Like many who visit us here, we became intrigued by momentum strategies as a result of a series of articles in Seeking Alphaby Frank Grossman. Specifically his articles on ETF market rotation strategies – GMR, GMRE and Bond. We were early subscribers to his newsletter(s) and experienced favorable returns using the strategy in the latter part of 2013. Like some others, the newsletter wasn’t enough, we wanted more. So we began to run our own back tests of various performance/volatility combinations, at first merely to see if we could accurately duplicate his predictions.
If you’ve read the comments associated with the original articles, you’ll know that “duplicating” Mr Grossman’s methods proved difficult, if not, impossible, for most who tried. He provided an outline of the methodology, but as they say, the devil is in the details, and there are just too many details we do not have. While many have tried, we’ve not seen anyone who has successfully duplicated Mr Grossman’s published back test results. Some may have come close, but we’ve not seen them duplicated. We certainly have not been able to.
While we’ve been unable to duplicate Mr Grossman’s results, our test results have indicated that his strategies are certainly among the most promising of those we’ve tested. His selection of funds provides an excellent foundation on which to construct a strategy. And his strategy is augmented by timely market analysis provided to his newsletter subscribers.
Another reason we began to do our own back testing was that there were elements of several momentum-based strategies about which we had reservations. For example, to be effective, you should only trade on the first trading day of the month, based on the ranking determined using closing data from the last. While that may be the case for the back-tests used to develop the strategies, is it some particular proven market characteristic? (We were actually able to determine other selection criteria which out-performed “first day/last day” in back-tests, but similar to “first day/last day”, they did not hold up across our stress tests for all periods tested.)
Why trust our findings/results? Simple answer – DON’T. We don’t. We keep testing and refining. We’ve reached a point where we’re comfortable that our observations/results will not be totally refuted, while at the same time, hoping they will be and a consistently high performing momentum strategy will be found. Please, do your own due diligence. Our general findings relate to thousands of back tests run across dozens of time periods and make NO specific recommendation regarding a specific strategy recommendation. We believe we’ve seen enough results and systems posted since Mr Grossman’s articles to realize, beating the market can’t really be that easy. As someone asked in response to one of the published momentum-based strategies, “If it’s that easy, why isn’t it more mainstream”. In other words, why isn’t everyone doing it? Why aren’t fund managers trouncing the market by employing relatively simple, momentum-based strategies? Our response to that question is “It’s not mainstream because in the real world, you can’t trade in the past, which is where the systems have been proven to be effective.” However, since it’s so appealing, we continue trying to develop a market-beating system based on momentum. We want people to understand up-front – there is NO single best strategy, nor is there any guaranteed winning momentum strategy that we’ve been able to develop or validate.
Efficient market theory contends “future prices cannot be predicted by analyzing prices from the past. Excess returns cannot be earned in the long run by using investment strategies based on historical share prices or other historical data. Technical analysis techniques will not be able to consistently produce excess returns, though some forms of fundamental analysis may still provide excess returns.”
And while we’re talking disclaimers, we must add regarding our own back tests: our testing relied on numerous external (downloaded) and computed values based on historical ETF performance, mainly obtained from yahoo. The values used as input and/or computed, may not be statistically significant, and may be subject to extrapolation or programming error. Of course, everyone published their results as if everything is “fact”, when we’ve seen too many contradictory and questionable results to blindly accept any as “fact”.
Our first attempts were to validate Mr Grossman’s findings, as closely as we could. Using yahoo quotes and software which we developed specifically for the back tests, we ran hundreds (eventually thousands) of permutations across a 2002 – 2013 period. We later split the time periods into several periods which we found to exhibit differing characteristics to see if results/predictions from one period would hold into another with differing characteristics.
We ran “sub tests” on every candidate strategy to determine if it was good across all time periods or only some. While we will publish additional detail regarding the specific tests and results, we characterize the results as follows:
Momentum strategies rely on momentum, which is great as long as the market is exhibiting positive (or negative) momentum. However, during turbulent periods or times of very low or shifting momentum, the strategies need to be monitored more closely.
Under certain market conditions, i.e. a momentum-exhibiting market, a properly constructed and implemented momentum-based strategy may add-value over a more general buy and hold strategy. However, since no one can accurately predict which set of conditions will prevail for a given future period, momentum should not be relied upon for consistent, superior results.
There is NO best strategy. A strategy which out-performs for any specific period will under perform in other periods.
Nor have we been able to determine a performance advantage obtained by developing a strategy backtested over several decades versus tested from 2002 forward. Back-testing using “fund equivalents” across very long time frames, does not appear to offer meaningful equivalency to stress testing utilizing multiple, shorter testing periods using the actual funds employed in the strategy.