I decided to dig into this a bit. Note the mutual funds are not the same as the traditional GEM. What I was trying to get at was how the various momentum look backs perform comparatively. Look backs are 3/6/12 months and always weighted .34/.33/.33 respectively. In columns J & K, J is weighted by lookback length and K is weighted by % of portfolio dedicated to that lookback traded independently. So when adding up the annual return of column K that is simply the sum of columns G*.34/H*.33/I*.33. The bottom row is the cumulative return and column M is if column J exceeded column k (which it did 13 out of 23 times. (I did not count 2019). Columns N and O are what 10K grew to.
https://www.dropbox.com/s/01491e5fq4rn2 ... 3.PNG?dl=0
I then performed a rolling 5 year summation of annual returns which I think is highly informative. You will note a couple of things. 1) Which single lookback is best clearly changes over time. 2) Equal weighted lookback averaging as compared to running a third each of your portfolio performs quite differently. Averaging lookbacks is clearly "streaky" when it comes to performance and tends to be correlated with the LT side of the equation whereas if you are looking for "the average" segregated is clearly the better way to go.
https://www.dropbox.com/s/lkf0but3qne94 ... 2.PNG?dl=0
Turning to behavioral finance...if you go with equal weighted lookbacks you are clearly going to have regrets when 3mo momentum is outperforming. take a look at 2009-2013. 12mo mo and equal weighted lookback massively underperformed 3mo. Can you psychologically do that for five years running. If you go with a segregated portfolio. You will have regrets every year if your bogey is the best lookback. Alternatively, you will be happy every year if your bogey is the worst lookback. Can you psychologically be "average" every single year? Additionally, one has to be extremely cautious of recency bias when looking at the data. It goes without saying that if 3 months is in the mix this portion is going to trade a lot. I looked at just the 3mo trades and they were pretty much double that of the 3/6/12 weighted option.
However, the happy news is that DM outpeformed all assets and an equal weight of all assets. So every option was "good."
My personal take: In a taxable account a good argument is to be made that anything longer term is better. However, you have to trade
less than annually to take advantage of slower trading and in most years, you are going to trade at least once which means a taxable event. In a retirement account then it's probably whatever you feel good about...however, when withdrawing money from your retirement account I don't think there is any question which is superior...segregated.
I'm glad I did this. I will be acting on it as to me there is a pretty compelling story here.