The Failed Promise of Market Timing

I’ve long believed the most difficult part of being a Couch Potato investor is resisting temptation. Index investors are asked to be content with market returns, but they are bombarded daily by fund companies, advisors and market gurus who promise more.

Back in May 2012, I wrote about one of these enticing strategies, described in The Ivy Portfolio by Mebane Faber and Eric Richardson. The so-called Global Tactical Asset Allocation (GTAA) strategy grew out of Faber’s widely read research paper, A Quantitative Approach to Tactical Asset Allocation, first published in 2007. It begins with a diversified portfolio inspired by the Yale and Harvard endowment funds, combining traditional and alternative asset classes. The “tactical” part involves using market timing to move in and out of these asset classes based on 10-month moving averages.

Faber updated the paper in early 2013 and it now includes four full decades of data. From 1973 through 2012, the GTAA strategy shows exactly one negative year: a modest loss of –0.59% in 2008. And over those 40 years, the GTAA delivered an annualized return of 10.48% with a standard deviation of 6.99%, compared with a 9.92% return and higher volatility (10.28%) for a buy-and-hold strategy using the same five asset classes (US and foreign stocks, bonds, real estate and commodities).

The rest of the story

In my original post I described The Ivy Portfolio as “an excellent book that I would highly recommend to index investors,” and I stand by that. The authors do an admirable job of explaining the benefits of diversifying across uncorrelated asset classes, which is a core principle of the Couch Potato strategy. But in my follow-up post, I warned investors not to be tempted by the idea of avoiding drawdowns and getting higher returns: “At some point in the future we’ll have another bull market that lasts three or four years, maybe more. Timing strategies are guaranteed to underperform during prolonged bull markets (as they did throughout the 1990s and in the mid-2000s).”

We’re now well into a bull market that began in 2009, so let’s see how the GTAA has done since the book appeared. Faber’s updated paper includes data for both the GTAA and buy-and-hold portfolios from 2009 through 2012, and the timing strategy underperformed in all four years:

Buy & Hold GTAA
2009 18.74 12.77
2010 14.04 3.65
2011 2.81 2.49
2012 11.57 1.26

Over this four-year period, the buy-and-hold benchmark had a four-year annualized return of 11.64%, during which $1 grew to $1.55. The GTAA had a four-year annualized return of 4.95%, during which $1 grew to just $1.21.

The real-world performance

All of these numbers are hypothetical, of course, so let’s consider how GTAA worked in the real world. In October 2010, AdvisorShares launched the Cambria Global Tactical ETF (GTAA) and hired Faber and Richardson to manage it. The ETF takes the simple five-fund strategy to a much more complex level: the fact sheet explains it’s based on “a diversified portfolio of approximately 50 to 100 ETFs.”

The fund’s performance makes it clear it bears little resemblance to the strategies laid out in Faber and Richardson’s writing. In 2011, when the five-fund GTAA portfolio returned 2.49%, the ETF lost 7.29%. The following year, when the hypothetical GTAA portfolio in the paper returned 1.26%, the ETF outperformed by five percentage points. Bottom line, if you were attracted to the strategy described in the research, start figuring out how to implement it yourself, because you won’t be able to access it through this ETF.

And how has the ETF performed overall? As of December 27, its one-year return was 1.93%, and its three-year return was 0.42% annualized. Let’s compare this to a pair of Faber and Richardson’s own buy-and-hold benchmarks. The ETF proxies for the five-fund portfolio are tracked by Faber on his website; those for the 10-fund ETF portfolio are suggested on page 191 of The Ivy Portfolio. (Returns are in USD. Three-year returns are annualized and assume no rebalancing. All data from Morningstar.)

Five-Fund Portfolio  ETF 1 year 3 years
20% US equities VTI 33.88 16.07
20% Foreign stocks VEU 14.18 5.49
20% 10-year government bonds IEF -6.30 4.42
20% US real estate VNQ 3.10 9.72
20% Commodities DBC -6.67 -1.53
Total 7.64 6.83
10-Fund Portfolio  ETF 1 year 3 years
10% US large cap VV 32.10 15.94
10% US small cap VB 39.04 16.19
10% Foreign developed stocks VEA 21.39 8.33
10% Foreign emerging stocks VWO -4.60 -2.28
10% US bonds BND -2.21 3.25
10% TIPS (inflation-protected bonds) TIP -9.19 3.47
10% U.S real estate VNQ 3.10 9.72
10% Foreign real estate RWX 9.06 10.70
10% Commodities GSG -0.70 -1.10
10% Commodities DBC -6.67 -1.53
Total 8.13 6.27

Over the full three years, both buy-and-hold portfolios enjoyed a cumulative gain of over 20%, while the ETF’s return was effectively zero. Every dollar invested in the fund would have grown to $1.01.

