Flies in the TAA Ointment

15Nov12

Despite the fact that most tactical asset allocation (TAA) models have fallen on mediocre times over the last couple of years (1), I’m still mostly positive about the concept, mainly because the core principles of most of these models (trend-following, momentum, etc) have worked across mature markets for generations.

Having said that, there are some flies in the ointment for TAA moving forward that are rarely fully embraced, but will make these models less effective in the future.

* * * * *

The most critical of these is the little potential upside still left in US Treasuries and other low-risk fixed income assets, and all of the problems that necessarily leads to.

If yields increased from a higher baseline (like Treasuries in the 1970’s), bond funds (which tend to maintain a somewhat constant maturity) could still be a long-term source of return, because today’s coupon makes up for some of the price change resulting from increasing yields.

But that’s not as much the case today with yields so low (see estimates of the hypothetical upside still remaining in 10Y UST). An increase in yields today would have a much more negative impact on bond funds, ETPs, etc.

Treasuries can still serve to reduce volatility in the portfolio (because Treasuries exhibit low to negative correlation with equities), but if it fails to be a source of return (or worse, a drag on returns) TAA will suffer.

That problem is made worse by the fact that correlation across most major asset classes has been increasing in recent years. No longer are international or emerging market indices, or individual (developed) markets, or private equity, or real estate, or commodities strong diversifiers of equity risk.

The only other major asset class that still acts as a strong counterweight to these equity-like investments, is gold and precious metals (and equities + gold do not a diversified portfolio make).

In short, over the long-term: without Treasuries, TAA volatility will rise, but with Treasuries, returns will fall.

Any attempt to reach for higher yields (ex. corporate bonds) is just going to increase the portfolio’s correlation with equities, et al. (especially during market shocks), and therefore increase portfolio volatility.

None of the above ramblings are new information, I’m just thinking aloud.

And lest I sound like Chicken Little, let me say that there’s still some juice left in Treasuries, so this problem doesn’t necessarily kick in tomorrow. And even without Treasuries, there’s clearly value in using trend-following, momentum, etc.

My point is that (despite the fact that TAA and all of its sophisticated alternatives have become all the rage as of late) there isn’t enough talk about the realities of a future where traders’ most significant source of returns that are uncorrelated with equities fails to deliver.

* * * * *

I’ve reached my self-imposed word limit.

This subject is important to me. I’ve always been successful at building strategies to generate big aggressive (but volatile) returns, but I often struggle with a low-vol, “sleep well at night” home for the core of my wealth.

In a follow up post I’ll share my own thoughts on solving this problem.

Happy Trading,
ms

(1) Of course, I’m referring to real-time, out-of-sample TAA models (i.e. the only kind that really matter), not the backtested variety burning up the blogosphere.

P.S. on further thought, I left out currencies as a major asset class w/ the potential to diversify equity risk. I haven’t seen much juice from currencies in any of the TAA models I follow, but I’ll throw it out there.

. . . . .

To stay up to date with what’s happening at the MarketSci Blog, we recommend subscribing to our RSS Feed or Email Feed.



22 Responses to “Flies in the TAA Ointment”

  1. Me too still believe in the concept of TAA. People tend to invest in a TAA model after seeing a glorious 20 or 40-year back-test without noticing it usually underperforms the market when the market is sugar high (something not that obvious in PnL curve). But the real advantage of this model, instead of consistently outperforming the market, is to resist drawdowns when the market performs really bad. Investors who walk into it with wrong expectations will probably find themselves missed out for QE hurrays and politic-quakes.

  2. 3 Steve

    Hi Michael, I’ve also been wondering about the underwhelming performance of TAA especially when compared to the results of many risk parity offerings including AQR, Invesco and Salient. I was curious about your take on this- do you think their equal risk budgets and leverage account for better performance? Also, I’d be interested to know if you’ve thought about using risk factors (premia) rather than asset classes to create TAA portfolios.
    Here’s a paper from DB on the subject-
    http://www.portfolio-cafe.com/wp-content/uploads/2012/06/Rethinking-Portfolio-Construction-.pdf

    • 4 MarketSci

      Hello Steve –

      Most TAA models are some variation of trend-following/momentum type trading. That approach has gotten killed in recent years. A nice post today from Mebane Faber on the subject:

      http://www.mebanefaber.com/2012/11/15/the-end-of-a-trend/

      Long-term plays like this (or value investing, or risk parity, or ) go through long periods of out or under-performing.

      As usual, investors will pile into whatever the flavor of the year is. But I see no more reason to call TAA dead today than there will be to call dead a few years from now when it goes through its rough spot.

      All of these are generational strategies. The basic math says that if they were always on, always outperforming, they would be putting up 20-30%+ a year (and no one should expect that from such a long-term play).

      michael

  3. 5 Vix-Trader

    I don’t believe changing markets are the reason TAA models are showing weakness. They just look weak right now because we’ve seen a pretty strong S&P 500. Everything is relative, and I think it’s a fair assumption that TAA models will have periods in the future where they again show relative strength compared to the S&P 500. It’s just the normal ebb and flow of the markets, and I really don’t think anything has changed in the last few years.

