The “Hot Potato”

April 18th, 2016 by Potato

When talking about passive investing and diversifying across many asset classes, a common quip is that if you knew in advance which asset class would perform best, you’d just put your money there. Since you can’t, you diversify.

Well, in a recent issue of MoneySense, Norm Rothery delved into the “hot potato” that does exactly that (he also wrote about it in August of 2015). Using a momentum strategy, it goes all-in on whatever the hottest sector was over the previous 12 months. The extra return over a more vanilla, diversified approach is sweet — too much to ignore without some further thought.

Norm puts in a number of disclaimers at the end of the article, but before going ahead it’s worth thinking about those risks and more.

Firstly, I have to point out that going all-in on something (even something as diversified as a broad-market ETF) opens you up to blow-up risk. It didn’t happen in the recent past, but that’s not saying it couldn’t. That’s a fundamental risk of concentration — you could go all-in on the one asset class that goes down one year, then switch and catch the next loser.

It also looks to me like a strategy that’s possibly not very robust and vulnerable to execution risk. I don’t have data on returns with monthly resolution going back very far to test out how robust it may be, but it’s easy to get for the ETFs available to Canadian investors for the 2008 market crash and do a spot check. Norm mentions that the monthly rebalanced version of the hot potato only fell 10% in 2008 and recovered by 2009. Going into 2008 the hot potato had you do a few little dances between XIC and XBB. In August of 2008 it would have had you buy XIC, just to eat a 12.5% loss, and switch back to XBB the next month. If you went on vacation and missed that switch by two months, you were down 33%; if you missed by five months you were down 40%.

Momentum is a weird thing in the markets: it’s there, it looks to be exploitable, but it doesn’t have a “mechanistic” basis: controlling costs does, as does diversification — you can predict that those “should” work in the long term. But for momentum, even if there is some psychology or whatever to the momentum effect in general that makes it a real effect that may possibly form the basis of a successful strategy, what is it about trailing 12-month returns and monthly rebalancing that worked so very well for the historical check of the hot potato? Could that shift in the future to 6 or 24-month returns being the momentum sweet spot, and leaving someone using 12-month/rebalanced monthly strategies in the lurch? Are the features of momentum stable and exploitable enough to bet it all (or even compromise with dynamic sector weightings) on something like the hot potato? Or is this an accident of over-fitting historical data? Those question marks may make it a difficult strategy to stick to when a period of underperformance eventually comes along.

Larry Swedroe has an article looking at momentum in general that’s worth reading, in particular the line about how the strategy becomes less valuable as correlations between global markets increase.

I don’t know, and haven’t invested a tonne of time doing research here — hopefully Norm or someone else has a more data-driven answer. For now, I’m not brave enough to try it with actual dollars in the uncertain future — I’ll stay diversified and lazy.

9 Responses to “The “Hot Potato””

  1. Says:

    Going all in, into one small segment of the market? That’s insane.

    Even if it does purportedly produce better results, the increased volatility is going to be crazy. And, while I’m no expert on this, I suspect that the downside of the volatility is that you could end up in a segment that never comes back – and you’re then toast.

    The whole point behind diversification is to reduce volatility. As long as you do the right diversification, you should be able to reduce volatility without killing returns. Why wouldn’t you just do that?

  2. Grant Says:

    Actually, Glen, dual momentum, as described by Gary Antonacci, in his book of the same name, shows there is deceased volatility and increased returns over very long time periods. It works best in periods like 2000 and 2008, when it gets you out of the market fairly early on avoiding losses, not so much in choppy markets when you can get whipsawed. A good summary of the strategy can be found here. I don’t do it myself, but it looks very appealing in many ways, especially for tax protected accounts.

  3. Richard Says:

    While reading the article in full I noticed immediately that it references data starting in 1981. I’m always suspicious of results over that timeframe because of the unusual bond returns.

    Also interesting that the inspiration was from James Montier who works at GMO, a firm that constantly talks about betting on reversion to the mean — essentially the opposite. Although I can see that if a market is coming off a bottom then its first good year might be followed by another good year, moreso than a hot market that’s been doing well for a few years. I’m not sure if this strategy tends more towards the former or the later.