Bad timing

The point here is not to pick on one active strategy or one fund. My goal is to help investors avoid being seduced by similar promises. There will always be active strategies that show past outperformance or reduced risk based on hypothetical models, but as the Cambria Global Tactical ETF demonstrates, there are no guarantees they will work in the future. And it is extremely difficult or impossible to execute them in the real world, where there are management fees, transaction costs, taxes and highly competitive money managers ready to arbitrage away market inefficiencies.

You also need to ask yourself how patient you will be when the active strategy underperforms, as all of them will eventually. According to Faber and Richardson’s own data, the GTAA strategy unperformed the buy-and-hold model in 12 of the 17 years from 1975 through 1991. And including 2013 (using the Cambria ETF as a proxy) it has now fallen short in five straight years, and in eight of the last 11. Active investors may be attracted to low volatility, but they also want comparable performance. How many would continue with any active strategy after multiple years of dismal results like that?

Successful investing is about more than the funds in your portfolio: it also requires a change in thinking. To stick to an indexing strategy over a lifetime, you need to let go of the idea that you can improve it with some kind of active overlay. Before you think you’re adding value with market timing or picking individual securities, remember that the probability of success is low and the potential payoff is small: even the best active strategies cannot hope to achieve more than an incremental outperformance over the long term.

More important, the whole pursuit is an enormous distraction from what’s really important, which is saving regularly, diversifying widely, keeping costs low and tuning out the chatter of those who promise what they can’t deliver.

62 Responses to The Failed Promise of Market Timing

  1. Jason January 24, 2014 at 1:24 am #

    The GTAA 13 AGG 6 strategy caught my attention in particular as a way to enhance returns over the more basic GTAA strategies. One measure of robustness is how well the strategy performs if the ranking / trading does not occur on the last trading day of the month, but some other day (say, the 2nd thursday of each month). It would seem that as long as a month was allowed to pass before re-ranking, the strategy shouldn’t care whether it is month-end or not.

    I set up a spreadsheet and looked at 2011-today using yahoo’s adjusted price date. As described in the paper, the top 6 of 13 assets are identified each month by averaging the 1 yr, 6 mo, 3 mo, and 1 mo returns. If they are above the 10 mo sma, they are bought or held. If they fall out of the top 6 or below the 10 mo. sma, they are sold. The process is repeated every 21 trading days.

    My numbers won’t match the published ones exactly, but since the beginning of 2011, trading at the end of the month yield a CAGR of 11.8%. So far, so good… the strategy is likely to underperform slightly in rising markets, and this does exactly that.

    However, if the monthly trading date is moved from the end of the month to another day, the returns fall drastically. Perhaps the strategy is quietly benefiting from some kind of end of month anomaly, but since that is not explained in any part of its reasoning, it is hard to trust.

    Days away from EOM : CAGR over last 3 yrs.
    +14 : 4.1%
    +13 : 2.9%
    +12 : 4.7%
    +11 : 3.3%
    +10 : 4.9%
    +9 : 5.6%
    +8 : 6.0%
    +7 : 6.2%
    +6 : 4.5%
    +5 : 6.7%
    +4 : 5.4%
    +3 : 8.0%
    +2 : 7.1%
    +1 : 10.3%
    0 : 11.8%
    -1 : 9.6%
    -2 : 9.8%
    -3 : 8.0%
    -4 : 8.6%
    -5 : 8.3%
    -6 : 5.9%

    I haven’t looked at the other GTAA strategies yet, but base on this, I have a hard time seeing how the GTAA 13 AGG 6 couldn’t be counted on to deliver anything like what is suggested by the paper.

  2. Brian G January 24, 2014 at 4:40 pm #

    @Jason, interesting finding but I don’t think it is wise to evaluate any stock investing scheme by only looking at a 3 year period. My rule of thumb is that you shouldn’t invest in stocks unless you plan to be there for 10 or more years. So, to evaluate a stock trading scheme requires one to look at least 10 years if not 100 years. 🙂

    Also, if you are correct that there is a monthly cycle that is so predictable, then that is interesting by itself and maybe it should be exploited until it goes away! There are some reason’s why there may be a monthly cycle. 1) timing of option contracts 2) timing of dividend payments 3) monthly incentives for institutional traders.