    Having said that, the biggest problem TAA models face is even in good years the returns aren’t very high. So in a world where people can make 10-20% a year trading in other ways, like option trading or trading volatility products, where does TAA really fit in these days?

    The fact that an XIV Term structure play in the last 12 months returned 150% isn’t helping the TAA model argument either. Now obviously that won’t last, since this may go down as the best year in history for that type of trade, but it has at the very least added some skew to the trading world.

    Compared to itself, I think TAA is alive and well and will soon start to perform well again. Compared to other trading styles, that’s another story entirely and I don’t see TAA catching up in that respect.

    • 6 MarketSci

      Hello B – I agree with a lot of that. If I were nitpicking I might say…

      Remember the golden rule: return is an illusion. Return relative to vol, dd, etc. is what matters. You’re not going to wake up tomorrow and watch half your diversified portfolio evaporate the way you very well could with a big XIV position.

      As you know, I also trade vol, so I’m not poopooing that trade, but the “A returns more than B, therefore A is better” argument needs to go the way of the dodo.

      And if I were to nitpick a little more I might say…

      You’re comparing apples and oranges in terms of strategy robustness. TAA is (usually) based on ideas that have worked for many generations. By contrast, the vol play has been with us for a half decade.

      Not so long ago, everyone (including me) was printing money playing short-term mean-reversion. Anyone who thought that party was going to go on indefinitely was fooling themselves. I don’t know what makes the music stop for vol strategies, but it will stop at some point.

      So the benefit of TAA isn’t return (b/c return is an illusion), it’s (a) return relative to risk, and (b) strategy robustness.

      michael

  4. Well you’ve said that too me a few times before, but stubborn me I don’t fully understand it :)

    When you say return is an illusion, and A returns more than B is not a good argument, what am I missing? What else matters besides that?

    Now I realize that volatility adjusted returns are also important, but they certainly aren’t the end all of trading. Example:

    Trader A has a killer TAA model, which doubles the S&P 500 returns during good years, and cuts losses in half during bad years, averaging 10% a year forever with drawdowns of no more than 10%.

    Trader B is a term structure XIV trader, returning a long term average of 20% a year, but suffers massive drawdowns as big as 50% here and there.

    So trader A makes 10% a year forever, and trader B makes 20% a year forever. I guess I kind of think i’d rather be trader B. My sharpe ratio never paid my rent, but my returns seem to be so far.

    I really don’t know…. Just an opinion. I never went to school for any of this, so I guess my understanding of volatility adjusted returns is pretty limited. Maybe it’ll dawn on me someday

    • 8 MarketSci

      For a number of reasons, but I think Condor Options sums it up nicely here (last paragraph):

      http://condoroptions.com/2009/04/23/moving-averages-and-human-risk/

      It’s very easy to look at a strategy that (for example) blindly follows the VIX futures term-structure today, with the benefit of hindsight, and say, yeah, I would stuck with that strategy.

      It’s very different to have been in a massive 50% drawdown, trading this brand new product, and honestly say you would have had the intestinal fortitude to stay the course. And it’s going to be just as difficult the next time that that trade goes off the rails.

      If you’re trading play money, fine. But when you’re managing real wealth and preventing big losses is just as important as earning abnormal returns, “massive drawdowns as big as 50% here and there” are unacceptable.

      And that all assumes that you have some reason to believe that the strategy is going to climb out of that drawdown. In the real world, we don’t get the benefit of hindsight. In the case of the term-structure play, we’re talking about a strategy with little historical data to consider.

      Really ask yourself, if in the next week XIV loses 50%, what are you going to do? Will you stay the course?

      P.S. returns are an illusion b/c they’re a function of exposure. Trader A went 100% long XIV a year ago. Trader B went 200% long. Who has the better strategy? Terminal value says trader B. But no sane person who understands what XIV is capable of would agree.

      michael

    • 9 qusma

      Sharpe matters because you always have the option to leverage up. Even with a reg T account you can go to 200%. Suddenly trader A is matching trader B’s returns with less than half the pain.

  5. 10 Vix-Trader

    Well yeah, the basics of it are pretty clear. We shouldn’t care about a strategy that yields massive returns, if in order to get those returns you’d be risking losing an enormous amount if you’re wrong. So of course volatility adjusted returns matter to a point. I just don’t think it’s quite so simple to just compare sharpe ratio for example and pick the winner. I don’t think returns are simply a function of risk. I think the concept of the model matters just as much. If we had reason to believe a high yielding system with big drawdowns would persist that way into the future, then it would actually be worth the risk to trade it.

