Studies of Life

Learning by doing.

Strategy Test 2015 – 4: March 31

01 April 2015 by Jim

It’s been a full quarter since I started testing a few different portfolio strategies as explained in this post. Here’s the next update.

strategy test

The first quarter overall was very positive, as you can see. The best performer is the all-stock Trending Value portfolio, the worst is the risk parity approach although it does what it is supposed to: it reduces risk, as is evident by the smooth curve. AAA is also very smooth so far. DM and Trending Value are the most volatile.

Here are the results for Sharpe and total return:

sharpe and return

For Sharpe, the best portfolio strategies this year are:

  • Trending Value / X (=my own portfolio, basically 65% TV + 35% bonds & gold & cash)

* Adaptive Asset Allocation

  • 60/40

Surprisingly, Risk Parity did not do as well in this regard.

In terms of return, the winners are the same ones as for Sharpe, but the Permanent Portfolio also did very well (2nd place in terms of return).


The most obvious from these stats so far are, in my opinion, that two strategies are clearly superior: my own portfolio (called ‘X’ in the graph, basically 65% TV + 35% bonds & gold & cash) and pure Trending Value. So the Trending Value method clearly seems to have an edge, both in terms of return and risk. That’s great because it is among the most tedious to implement so it’s nice to see a reward for that work. Here’s the evolution of my own portfolio:

own portfolio

After TV, the best choice so far seems to be AAA and a simple 60/40 portfolio. If this remains this way for the rest of the year (and in the future) then AAA will be useless, because it’s also very tedious to implement. So if a simple 60/40 allocation using 2 ETFs does the same thing, there is no use to stick with AAA. But we’ll see. It is, by the looks of the portfolio evolution below, the smoothest of all so far.


Compare this with Dual Momentum and Risk Parity, which are much more jagged:

Dual Momentum

Risk Parity

8 comments | Categories: Investing, investment, portfolio | Tags: , ,

Comments (8)

  1. Hi Jim

    Thanks for the interesting blog. I’ve one observation which you may want to think about in future.

    I do not think that a comparison of ETFs in a dual momentum system is strictly comparable to a trending value system which picks individual stocks. While your implementation of each is similar to the intention of the original proposers of these systems, I suspect that were you to use a dual momentum system on individual stocks, this would give a much better comparison to the trending value system.

    In my own system tests I have found that individual stocks in most cases always beat (by some margin), ETFs or Mutual Funds in a dual momentum system.Having said that I do not implement value in any of my systems, simply pure momentum as I find the value implementation very tedious and time consuming, and pure momentum gives me excess returns and sufficient risk control.



    • Thanks for your comment, Jay! I suspect that you may well be right that, in a bull market as the current one, using dual momentum on individual stocks will yield better results than applying it to ETFs. Using dual momentum with individual stocks can definitely be part of a good portfolio, but I would not use it as the single portfolio component due to the inherent volatility and possibility of big drawdowns. Since ETFs of sectors or asset classes or countries are automatically diversified to some degree, I feel that having a dual momentum 5 ETF portfolio is safer than a 5 stock portfolio.

      It’s great to hear that you’re having good results with poor momentum. I am relatively new to investing and have, in terms of stocks, entered the arena in 2012 so I have only seen a bull market so far. I suppose momentum works great there, but I’m also kind of worried about what crashes would do to a pure momentum strategy, especially since I haven’t been investing for a full bull-bear cycle so I don’t yet have any experience with that. That’s why I am trying to play it safe to some degree.

      But I would definitely like to test a dual momentum individual stock strategy. How do you think that would look with a large number of stocks, i.e. 20-30, chosen purely based on dual momentum?

      • Jim,

        I only apply dual momentum to a choice from 200+ funds/stocks as I have found that the best results and least risk occurs when the sample size is large, very large. I then invest in the top decile, rebalancing monthly. I use a 100 day SMA filter as a further filter and go to cash for the portion in the top decile not above the 100 SMA which happens very seldom in a bull period. The 200 day SMA works okay when you are picking only 2-5 funds but with stocks and a large selection sample I find that 100 SMA is better.

        I find also that because of the large sample of stocks/funds that the 100 SMA serves as a good stop in faltering markets and gets one out at mostly the right times without too many whipsaws. I run 2 systems, one which trades the top decile monthly of a sample of 250 GBP mutual funds, and a second which trades the top 350 stock in the UK. I do the same for family who live in South Africa on the top 200 stocks there.

        When I find time I’ll summarise the backtests and the live results and share those with you, also how they performed back in 2007-2009.


        • Wow, I was not expecting such an extensive answer. It would be great to see your data on this. Is the monthly rebalancing actually necessary? I suppose it helps to smooth returns, but if you rebalance 35 stocks every month that would, in my case, make for a lot of commissions. In my own portfolio, I rebalance once per year and only between asset classes (i.e. I want 50, 60 or 70% stocks vs bonds) but not individual stocks to avoid paying too much in terms of commissions. I do however buy a new set of stocks every once in a while to freshen up the stock portion, but I don’t do so more than once in 6 months.

  2. For my UK mutual fund system I have found that monthly rebalancing is very necessary. I use as broker and there is no charge to switch so the monthly rebalance is frictionless apart from the time to process by each fund house. Some months it works to my advantage others not but nothing which causes concern. I have found that the trend effects for the top 4 picks lasts about 3-4 months but for those from 5-25 only a month, hence my need to rebalance monthly. I also do not balance 100%, only where the change on position is above 10% of the position size or where a fund falls out the top decile. As most of the top rated funds trend for some time you will usually find that changes in these are every 3-6 months, in exceptional cases 12 months. The bottom 5 funds of the top decile usually swap chairs each month. Again no costs to these changes as the swapping from one fund to another is free and in most cases one finds that getting rid of these funds is a good thing and protects one from large losses.

    For the stock system of course there is a brokerage charge and I would not recommend using a monthly rebalance unless you have at least £100k to invest, With switching costs being around £7 per change plus spread the larger the account size the less friction on performance. I generally find that as long as I can keep costs in line with a typical mutual fund charge (.75 – 1% p.a.) then costs do not effect performance too much. I also find that with stocks, as opposed to mutual funds, that the trend effects last for at least 12 months on individual stocks. Remember though that I use a 100 SMA cut off so if you did not want to rebalance monthly you could in theory only rebalance every time a stock fell out and then replaced only that stock. Not tested or tried that yet but it may actually work better – perhaps something to test!

    One only learns with real investment experience what works in practice for rebalancing and how cost effective this is. You learn in time how to take short cuts and what works best. My data is a jumble of spreadsheets at the moment. I’ll hopefully make an attempt one day to summarise these and throw them your way. In summary thought while actual performance approaches the theoretical model performance, in practice because of the reasons above some months it’s above and some months below, but never by very large margins.

    • What you’re doing sounds really mature and thought-through. When I started out investing, momentum seemed like the bad choice compared to value, for reasons like volatility and the need for more active management. That’s why so far, I did not look into it further. Since academic research does however clearly show that the momentum effect does exist, I’ve started reading more about it, and you’re the first person I’ve talked / written to that is using a systematic momentum approach. I think I will definitely be testing some of your suggestions myself. Thank you very much for your input!

Leave a Reply