    Finally, are you sure you didn’t make any mistakes in your spreadsheet? I’ve made them before when doing work like this. You must be really careful.

  3. Jason January 24, 2014 at 10:32 pm #

    @Brian, yes, 3 years isn’t enough time to prove that a strategy is sound, but I think that it can be enough time to prove that a strategy is *not* sound. 100 year backtests are fun, especially if the results are broken up into time periods of 10-30 years that matter more to typical investors, but getting data for anything except the most basic market indices is tough for amateurs. Survivorship bias is more likely to sneak in the longer you look back, and avoiding that can be difficult sometimes too.

    My point is less about the fact that the end of month trades appear to be working, but more to support one argument of your post… as complications have been added to the basic gtaa premise, there is brittleness introduced that may not be obvious from the published results. For anyone baffled by the idea that the GTAA ETF is underperforming the basic strategy so much, the results I’ve found may be one clue as to why.

    Sure, it is always possible the spreadsheet has an error, even though I’ve reviewed it closely for accuracy. I’m not holding myself to the standard of an SSRN publication… I just wanted to point out that anyone seriously considering this strategy ought to look carefully at how the timing of the trades matters in ways not discussed by the original paper. I am not willing to trade a momentum strategy that can be completely undone if the trading is offset by a week. Others may be more confident in their ability to consistently make the necessary trades on the last day of each month, and that the end of month anomaly will persist. Good luck to those of you that do.

  4. Jason January 25, 2014 at 1:36 am #

    I’ve looked at the other GTAA extensions. The range of possible returns represents how the strategy does if the trading day is offset from the end of the month and assigned to any other particular day of the month. The CAGR listed below is the average of each of those 21 possibilities.

    GTAA 13 Moderate : 3 yr CAGR = 3.4%, range = 2.2% to 4.4%
    GTAA 13 Conservative : 3 yr CAGR = 3.5%, range = 2.5% to 4.2%
    GTAA 13 Aggressive 6: 3 yr CAGR = 6.8%, range = 2.9% to 11.8%
    GTAA 13 Aggressive 3: 3 yr CAGR = 8.6%, range = 1.7% to 12.4%

    The Moderate and Conservative approach offered much less variation in the return when the trading day is moved from the end of the month. This would be expected since each asset goes to cash when falling below the SMA, instead of aggressive (and time sensitive) betting on momentum like the AGG strategies do. Note that I haven’t included any return on cash… parking it in T-bills or 10 yr notes could give different results, but I view that more as of an enhancement than a main source of returns from the strategy.

    The AGG3 actually looks pretty decent, there was only one really bad choice for a trading day but for the most part the returns looked good. In this case, the max return was not achieved with an EOM trading day, but instead trading about a week before month end.

  5. Brian G January 25, 2014 at 4:12 pm #

    @Jason, the past 3 years is not enough to prove that it’s not sound; the past 3 years has been an odd Fed fueled bull market. This is only one type of market.

    Also, I don’t like the fancy 13 asset variants of GTAA. That many assets doesn’t work because the asset classes are not going to be non-correlated. As I’ve stated before, I don’t consider two asset types to be different if they get correlated in market downturns.

    E.g. the past week all stock markets fell together. They always do that. So let’s stop kidding ourselves… they are one asset class. So what moved independently? Government bonds? Gold? CDN Cash? US Cash? Harry Browne was on to something.

    The ability to perform in a bear market or during a sudden downturn is my measure of a good strategy. Bull markets hide everything, as Warren Buffet says “you only find out who is swimming naked when the tide goes out”.

  6. will January 25, 2014 at 10:42 pm #

    I’m also not a fan of that complicated asset class strategy with the 10 month average timing.

    I’m far more interested in the general market and what it’s doing, because as you say, all of these different asset classes are really mostly correlated, and I’d rather just grab the whole picture with the index. It’s cheaper and less complicated.

  7. daniel morton March 6, 2014 at 7:49 pm #

    Thanks Dan. The tracking error of the GTAA ETF versus the manual approach is very disconcerting.

    What is the longest GTAA portfolio track record (active and backtested) that you know of for an American investor? I look forward to reading your book and hope our paths cross when we come to Canada.

