocks, is the main chunk of my personal portfolio.
What exactly is momentum? In short, it’s a tendency of stocks to keep on rising if they’re currently rising, and to keep on falling if they’re currently falling. However, the difference between rigorous momentum use and the kind of TA investors that I’ve criticised is that momentum can be quantified with relative strength or a moving average, whereas the TA and its imaginary patterns and arbitrarily drawn trend lines are completely outside the scope of research.
Mebane Faber wrote a paper on a system using moving averages with whole asset classes. His system basically is:
– setup a portfolio of a few ETFs (5, 10, 15, however many you like)
– buy every ETF if its price is above the moving average
– sell the ETF if it drops below the moving average (and stay in cash until you get the signal to buy again)
The system is updated on a monthly basis, so it requires little intervention, and it seems to have performed admirably in the past 40 years (CAGR 10.48%, Sharpe 0.73) as you can see here in his backtest from 1973-2011, compared to just buying and holding the same assets (CAGR 9.92%, Sharpe 0.44). The Sharpe ratio basically doubles.
Here’s the data:
So it is definitely true that this system can improve your returns with high probability. But there are a few caveats. First, Faber works for a company that launched a fund based on this Global Tactical Asset Allocation (GTAA) and that fund has underperformed for the last few years, even though there has been a huge bull market in stocks, so the comparison to the SP500 is not very favourable:
This can be a few explanations for this. First, as a strategy supposed to guard against big drops, this system will shine mostly when there is big volatility in the market and not when there’s a huge bull market. Since it does not allocate 100% of holdings to stocks, it cannot beat the SP500 in a bull market for stocks. Secondly, while Faber’s paper outlines a portfolio holding around 5 – 15 ETFs, the GTAA fund uses many more instruments. The top 5 holdings by weight represent 37% of the total portfolio, but after those bigger positions there are many 1- 2% weightings, which means the portfolio may easily hold more than 50 different positions, which is a lot more than were used in the paper.
No matter why the fund is performing dismally, any strategy can under-perform for 3 or even 5 years. Good strategies do not do so often, but they still do. Look at the first GTAA chart. You can see that in the year 2000, the blue line is ahead of the red one for the last time: so if you had stuck to the strategy from 1973 to 2000, a full 27 years, you’d still be under-performing a simple buy and hold portfolio! No matter if the next few years would have been spectacularly good to you. That’s what makes strategies difficult to stick with, and why backtests over 10 years may look great but are difficult to replicate in real life. It’s very difficult to stick with something that hasn’t worked in the past 5 years. Honestly, would you buy the GTAA fund right now? I wouldn’t. So look at this as a wake-up call regarding (new) strategies and their possibly extremely good looking returns. They may really work, but they may also not work, or not work for a few years (making them almost impossible to stick with) or they may not work for human beings (because their risk, or the work needed to implement them, maybe too high a price to pay for most of us).
Always ask yourself: how practical is this strategy, really? Can it really be implemented by myself, or is it too much work, too complicated, too unrewarding? And of course: how likely am I to stick with it?
GTAA may well be a good, logical strategy that out-performs over the long run, but at least for now it does not look like something you would want to invest in. And most strategies, if not all, are that way. So ask yourself ahead of time how committed you are to it, and how much you want to risk. Don’t put all your eggs in one strategy basket, ever. It’s much easier to stick to a strategy if it’s only 30% of your portfolio than if it represents all of it.
Update January 2015: I’ve decided to test a 5 ETF portfolio (among other strategies) myself in 2015 and post monthly updates. The only difference to the GTAA described by Mebane Faber is that it doesn’t buy every ETF above its moving average but instead allocates the portfolio in order to achieve minimum variance.
 Technical analysis is, to explain it simply, a system using stock price charts to predict future movements with imaginary lines and patterns like the “head-and-shoulders” pattern.
Technical analysis – Wikipedia, the free encyclopedia.” 2003. 25 Nov. 2014 <http://en.wikipedia.org/wiki/Technical_analysis> : “Whether technical analysis actually works is a matter of controversy.”
 Wong, WK. “How rewarding is technical analysis? Evidence from …” 2010. : “The results indicate that in general, single moving averages produce the best results, followed by the dual moving average and the relative strength index using the ‘50 crossover’ method.”
 “The Piotroski Score — The Graham Investor.” 2009. 25 Nov. 2014
 Faber, Mebane T. “A quantitative approach to tactical asset allocation.” The Journal of Wealth Management, Spring (2007).