Asset Class Update: Cross Market Momentum

I spend a lot of time thinking about momentum. One of the problems with the cross-market momentum approach is that it can leave you undiversified and exposed to large drawdowns due to concentration risk. ie if someone followed the approach with the asset classes below, selecting the top 10 based on 12-month returns, he would have returned about 10% for the month. However, most all of your positions would be in emerging markets.

Has anyone seen any research on cross-market momentum that selects the top X-positions from each asset class? For example, take the top two asset classes from each asset class category (top 2 from US Stocks, top 2 from Foreign Stocks, top 2 from Bonds, etc, etc)? I have not seen much research in this space.

click to enlarge

Comments

[AFJ Garner:]

I agree with your approach (I am testig this in Trading Blox) and no I have seen no such research. Interesting that Fundx does not do this - each asset class is treated separately. Trouble is, as both of us know only too well, ETF history is too short for meaningful test results. So, you either have to work with Mutual Fund data or create your own total return indices with suitable assumptions for tracking error.

2007 Nov 02 05:31 PM

[AFJ Garner:]

I agree with your approach (I am testig this in Trading Blox) and no I have seen no such research. Interesting that Fundx does not do this - each asset class is treated separately. Trouble is, as both of us know only too well, ETF history is too short for meaningful test results. So, you either have to work with Mutual Fund data or create your own total return indices with suitable assumptions for tracking error.

2007 Nov 02 05:31 PM

[AFJ Garner:]

I am also not entirely sure that I am happy working with monthly data points which no doubt downplay max DD and possibly boost CAGR. I have worked with both monthly and daily data recently and tend to prefer the latter. Which makes the data collection problem even more acute.

2007 Nov 02 05:58 PM