    Daniel

  8. Canadian Couch Potato March 7, 2014 at 12:09 am #

    @daniel: In The Ivy Portfolio, there are some charts that go back as far as 1900, though that needs to be taken with a grain of salt. The good quality US data is generally considered to go back only as far as 1926.

  9. daniel morton March 12, 2014 at 12:24 am #

    Thanks Dan. I am trying to understand the constraints of the GTAA 13 AGG 3 system. Faber writes “The assets are only included if they are
    above their long-term moving average, otherwise that portion of the portfolio is moved to cash. We also include the effects of only investing in the top three out of thirteen assets. ”

    I don’t see how this is possible, surely I am missing something obvious. If you are investing only in the top 3 of the 13 assets then how is that not contradictory to “otherwise that portion of the portfolio is moved to cash”? I am trying to confirm what happens in each scenario:

    1. All 13 10 Month SMA then you are 33% in that 1 asset and rest in cash?
    3. 3 Assets > 10 month SMA and 10 assets < 10 month SMA then you are 100% invested with 1/3 in each top 3 assets in which case you have 0 in cash which contradicts the rule stated before.

    Thanks for any clarity.

    Daniel

  10. Brian G March 12, 2014 at 1:53 am #

    @Daniel, I believe Faber’s system works by starting with 13 asset class and eliminates any asset classes that are under their current 10 month SMA. The remaining asset classes are ranked by their 10 month SMA. Finally, the top 3 are used to fill up 3, 33.3% buckets. If there are not 3 that pass the above filter, the remaining buckets are left in cash.

    I hope that helps.

  11. Pat September 13, 2016 at 5:47 pm #

    Hi Dan,

    I’m wondering if the current situation in Canada is an exception to timing the market. It’s widely acknowledged that Canada has a housing bubble, and some say it is extreme. The warnings are everywhere now, even coming from the federal government, and the banks, in addition the to the IMF, the OECD, The Economist, leading American economists, and many other knowledgeable bodies and people. Basically, every economist in the world worth their salt is pointing out how bad the Canadian real estate market and Canadians’ indebtedness have become. By any traditional measure (price to rent ratio, average price to average income ratio, etc), real estate in Canada is at extreme prices. Also, Canadian households have hit record high levels of indebtedness, most of it mortgage debt. These indicators are at worse levels than they were in the US leading up to the 2008-09 financial crisis. And subprime lending in Canada is significant.The bubble will eventually pop. Already, Vancouver’s real estate market is getting hit.

    When the bubble pops, it seems likely the Canadian stock index will suffer a big drop. I’m not sure I’m comfortable having 1/3 of my equity investments in the Canadian index, and continuing to add to it monthly when the writing seems to be on the wall. I’m inclined to transfer my Canadian index investments to the US and International indexes and stop contributing monthly amounts to the Canadian index, until the bubble has popped and the Canadian index has suffered significant decline. I made almost all my Canadian index investments in late January, so they’ve done really well, and I don’t mind selling them and buying the US and International indexes with the proceeds. I know I can’t time the Canadian index bottom, but that’s not necessary. I just need to wait until it drops a lot before starting to add monthly contributions to it again to build it up to 1/3 of my equity holdings. I could start making contributions again to the Canadian index before or after the bottom , it doesn’t really matter, the Canadian index will probably be at a low price. I understand that it is recommended to have 1/3 of equity in the Canadian index to help offset currency risk, but does the impending risk of a Canadian stock market drop outweigh this? By keeping my equity/bond allocations the same throughout this process (except without Canadian equity, just US and Int.’l indexes) this seems like it may be prudent. The extreme Canadian real estate market, and it’s potential impact on the Canadian stock market does give me pause for thought.

    Your thoughts would be much appreciated.
    Thanks

  12. Canadian Couch Potato September 13, 2016 at 7:34 pm #

    @Pat: It all sounds so logical, but a few things to consider:

    – If “every economist in the world worth their salt” knows a crash is coming, why would that not be priced into the Canadian stock market already?

    – The Vancouver market has cooled considerably this year. Canadian stocks are up about 14%. Are you sure the market would react to a real estate crash they way you expect it to?

    – People have been calling a housing bubble for about 10 years now (see Garth Turner’s blog, for example). Eventually they will be right, but at what opportunity cost?

    – When are you going to get out, and when are you going to get back in? Do you have a strategy, or will you wait until it “feels right”?

    – If you underweight Canada, what are going to overweight, and why?

Leave a Reply