asset management

Dual Momentum Analysis

J. González

23/02/2017

No Comments

Why dual momentum? Because strategies based on highest relative momentum show great results in the long run, but can experience deep falls and have little participation in the posterior rebounds after large market falls.

To sidestep these drawbacks, here’s a strategy laid out based on Gary Antonacci’s studies about Dual Momentum and Absolute Momentum, with the difference that, while he used the so-called Dual Momentum in multiassets universes (Index ETFs, Bonds, etc), the universe we are presenting is made up exclusively of stocks. We’ll also change the momentum calculation window length to improve the reactivity in rebound situations.

Results

To avoid the survivorship bias effect or liquidity inconsistencies, the tests have been made exclusively using the components of Euro Stoxx Index and S&P 500 Index in each moment.

As seen in the results of tables 1 and 2, in the strategy that selects the quintiles weekly according to 1 year stock price momentums in Europe and USA, the fifth quintile (highest momentums) achieves a better annualised performance, as well as a better Sharpe Ratio in the American case.

Table 1. Momentum quintiles with S&P 500 universe, weekly review. 1/1/1991 – 8/5/2015

Momentum quintiles with S&P 500 universe, weekly review. 1/1/1991 – 8/5/2015

Tabla 2. Momentum quintiles with Euro Stoxx Index universe, weekly review. 1/1/2000 – 8/5/2015

Momentum quintiles with Euro Stoxx Index universe, weekly review. 1/1/2000 – 8/5/2015

You can see in the annual returns how in bearish market years, like 2008, the fifth quintile falls too, although less than the first quintile (losers quintile). The winners quintile falls behind in the recuperation years, like 2003 and 2009, while the losers quintile reacts much better.

 

Winners and loosers USA quintile            Winners and loosers europe quintil

Alternative momentum strategies

To offset the poor statistics of the high momentum strategies in years of fall and recuperation, we propose two approaches:

  • Dual momentum: when the absolute momentum is negative, the strategy disinvests. Absolute momentum refers to the momentum of the last year of each stock minus the money momentum (in this case we will consider 1 month EUR and USD interest rates respectively). The portfolio will be built with the quintile of maximum momentum, selecting only those with positive absolute momentum, leaving the remaining weight, if any, in cash (interest rates).
  • Dual momentum plus: same as the previous approach, but the weight left by the stocks with negative 1-year absolute momentum with those stocks which present the highest positive absolute momentum along the previous months. The weight not covered goes to cash.

With the first proposal, we expect to reduce the falls in bearish markets. The second should improve the rebounds after large falls.

It is shown in the tables that the dual momentum strategy has a notorious improvement in terms of risk and Sharpe ratio. On the other hand, the Dual plus strategy holds similar values of performance and risk to the pure maximum momentum strategy.
usa com
eu com

In the next graphs you can see how the disinvestment is quite punctual, happening in 2002, 2003, 2008 y 2009 in the USA, which are the troublesome years. In Europe there is desinvestment in 2001 and 2011.

cash cash1
The Dual strategy reduces the fall quite a lot in 2008, not so in 2001 and 2002 (in the USA). In Europe, in 2008 it achieved a spectacular improvement. In 2009, the annual return gets better due to a lower loss until March.

The Dual Plus strategy, on the contrary, does not achieve reduction of negative returns. It does, however, improve during market rebounds, mostly in 2009, where the improvement is considerable in both regions.

years usa            years europe

Conclusions

The two strategies proposed improve portfolios based on momentum factor.

Dual momentum: acts to reduce large draw downs, achieving better return/risk ratios.

Dual plus momentum: plays with the momentum calculation length to improve the portfolio reaction in rebound years after a market drop. On the other hand, it does not achieve risk reduction.

add a comment

wpDiscuz