I recently received a message asking whether I thought that the Dual Momentum (DM) Model was less effective than other systems.
First let me define how I interpret “Dual Momentum Model”. In this case I interpret it in the same way as used by Gary Antonacci in his book “Dual Momentum Investing” i.e. the selection of the single best performing asset, as determined by measuring “momentum” (Rate-Of-Change of price), from a short list of (usually three) assets. The “Dual” aspect comes from a consideration of both “absolute momentum” or whether prices are moving up or down over the look-back period and “relative momentum” or whether prices are moving up or down faster or slower than other assets being considered.
In his book, Antonacci presents evidence to support the efficacy of this system (with a single 12-month look-back measure of momentum) over a 40-year period – and the evidence is compelling that, over the long-term, the model/system works very well. Of course, the back-tests were based on End-Of-Month review/adjustments and do not consider timing/review date “luck” – but it is difficult to refute the validity of the conclusion that the Dual Momentum Model performs well over the long term.
Antonacci’s back-tests show that, using his suggested 12-month look-back period, we should be prepared to see an ~18% draw-down using the DM investment model.
At ITA Wealth, since many of us tend to be “in our advanced years” 🙂 , we have become sensitive to shorter-term performance and the preservation of capital – and this presents us with a problem – can we reduce draw-down/risk without significantly reducing returns?
To address this question we have taken the basic concepts of Dual Momentum and modified them to:
- Apply multiple look-back periods to measure “momentum” (and added a volatility component);
- Allow more diversification by selecting more than one asset (class);
- Use alternative methods to measure/calculate “momentum”;
- Use other indicators/signals to complement the basic “momentum” measurements.
As a simple example, the Kahneman-Tversky Portfolio uses Dual Momentum to analyze a 3 asset portfolio but provides diversification through the use of different look-back periods and allowing more than one asset to be held in the portfolio.
However, at the heart of all these systems we apply (or at least consider) the basic concepts of Dual Momentum i.e. absolute and relative momentum. Therefore, maybe all systems could be classified as Dual Momentum Models.
Irrespective of what system/model we might be using let’s take a look at what we might have expected from market performance year-to-date (YTD).
Let’s start with the broad US equity markets as represented by the S&P 500 Index (SPX):
- Over the past 10 months the SPX has traded at or within 5% of All-time highs (grey shaded area) for ~one-third of the time – quite impressive;
- We have seen two ~10% corrections in the period (February and October) and a trough-to-peak move of ~16% over 7 months;
- The 200-day SMA provided support for most of the year but price has recently broken through this support and is re-testing it as possible resistance;
- The year started with SPX at ~2700 (blue line) – and is presently sitting at 2781 – up 3% YTD (blue line reference);
So, what might we (as investors) expect to get out of this? Even with hindsight, 3% on a buy-and-hold strategy is probably about as good as we could reasonably expect. The pullbacks were so rapid that no “investment” strategy is going to pick this up – even short-term “traders” probably had problems – and so it would have been difficult for any (long-only) strategy to avoid these losses. Obviously, this leaves the question of how much of the trough-to-peak move could be picked up and this will depend on the amount of confirmation considered desirable to identify a change in trend. A reasonable goal might be to pick up at least one-third of a significant trend (without unacceptable draw-downs).
The next broad asset class might be International Equities – so let’s take a look at the performance of VEU:
Not too much to be said here – it’s been downhill all the way since the end of January and prices are currently ~15% lower than where they were at the beginning of the year.
So, maybe Bonds would have been a better place to be due to their inverse correlation with equities?
Well, hardly – Bonds (as represented by the AGG Aggregate Bond Fund) has not had it’s head above water all year and is down ~4.5% YTD. Other Bond Funds show similar trends.
So, there goes the major asset classes for simple Antonacci-style DM portfolios – but I suppose money must be going somewhere – maybe Real Estate? Let’s take a look at VNQ (US REIT’s):
VNQ has had it’s moments and moved up ~17% (comparable to SPX) between February and August but is down ~3% YTD. Momentum systems with long look-back periods are going to have difficulty picking up these short-term price changes.
Would any other asset classes have helped us out over the past ~10 months? Let’s take a look at Commodities (DBC):
This one probably flew under the radar but stayed mainly ahead for most of the year moving up ~13% over the first 9 months before falling back recently and currently sitting unchanged YTD.
All this is a little depressing – with no obvious winners YTD – so why would we expect any long-only intermediate/long-term investment system to generate significant returns over this period? A system – any system – can only generate positive returns if allowed to by market price movement and, if there is no (average) movement, we cannot generate returns that aren’t there.
Ok, so that’s the first simple (or not-so-simple) answer – but, are there other things we can do to “improve” performance? Well, there are other things to consider that may/may not be “system/model” related. For example, the composition of the portfolio asset list may change “system/model” performance significantly – but this is generally not a reflection of the system but on the “art” of portfolio construction. For example, the healthcare sector (XLV) has seen a ~21.5% move from the lows to the highs and is currently trading up ~12% YTD…
.… so an “artful” investor might have picked up nice gains here. But I think we are a long way from the original question related to the DM model – single best performing asset chosen from a small list of broad market assets (classes) – and whether there are better systems/models.
As far as different systems/models are concerned we can “fine-tune” the system to improve risk-adjusted returns (maybe accepting lower returns in exchange for lower volatility and/or smaller draw-downs). This might be done through control of the number of assets to be held in the portfolio that, in turn, might be influenced by the diversification/correlation between assets under consideration.
Bottom line – for investors looking for a simple investment model I do not think that this is the time to throw out the DM Model just because performance YTD has not been positive. I doubt that many money managers or investors are showing outstanding returns YTD – especially using “simple” systems focused on an intermediate to long-term investment horizon. Even Hedge Fund managers using advanced algorithmic trading methods are struggling right now.
If an investor is breaking even, YTD, this is probably pretty good – and better than most investment advisers are doing – and if the market goes into a bearish dive the DM model should keep us out of harms way.
In future posts we (Herb and I) will be reporting on recent back-tests that we have been running on the latest modifications to the LRPC (momentum) system and I will try to remember to highlight a few observations that come out of those tests that have relevance to the issues raised by the question that stimulated this post.