Timely and worthwhile read for those using or considering the Dual Momentum approach:https://blog.thinknewfound.com/2019/01/fragility-case-study-...Among do-it-yourself tactical investors, Gary Antonacci’s Dual Momentum is the strategy we tend to see implemented the most. The Dual Momentum approach is simple: by combining both relative momentum and absolute momentum (i.e. trend following), Dual Momentum seeks to rotate into areas of relative strength while preserving the flexibility to shift entirely to safety assets (e.g. short-term U.S. Treasury bills) during periods of pervasive, negative trends.In our experience, the precise implementation of Dual Momentum tends to vary (with various bells-and-whistles applied) from practitioner to practitioner. The most popular benchmark model, however, is the Global Equities Momentum (“GEM”), with some variation of Dual Momentum Sector Rotation (“DMSR”) a close second........When a portfolio’s returns are highly sensitive to its specification – i.e. slight variation in returns or model parameters lead to dramatically different return profiles – we label the strategy as fragile.In this brief commentary, we will use the Global Equities Momentum (“GEM”) strategy as a case study in fragility.
Nice little article.I think he's wrong about this family of strategies lacking mean reversion, though.There are subtly different types of mean reversion one could talk about.He seems to be making the tacit assumption that market movements are random walks: zero serial autorrelation.Relatively few adherents of trend following start with that as an axiom.If one posits that market trends do exist at some range of temporal frequencies and those choice frequencies are somewhat stable over long time periods, then it follows that each strategy (GEM with each different lookback in months) will have a certain very very long run average level of outperformance relative to market.Observing any one such lookback through time, the performance relative to market in any given year will likely be seen to have a somewhat bell-shaped distribution.The more samples you take, the more your average level of outperformance would be expected to approach that number.So, though it's true that a bad year is no more likely to be followed by a good year just because the first bit was bad,the existence of a very very long run mean level of outperformance means that over time one *should*expect mean reversion towards that lookback's level of improved CAGR relative to market.In short, a really good year isn't to be expected after a really bad year. No short term mean reversion.But really good years (relative to market) are expected to be as common as really bad years. Slightly more so in fact, if it's a bell curve centred around a long run average positive level of outperformance.So, pretty strong long term mean reversion in relative-to-market performance is expected.If you just had a really bad year relative to market, or even if you didn't, you're likely to see a pretty darned good year relative to market in the next decade.Those will all feed into some long run average level of outperformance, which we expect to slowly converge on that strategy's "true" number.If the overall strategy is a bad one, that long run average outperformance might be tiny or zero.That's a much more interesting question, but a hard one to answer.I'm somewhat hesitant to put too much faith in momentum strategies that that worked in the last few bears, but not before.Jim
At a shallower level, I'm skeptical about how practical his proposed approach would be for individual mechanically inclined asset allocators (the few crazy people that try this stuff) to do. I liken the workaround to a dozens approach. Instead of investing based on re;ationship to one moving average, you'd be investing in based on what, 6 slices at a time, potentially up to 6x the trades?slice 1 5m MA - USslice 2 6m MA - Europeslice 3 vs 7m MA - Treasuriesslice 4 vs 8m MA - USetc
I've only skimmed this, but it is Gary's response to the fragility issue:https://www.dualmomentum.net/2019/01/whither-fragility-dual-...
At a shallower level, I'm skeptical about how practical his proposed approach would be for individual mechanically inclined asset allocators I read the first half of his paper and skimmed the rest. Pretty ho-hum. Different values of parameter have different results. Next up: water is wet--panic now.Different start dates have different results. I kinda figured that out the first time I ran a screen backtest at portfolio123.Heck, we all know that sometimes even doing a screen's purchases with a 1 hour time difference will have different outcomes.You know, it doesn't have to be all one thing or nothing. I run GEM strategy in a few distinct accounts and I have them on different cycles. Also have some of them doing monthly (end-of-month) and some of them on 1st Monday of the month, 2nd Monday, etc. By happenstance, they don't all have the same lookback period. Some 7 months, some 10 months, etc.
... 6 slices at a time, potentially up to 6x the trades?I don't think that's a big worry.There are only so many choices, and you only do the NET trades.If one lookback is shifting from US equities to bonds and another is switching from non-US to US equities, there is no trade for the US equity holdings.Plus, each trade would be 1/6 as big, so 6 times the trades isn't 6 times the cost, just 6 times the typing.Assuming you're with a broker that charges in proportion to the actual execution costs, as mine does.Jim
Best Of |
Favorites & Replies |
Start a New Board |
My Fool |