Thanks for the interesting post. An interesting side topic might be "the power of an individual to shape thinking, as an accelerant (band wagoners) or retardant (skeptics) and their roles on the persistence or termination of trends.
For example, I saw former treasury secretary Paulson speak the other day and he went into TARP and the financial crisis. He seemed to feel the program, while very unpopular, really saved the economy. It is possible, one or two people being unwilling to buck popular notions (bankers should not be bailed out, despite the consequences) can really swing the outcome of the global economy.
Can belief in the economics of Amazon or Twitter due to popular culture make the success happen? Interesting to consider. Many real estate deals pencil out or fail based on perceived desirability of the location.
Much food for thought. Thanks.
Thanks to
pizzasteve posts as well as links provided by
brooklynguy in this thread. They provided a lot of food for thought indeed. This is a long post. I have put together my thoughts on what a DMI investor (specifically GEM) should worry about. You need to be familiar with GEM and its methodology to understand this analysis.
Followers of DMI (specifically GEM) are attracted to this investing methodology because of its eye popping over-performance relative to B&H in backtesting. However, this excellent Philosophical Economics blog
http://www.philosophicaleconomics.com/2015/12/backtesting/ posted by
brooklynguy demonstrates that relying on back testing alone without understanding why and how GEM works is very dangerous. Also it would be unwise to take comfort in a theory that momentum anomaly is real and it is sustainable. Even if that is true, GEM is an algorithm, it has certain parameters with certain values recommended for those parameters. GEM can fail if the recommended values no longer work in the future. So one really need to understand how GEM parameters values interplayed with its context during the period of back testing. You also need to think if the environmental conditions that allowed past success will continue to persist in the future. Lets break this down to examining absolute momentum and relative momentum components of GEM.
Absolute Momentum (or Trend Following)Again another article from Philosophical Economics blog
http://www.philosophicaleconomics.com/2016/01/gtt/ (also link posted by
brooklynguy) gives a very good framework to understand the variables involved. Absolute Momentum trades down-market gains (gain relative to B&H that is) to losses (again relative to B&H) due to whipsawing. Thus, Absolute Momentum’s performance depends on extracting higher gains than losses. The parameters for implementing this strategy is monthly testing and 12 months of look back period. Both are critical - if you reduce/increase the testing frequency and/or look back periods, the returns from this strategy suffers. With these two parameters the way they are, the tug-of-war between gains and losses of this strategy breaks in favor of gains. Given the above background, we can draw the following observations about Absolute Momentum’s promise going forward.
- Duration of business cycle: Lets say there are no business cycles that cause bull-bear market cycles. Then Absolute Momentum will underperform B&H. So performance of Absolute Momentum is inversely related to the average cycle time of one full business cycle (within reason of course - if the cycles are too short - say less than a couple of years, this won’t work either).
- Depth of business cycles: The distance between peak to trough, the amplitude of the cycle is important. If this amplitude is shallower than historical averages, Absolute Momentum will underperform back testing promise. At a certain threshold, it will underperform B&H as well. This depth is proportional to momentum in the market. Herd mentality will drive the valuations up in a bull market and depresses them in bear markets. If momentum is arbitraged away in future, markets stay perfectly valued all the time. Obviously, that is NOT good for DM.
- If the medium term (monthly time frame) volatility becomes much higher than historical averages the performance degrades due to increase in whipsaw losses.
- If the market-drops at the end of bull markets and market-raises at the beginning of bull market are steeper than historical averages, then performance degrades. The test frequency and look back periods that worked in the past won’t work as well.
- If US is no longer the leader of world markets in terms of leading world markets up/down, Absolute Momentum may not work as well in the future. The reason for this is GEM uses exclusively US market performance to get out of stocks.
Relative Momentum - If US and World ex US are not about the same size overall, the performance degrades. GEM depends on both them to be large so that they have enough, but just enough volatility in them. If you want to divide the entire world market into two, you get largest minimum size when the entire market is divided into equal pieces. If you take it extreme, if US becomes very small, it may be too volatile to make this work. That is why sector rotation underperforms GEM (because sector indexes are too volatile)
- There is a phase difference between US and world economic cycles and in terms of their monetary and fiscal policies. One of the main reasons GEM is able to milk about 300 bps from Relative Momentum is not because US and world are not correlated, but because of dollar’s raise and fall. This phase difference in cycles must continue in the future for Relative Momentum's promise to be fulfilled.
Most people who do not fully understand GEM dismiss this as a market timing strategy. It IS based on marketing timing. However, it is also a very cleverly designed algorithm to capture momentum anomaly. Whether or not it will work in the future depends whether or not if we expect to have above outlined conditions of the past persist into future.
What I claim here is that the conditions outlines above captures most of the uncertainty with regard to GEM future performance. Can you think of any other factors?
How would you rank these risk factors?
My rank from top to bottom: AM-3, AM-4, AM-5, RM-1, RM-2, AM-1, AM-2
(AM - Absolute Momentum, RM - Relative Momentum, 1-5 refer to the numbered list in the corresponding sections)
How big are these risks? Based on this analysis, are you more confident or less confident that GEM can hold its promise?