  4. Grant Says:

    Richard, in the paper linked in my comment above the data goes back to 1974. It shouldn’t matter that much what bond returns are, because the focus of the strategy is that is gets you out of the market early on in a down turn, and back into the market early on in the upturn.

  5. Richard Says:

    I think bond returns do matter, if the strategy is switching you into bonds frequently during a period where they may have annual returns of 7 – 10%. Try that when they have returns of 2 – 4% and watch the difference :)

    I checked some historical return and it would seem that it selects bonds in 13 out of 43 years when the other choices are Canadian, US, or EAFE. Over that time there were only 17 years when the top-performing asset class didn’t last for more than one year and 28 switches overall. There are a few instances where the best-performing asset class has a swing of 30 – 40 points and does really badly the next so it’s not sensitive enough to avoid downturns in a reliable way.

    I doubt this would hold up well to transaction costs. Maybe the fact that it can only be attempted in tax-free accounts makes more persistent, since larger traders are more likely to pay taxes than individual investors.

  6. Potato Says:

    Thanks for the discussion guys (and sorry I’ve been AWOL with work and getting taxes filed).

    For the timeframe, I wonder if a period with more choppy/sideways markets might throw this strategy for a loop (IIRC the 60’s and 70’s had more of that, though the Antonacci paper Grant linked would have included some of that early on in its set).

    Transaction costs and taxes would be higher in a strategy like this, but for the monthly version in Norm’s article the edge (before taxes/fees) is ~6%/yr — don’t even have to pull out the calculator to know there’s still going to be an advantage even after taxes and fees there.

    Thanks for the link, Grant. I love the line in the intro: “Despite an abundance of momentum research over the past 20 years, no one is sure why it
    works.” That gets to the heart of my fear with it, that it might stop working.

    Of course, a robust phenomenon doesn’t have to have a known mechanism to exploit if it’s robust enough… Reminds me of a quote a friend in EEG research had pinned up a few years ago: “Despite decades in sleep research the best answer I can give as to why humans need to sleep — what function it serves — is because we get sleepy if we don’t.”

  7. Weekend Reading – Pricey mutual funds, $1 million TFSAs, Tesla and more - My Own Advisor Says:

    […] Potato is not convinced the “hot potato” passive investing approach is right for […]

  8. Richard Says:

    I wouldn’t discount the risk / cost of trading your entire portfolio every month.

    It strikes me that this is increasing the risk and returns with a lot of work, and it’s only demonstrated in theory with none of the real-world costs of trading.

    If that’s the aim just leveraging your portfolio is probably far easier.

    Using the Libra history of returns I found the following annual returns: 10.48% for a balanced portfolio with 25% in bonds, 11.94% for annual trading (again this is the perfectly theoretical version), and 11.76% for a balanced portfolio with no bonds, 25% leverage added, a 5% annual interest cost, and no trading other than rebalancing.

    There’s risk and then there’s risky risk. I keep it simple and use a little leverage.

  9. Grant Says:


    1. The actual bond returns don’t matter because you are only switching into them during times of stress, when equities are in a bear market when bonds generally do well, eg in the crash of 2008 bonds were up significantly. The majority of the time you are in equities.

    2. I agree the strategy would work best in a tax protected account to avoid capital gains taxes.

    3. You do not trade your portfolio every month. You only trade when the ETF you are in goes negative relative to t-bills over the last 12 months. I think there is on average about one trade per year.

    4. I’d encourage you to read the book “Dual Momentum” which describes the strategy in detail. It is based on US data and in fact he back tested the strategy for non US investors and concluded we should not use our home country equities, instead use a US domiciled ETF (large cap, not broad market), for US equities and international equities, and either Candian bonds or US bonds (hedged) for the bond component. So Norm’s version using Canadian equities appears not to be optimal – more ETFs so more trading.

    5. The big risk of buy and hold is the draw down and the big risk for dual momentum is a choppy market where you will get whipsawed. Perhaps buy and hold for taxable accounts and dual momentum for tax protected accounts? This would divery these rwo risks. In other words strategy diversification. But then you have added complexity and activity which I really dislike. I’m tempted, but I’m still 100% buy and hold.