Dual Momentum: Can I Use An All-World Fund Instead?

I've been wondering about the relative momentum aspect of Gary Antonacci's Dual Momentum strategy for some time now. Does the relative momentum component add a meaningful return to the strategy if you look at the entire world stock market?

To refresh memories, Dual Momentum considers the two main aspects of momentum. Relative momentum is comparing one asset to another asset when both are in the same broader asset class. This can also be called relative strength, or cross-sectional momentum. In the Dual Momentum strategy, we compare U.S. stocks to International-ex U.S. stocks when measuring relative momentum. For those that are curious, U.S. stocks are approximately 80% correlated to International stocks over the past few decades.

The second momentum component is absolute momentum. Absolute momentum is also called time-series momentum. This compares one asset's price now to the price at a prior specified time, such as the performance over the previous twelve months. If the asset has a positive return in excess of the risk-free rate of return (represented by short-term government debt), then the asset it can be said to have positive momentum.

Absolute Momentum Only Studies

Gary Antonacci has done comparisons using absolute momentum himself. But interestingly the only comparisons I can find use a single component: either the absolute momentum of U.S. stocks, or the absolute momentum of International stocks. Both of these types of absolute momentum under-performed Dual Momentum—but that makes sense. Sometimes U.S. stocks do better than International stocks for a period of time and sometimes they don’t. In the long run that means poorer performance if you only use one of these two assets.

What I couldn’t find is an absolute momentum back-test, using Gary Antonacci’s standard twelve month look-back, on Global Stocks as a whole (represented by the MSCI World Index before 1988 and MSCI ACWI All County World Index after 1988).

Since it piqued my curiosity, I decided to do a long back-test—taking things back to 1970. I call it the Global Stocks Absolute Momentum (GSAM) strategy. I had to use 1970 as a start date since I could only find global stock index information going back that far.

For global stocks, I used the MSCI World Index from 1970 through 1988 and I used the MSCI ACWI from 1988 onward. Unlike the MSCI World Index, the MSCI ACWI includes emerging markets as well. Prior to 1988, it's very difficult to obtain good data on emerging markets.

For bonds I had some difficulty locating good monthly data. While the bond rates information is out there, rates on bonds—especially medium term or longer—do not translate easily to monthly performance. So I used U.S. T-bill data from 1970 through 2004 and the ICE 7-10 Year Treasury Index from 2005 onward.

While the bond data issue would have a small effect on returns, it shouldn’t be that significant. If anyone has and can legally share good bond index monthly data that goes back to the 1970s, please let me know and I will tweak my spreadsheets.

My Hypothesis

My idea in doing this study was to test if there was a meaningful advantage in the relative momentum aspect of Dual Momentum when testing the strategy using better stock data.

While U.S. stocks only, or International stocks only, would clearly show a less favorable result, the issue with this comparison is that you only are looking at one-half of the equation. By comparing global stocks as a whole, you are looking at the performance of both U.S. and International stocks in a single package.

As many of the trades in Dual Momentum involve a switch from U.S. stocks to International stocks, and U.S. stocks are quite highly correlated to International stocks, my theory going in was the results of GSAM would be a bit lower performance than Dual Momentum with a lot less trades.

In theory, when U.S. stocks perform better than International stocks, the MSCI World/ACWI indices should go up as a whole somewhere in the middle of the performance of the two asset groups. Likewise when International stocks outperform U.S. stocks. Also, when one stock market group falls, the other tends to follow soon after.

The Results of GSAM

GSAM Logarithmic Scale 1970-2017. Sources: TheRichMoose.com, MSCI Inc., ICE (Interactive Data LLC), FRED (Federal Reserve Bank St. Louis)

After compiling all the information and synthesizing it in a nice spreadsheet, the results are in: eliminating the relative momentum component and comparing only Global Stocks to bonds is not a good idea if you're looking for comparable performance to Dual Momentum.

GSAM tended to do somewhat worse than Gary’s Dual Momentum (GEM) model in nearly all years. In up years, the gains were not quite as high and in down years the drawdowns were a bit worse.

Performance of GSAM vs. Stocks Only

GSAM produced a tidy 10.28% compound annual return from 1970 through 2017. This nicely beat the 8.04% annual return of the MSCI World and MSCI ACWI indices since inception.

The worst year for GSAM was 2011 with a -13.98% result for the year. The best year was 1986 with a 42.81% gain.

This compares to stocks only which would have seen the portfolio fall -43.5% in 2008. The peak to trough decline of stocks by end-of-month data from 2007-09 was -56.22%.

GSAM’s drawdowns were much smaller than investing in stocks only and the out-performance was amazingly consistent over the decades.

The results of GSAM clearly showed out-performance to a global all-stock strategy. $100 invested in January 1970 would have grown to $9,900 on December 2017 following the GSAM strategy. That’s nearly more than double the performance of global stocks only which would have grown to $4,260 in that same time period.

