I like Cliff Asness. Although I'm not a fan of momentum applied to individual securities, Asness' firm, AQR Research, created the first fully disclosed momentum funds available to the general public (AMOMX,ASMOX,AIMOX). AQR's low volatility risk parity fund (AQRNX) has outperformed the other publicly available risk parity funds during the first two years of its existence. Asness even has a sense of humor, saying he has recurring nightmares about being hacked to death by a pack of rabid black swans. More money managers should have such nightmares.
In 2009, Asness and his colleagues released a working paper called "Value and Momentum Everywhere". The paper did a nice job showing that momentum works well with many different asset classes - foreign and U.S. stocks, country bonds and indices, commodities and currencies. Asness et al also tried to show how value would have performed with these same assets. Identifying value with equities is pretty straight forward; book-to-market is commonly used. However, trying find comparable value metrics for other assets is not so easily done. There are no established ways for doing this, and efforts to do so may seem rather contrived. Nevertheless, Asness et al reach the conclusion that value and momentum do well when mixed together 50/50, since the correlations between them are weak.
The problem with that conclusion (besides the non-trivial one of determining value in non-equity markets) stems from the Asness momentum results.They are not as good as the ones we get from multi-asset momentum using asset classes rather then individual equities and a combination of relative and absolute momentum.
Since determining non-equitiy value can be problematic, let's look at just equities to see what I mean. Below are the returns from January 1975 through December 2011 for the MSCI U.S. Value and MSCI EAFE Value indices, as well as a dual momentum strategy that is long either U.S. Treasury bills, the MSCI U.S. index, or the MSCI EAFE index based on 12 month momentum. Positions are adjusted monthly. Transaction costs for momentum switches are negligible, with just 1.4 switches per year.
You can see that dual momentum easily approach outperforms the value indices. Let's now combine momentum with value as per Asness by weighting momentum 50% and each of the value indices 25%. Which results would you choose?
The superiority of momentum over value is greater when you construct multi-asset dual momentum portfolios, like we do, instead of using a portfolio that is limited to individual equities. Individual stock momentum may complement individual value stocks. But should there be value and momentum everywhere? I don't think so - not here anyway.
![.]()
In 2009, Asness and his colleagues released a working paper called "Value and Momentum Everywhere". The paper did a nice job showing that momentum works well with many different asset classes - foreign and U.S. stocks, country bonds and indices, commodities and currencies. Asness et al also tried to show how value would have performed with these same assets. Identifying value with equities is pretty straight forward; book-to-market is commonly used. However, trying find comparable value metrics for other assets is not so easily done. There are no established ways for doing this, and efforts to do so may seem rather contrived. Nevertheless, Asness et al reach the conclusion that value and momentum do well when mixed together 50/50, since the correlations between them are weak.
The problem with that conclusion (besides the non-trivial one of determining value in non-equity markets) stems from the Asness momentum results.They are not as good as the ones we get from multi-asset momentum using asset classes rather then individual equities and a combination of relative and absolute momentum.
Since determining non-equitiy value can be problematic, let's look at just equities to see what I mean. Below are the returns from January 1975 through December 2011 for the MSCI U.S. Value and MSCI EAFE Value indices, as well as a dual momentum strategy that is long either U.S. Treasury bills, the MSCI U.S. index, or the MSCI EAFE index based on 12 month momentum. Positions are adjusted monthly. Transaction costs for momentum switches are negligible, with just 1.4 switches per year.
Dual Momentum | U.S. Value | EAFE Value | |
Annual Return | 14.3 | 13.1 | 14.1 |
Annual Std Dev | 12.4 | 15.1 | 17.7 |
Annual Sharpe | .66 | .47 | .45 |
Max Drawdown | -26.0 | -54.6 | -58.6 |
You can see that dual momentum easily approach outperforms the value indices. Let's now combine momentum with value as per Asness by weighting momentum 50% and each of the value indices 25%. Which results would you choose?
Dual Momentum | Momentum w/ Value 50/50 | |
Annual Return | 14.3 | 13.6 |
Annual Std Dev | 12.4 | 12.3 |
Annual Sharpe | .66 | .61 |
Max Drawdown | -26.0 | -37.8 |
The superiority of momentum over value is greater when you construct multi-asset dual momentum portfolios, like we do, instead of using a portfolio that is limited to individual equities. Individual stock momentum may complement individual value stocks. But should there be value and momentum everywhere? I don't think so - not here anyway.