Posted on

Dual Momentum System on iShares Country Funds

With reference to and apologies for plagiarism to Gary Antonacci

I have recently run tests on a simple trend following system on US shares going back to 1870 and UK shares going back to 1753

Simple System US Markets Back to 1870

UK Markets back to 1753

The “binary” system looks at momentum over a given lookback for the stock market, short term bonds and intermediate term bonds. Here, the lookback used is 1 month. For the following month, the investment will be wholly in that instrument with the greatest lookback.

The 50/50 system takes the same lookback for the same three instruments and invests in the top 2 instruments for the following month. Thus the 50/50 system is never more than 50% invested in risk assets (stocks) at the beginning of each month.

My conclusion on the deep history tests of the “binary” approach is that such an asset allocation system has shown success in the past but that the level of maximum drawdown can be far higher than might be expected from tests run on recent data. Testing over the periods of the Great Crash of 1929 and the South Sea Bubble of 1721 showed that such a system is not a “get out of jail free card”. It might protect you from the worst of an equity collapse for much of the time but there will be occasions where it will not.

Modern commentators are fixed on the recent past. In my opinion this is grossly misleading. Having said that, as long as you are aware that such a system is not bombproof, it is about as good as you are going to get. There is absolutely no guarantee that momentum will continue to work going forward; equally there is no evidence that it will not. In the past it has often acted to protect an investor from the worst of an equity market drawdown (at least in back testing) and it may prove an effective method in the future.

The 50/50 approach goes one step further: by accepting a return somewhere between pure bonds and the “binary” approach one might reasonable hope for a lower drawdown and lower volatility by ensuring that each month the equity element of the portfolio is cut back to zero or a maximum of 50% of the portfolio’s value.

You can of course experiment with any other proportion.

As can be seen from the amalgamated test results below, for the assets chosen and the period for which asset history is available a very simple asset allocation such as the “binary” or the 50/50 system seems to hold promise when compared to the very steep drawdowns of Buy and Hold and the commensurate volatility.

Risk assets are represented by the following iShares Country Funds:


Reserve assets are represented by the following short term and medium term US government bond fund ETFs:


Here is what I did: I ran a buy and hold “test” to assess the risk return metrics of simply buying and holding each of the 18 country funds and the two bond funds. You will find the aggregated results set out below. Not a pretty picture: on average the return for the equity funds was 7.65% for the period with a huge average maximum drawdown of over 63%.

For the binary tests, I ran 18 separate tests – one for each of the country funds; and then produced aggregate figures for the results. For each test I coupled the relevant country fund with IEF and SHY. I repeated the 18 separate tests for the 50/50 system and again you will find the aggregate averaged results set out below.

The results are most satisfactory compared to simple Buy and Hold.

Start Date: 30th July 2002

End Date: 15th July 2016

On a Buy and Hold basis, a collective view of the risk reward metrics of the country funds is as follows:

18 iShares  Buy and Hold  
 Country Funds
mean 7.65 63.14 21.50
median 7.75 63.69 22.04
min 1.06 51.62 14.26
max 12.12 75.69 26.95


The reserve assets of a Buy and Hold Basis performed as follows:

IEF 5.81 10.4 6.66
SHY 2.22 2.23 1.39


A collective view of the risk reward metrics of the country funds invested in through the “binary” system using a 1-month lookback is as follows:

1m LB
mean 9.64 26.98 13.89
median 9.24 27.99 13.88
min 5.56 15.93 9.68
max 16.28 41.41 17.04


A collective view of the risk reward metrics of the country funds invested in through the “50/50” system using a 1-month lookback is as follows:

1m LB
mean 6.86 12.73 7.61
median 6.88 12.95 7.63
min 4.84 8.79 5.35
max 9.26 16.34 9.15





Leave a Reply