Performance of GSAM vs. Dual Momentum (GEM) Strategy

While Gary Antonacci only has his Dual Momentum performance tracked back to 1974, the results of Dual Momentum are unequivocally better than GSAM. GSAM produced an average annual return of 10.28% compared to Dual Momentum’s 17.5% annual return.

That 7% difference every year adds up significantly over time. Over a typical person’s 35 year timeline of investing, if they save $1,000 a month, that’s the difference between ending up with $3.6 million instead of $20.8 million. Yes… you read that correctly.

As in Dual Momentum, GSAM’s drawdowns were much smaller than investing in stocks only. However, GSAM had somewhat larger drawdowns than Dual Momentum—something that surprised me a little. The drawdown aspect may have something to do with the bond data issue, but I don’t think the difference would be too significant with better data.

One thing the GSAM back-test did show, as suspected, was a lot less trades would have been made. Less than 1 trade per year, on average, was made with the GSAM strategy. That equates to nearly half of the trades made in the Dual Momentum strategy.

While the overall performance was not as good as Dual Momentum, GSAM still has its merits when compared to a simple buy-and-hold Couch Potato strategy. It’s easy to execute, the drawdowns are quite minimal, the strategy is simplistic, and you get decent results.

Unfortunately my hypothesis was only partially correct. If I had to choose between Dual Momentum and GSAM, I would easily choose Dual Momentum. The 7% difference in annual returns is just too high; a 2% or 3% difference might be acceptable if reducing trades was important to you.

Comments & Questions

All comments are moderated before being posted for public viewing. Please don't send in multiple comments if yours doesn't appear right away. It can take up to 24 hours before comments are posted.

Comments containing promo links or "trolling" will not be posted. Comments with profane language or those which reveal personal information will be edited by moderator.

Trend Investing: Tools To Determine The Trend

I make it no secret that I am a trend investor. I do not buy positions to hold them forever. Instead, I buy positions when the price is in positive territory based on a specific metric and sell those positions when the price is negative based on the same metric.

How do you determine what the trend is? What kind you do to find out if the trend is up or down? Once you have determined the trend direction, how do your determine if it's time to buy or sell?

There are many forms of trend investing which are acceptable. Although the tools might be different, they all work in a similar fashion with similar effects. The biggest variable is your chosen time cycle. Longer cycles mean less trading, shorter cycles mean more trading.

Common Trend Metrics

There are several common trend metrics that are used to determine if the price trend is up or down. Moving averages, simple price breakouts, and time-series lookback momentum are all popular. A trader can get much more technical, but the benefits start to become a lot less clear. As a general rule, simple metrics tend to perform better over the long term.

Moving Averages

Moving averages are a simple tool that can be found on many brokerage charts and Yahoo! Finance. A moving average is simply a calculated average of the price over a specified number of past trading days. Using the closing price of each day is most common.

While there are many forms of moving averages, simple moving averages (SMA) and exponential moving averages (EMA) are the two most common measures. A simple moving average is a straightforward calculation. For example, you use a 5-day simple moving average of XYZ. On Day 1 the closing price was $8, on Day 2 it was $9, on Day 3 it was $8.50, on Day 4 it was $9.25, and on Day 5 it was $10. Your simple moving average would be ($8+$9+$8.50+$9.25+$10) / 5 = $8.95.

Exponential moving averages use a mathematical formula to apply more weight to recent days. This causes the EMA to be higher than the SMA if the price is moving up and lower than the SMA if the price was moving down. In our 5-day example, the EMA would probably land around $9.25. This makes the exponential moving average somewhat more responsive than the simple moving average.

Most traders using a moving average strategy would put in stop losses on each position at the SMA or EMA number so the system automatically sells the position for them if the price drops below the SMA or EMA. They would update the stop loss price at the end of each trading day or week.

Here’s a chart of XIC.TO with a 200-day Simple Moving Average (shown in purple):

Source: TheRichMoose.com, Yahoo! Finance

As you can see, you would have purchased a position in XIC.TO on September 20, 2017 when the price crossed $24.40 per unit. Your current stop-loss price on that position would be $24.73.

Here’s the same chart of XIC.TO with a 200-day Exponential Moving Average (shown in red):

Source: TheRichMoose.com, Yahoo! Finance

You would have purchased a position in XIC.TO on September 14, 2017 when the price crossed $24.07 per unit. Your current stop loss would be $24.82.

Moving average strategies can cause a lot of buying and selling action when the price hovers around the moving average line. For this reason, when a position is initially entered a trader will drop the stop loss price to 5-10% below their purchase price to try avoid these frequent whip-saw trades.

Simple Price Breakouts

Breakouts can be very complex, especially combined with other tools. For this purpose though, we’ll use a simple price breakout. It can also be called a “box breakout”. A simple price breakout would be the price making a new high, or a new low, based on a specified look back period.