    Now I don’t actually trade the XIV, but I use it as an example because so many other people do. I personally just option trade, but keeping with the XIV example, there actually is reason to feel the risk is worth it, even if we did see a big 50% drawdown and you were actually in the trade when it happened (very unlikely, but lets assume)

    The concept of contango goes beyond just XIV and the stock markets, it’s actually human nature. XIV as an ETF is new yes, but contango is as old as the markets. Until human nature changes, I think there will always be a way to profit from it year in and year out. Now obviously not at 150% annually like we saw this crazy year, but probably enough to put TAA to bed, not to mention all the active fund managers out there who can’t seem to even beat the S&P.

    Drawdowns and a poor sharpe ratio may just be part of trading XIV. As much time as I’ve spent on it, I can’t find a way to avoid those. The best you can do is come up with a formula that includes the standard deviation of the VIX, the trend of the EMA, the ratio of implied over historic vol, and VIX price stops. But regardless of how geeky you get into it, drawdowns are just part of the system.

    I still stubbornly stand by it…. Even if the volatility adjusted risk is worse than a good TAA model, I still say XIV trading wins. I hope you don’t feel like you’re talking to a brick wall :)

  6. Great conversation and a good post.

    To the original post, what about looking at foreign currency or alternative strategies like managed futures or long/short managers to add more uncorrelated returns to the portfolio?

    To the conversation, the reason that risk adjusted returns trump absolute returns is because (theoretically) you can leverage up a lower return, better risk adjusted return strategy to hit higher absolute return objectives.This isn’t as easy as it sounds without access to margin or products that give you leverage like futures, so an argument can still be made for looking at absolute returns.

    Keep up the great work!

    • 12 MarketSci

      Hello CIM – I mention currencies in the post script – I’ve never seen a model get a lot of juice out of currencies, but I agree that it’s another source of low/neg. correlation w/ equities. Should have mentioned that in post.

      As to the other investments you mention, b/c TAA in the usual sense of the word (a) frequently moves in and out of assets, and (b) is based on some instrument that can be modeled (backtested), any asset traded would need to be (a) liquid and (b) based on some index or fixed set of transparent rules. As far as I know that doesn’t exist in the managed futures or long/short space, but I’m always open to being proven wrong.

      michael

  7. 13 K

    Hello
    I am located in Europe ..How about the Permanent Portfolio as a replacement or viable alternative to TAA … PP is hyped here http://crawlingroad.com/blog/2008/12/22/permanent-portfolio-historical-returns/ and in many other blogs… Would love to have your opinion Thx KV

    • 14 MarketSci

      Hello K – I’m a fan of the perm port, but it will also be negatively impacted to some degree by the issues discussed in this post. I think over the long run TAA outperforms PP b/c of the ability to reduce equity exposure in downturns, but that’s just a semi-educated guess. michael

    • 16 MarketSci

      RE to Steve: possibly agree on narrow “improvements” to TAA, et al. Disagree on the fundamental concepts of trend-following and momentum. Those are concepts that have worked across mature markets for many generations (and many generations after being initially “discovered”).

      Doesn’t make sense to call the death of an approach when it’s performed well within historical norms.

      michael

  8. 17 finsovet

    Michael, thanks for the post. I also spend time thinking of whether UST will sustain this cushioning ability in portfolios with adequate returns. What made me sanguine on the matter are two thoughts. 1) Who has cancelled the long bond trade or predicted its demise? http://advisorperspectives.com/dshort/guest/Dominic-Cimino-121112-Great-Long-Bond.php There is no evidence so far. 2) UST yields act like gravity on all yields, including equities, junk bonds etc., so _relatively_ to all other asset classes UST will always be a proper diversifier and there is no point to blame them for poor returns going forward, as _all_ yields are getting lower.

    • 18 MarketSci

      Hello Finsovet – as I mentioned in this post and shown previously, there’s room for Treasuries to run:

      http://marketsci.wordpress.com/2012/10/16/follow-up-to-timely-portfolios-what-if-we-go-to-zero/

      But the basic math says that (a) there is limited upside left in Treasuries (and much less than many models assume), and (b) when yields do increase, Treasury funds, ETPs, etc. are going to be more sensitive to that increase b/c yields are so low today.

      That’s not my opinion – that’s math.

      The “it’s worked for 30+ years and I see no reason for it stop now” argument doesn’t work the same way for Treasuries as it does for other assets because there is (realistically) a floor on how low yields can go from here.

      michael

  9. 19 Jon

    You’re killing me making me wait for Part 2. I still think we have another ~2 years of earning just the coupon return of bonds, but I’m really curious as to your approach for TAA purposes when yields start to go up and prices start to fall. Some sort of timing mechanism to effectively short bonds would be almost ideal in my mind. Probably a higher correlation with equities but at least you (probably) get positive expected returns along with some volatility-smoothing effects for your overall TAA portfolio…

    • 20 MarketSci

      Hello Jon – sorry about that – went down the research rabbit hole on this concept and then doing some traveling – will be back in the saddle shortly to finish off this thought. michael


  1. 1 Thursday links: marble madness - Abnormal Returns | Abnormal Returns
  2. 2 Balanced Portfolio Risks | John Standerfer

Leave a Reply

Please log in using one of these methods to post your comment:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s