For example, let’s say you are tracking an ETF where the highest price in the last 6 months is $20 and the lowest price is $15. Using a breakout strategy, you would buy the ETF if the price exceeded $20. You would sell the ETF, or even short sell the ETF, if the price falls below $15.

Most traders using a breakout strategy would put in stop losses at their sell points. They would update those stop loss prices at the end of each trading day or week.

Here’s a chart example of breakout prices using XIC.TO using a 6 month lookback:

Source: TheRichMoose.com, Yahoo! Finance

The breakout prices are shown by the black lines. If you are not already in the position, you would enter a new position if the price exceeds $26.16 (the highest price in the last 6 months). If you have a position on, your stop loss price would be set at $23.69 (the lowest price in the last 6 months). You might also short sell a position if the price falls below that number to try profit from a further price drop.

Since the difference between the high price and low price can be quite large depending on your lookback period, many traders will set their initial stop loss price at 5-10% below their purchase price to avoid a large drawdown on that position.

Just as with the moving average, the lookback period is carried forward from day to day or week to week.

Time-series Lookback Momentum

The last strategy we’ll look at might be the most simple form of momentum investing. It forms the basis for the Dual Momentum strategy that I share on this blog and was popularized by Gary Antonacci.

In this metric, you buy a position if the current price is higher than a prior price specified in your lookback period. You sell—or take a short position—if the price is lower. You would only look at the price once a week, once a month, or once every two months. Generally a stop loss is not used.

This strategy differs from a breakout strategy—aka box strategy—because time-series lookback ignores the price movements that occurred in between the current price and the lookback price.

Here’s a chart example using XIC.TO with a 6 month lookback:

Source: TheRichMoose.com, Yahoo! Finance

Since the price 6 months ago was exactly $24.02, you would buy a position if the price was higher than $24.02 and you would sell your position if the price was below $24.02. What happened in between is completely ignored.

Using the Trend to Make Trades

Getting into (or out of) a position is the easiest part of trend investing. The key to remember is the highest potential return/lowest drawdown occurs at the first signal. After that, the return potential of that trade will drop.

Getting into a position sometime after the signal is not catastrophic—it can still be very profitable if the trend goes for a long time. Using a less-frequent system will reduce whip-saw trades as the trend solidifies.

You will enter a position when the signal first occurs based on your system parameters. Those system parameters might have you check your signals daily, weekly, bi-weekly, monthly, etc.

If you trade only once a week on Fridays, you will enter a position if the price is higher than the trend tool signal on Friday. If you trade once a month, as in Dual Momentum, you will buy only if the price is determined to be in an uptrend when you log into your account that month.

It might not be worthwhile moving into a position if the trend signal has been positive for some time. For example, using the 6-month breakout system, you would have entered a trade in Canadian stocks (XIC.TO) on October 11, 2017 at a price of $25.02 (the previous 6-month high on April 25, 2017). It might now be worth entering anymore—three months after the signal—now that the price is 3.6% higher and the sell signal (last 6-month low) is $23.69.

The Effects of Time Cycles

The frequency of trading will largely depend on the time cycle you choose for your trend investing strategy.

If you choose very long time cycles, such as 10 months (200-day moving average) or more, you will not be trading very often. Many positions will be turned over just once or twice a year—possibly less. Longer time cycles also reduce whip-saw costs. Don't go too long though, even 18 months is probably too long for measuring time cycles in this format.

If you choose medium time cycles, such as 6 months or 3 months (150-day or 100-day moving average), you will trade more often. You can expect most positions to be traded several times a year. This can somewhat increase your whip-saw costs while reducing your drawdowns on a position.

Short time cycles, such as 1 month, 1 week, 1 day, or less, will result in frequent trading. This is day trader territory. Most positions will be traded daily or even many times throughout the day. Not only will your commission costs start to eat into any profits, you will be whip-sawed frequently. You may also be considered a trading business by the CRA and be taxed at higher tax rates. I will not be referring to short time cycles when speaking about trend investing in this blog as I find these systems to be foolish.

The length of time cycle chose will determine the responsiveness of your trading system. The shorter the time frame (to a limit), the lower your drawdowns will be. This is because the spread between entry prices and exit prices will be tighter. Long time cycles mean you will buy at higher prices and sell at lower prices, assuming you don’t short sell. Surprisingly, this doesn’t really matter; longer time cycles are often just as profitable as medium time cycles because of the reduced whip-saw effect.

Comments & Questions

All comments are moderated before being posted for public viewing. Please don't send in multiple comments if yours doesn't appear right away. It can take up to 24 hours before comments are posted.

Comments containing promo links or "trolling" will not be posted. Comments with profane language or those which reveal personal information will be edited by moderator.