9. Empirical Results
9.5 Practical Observations Regarding the Momentum Strategies
months, which is the highest of any of the strategies. Among the winner strategies there is a clear tendency that those with a longer formation period experience relatively more months with positive returns, which acts to support previous indications that winner strategies with longer formation periods were preferable. However, the same pattern is not clear among the zero-‐cost portfolios. In general, it is difficult to deduce anything meaningful from the zero-‐cost strategies in the table, partly because there is less of a spread among the strategies. Thus, for now we will just conclude what was likely already expected, which is that there seems to be a positive correlation between the winner portfolios average monthly return and the number of months in which the strategies create a positive return. Further, this indicates that longer formation periods create better returns.
Table 9.11: Total return (Equally weighted and adjusted for transaction costs)
The table shows the total return for the winner and zero-‐cost portfolios for all 16 distinct strategies. The total returns are based on an equal weighting scheme, with partial monthly rebalancing and reinvestment of profits and losses.
Furthermore, the returns have been adjusted to account for transaction costs.
*Total return on benchmark from 2000-‐2016 was 387%
For most investors, the number of utmost importance is the return on their
investment by the end of the investment period, but the average returns presented earlier doesn’t offer any insight into this matter. In table 9.11 the total return accumulated over the 16-‐year period of analysis is shown. Not surprisingly, the strategies with the highest monthly average return are also among the strategies with the highest total return. And just like the results in
3 6 9 12
3Winner 645% 634% 665% 755%
Zero-‐cost 319% 392% 495% 496%
6Winner 991% 1276% 1194% 955%
Zero-‐cost 425% 714% 777% 542%
9Winner 1714% 1618% 1377% 1415%
Zero-‐cost 888% 962% 824% 515%
12Winner 2008% 1591% 1879% 1373%
Zero-‐cost 774% 697% 617% 430%
Holding (K)
Formation (J)
previous sub-‐sections, the table shows a relatively large spread between the return on the different strategies, clearly indicating that some strategies outperform others. The 12/3-‐winner portfolio takes the top spot with a staggering total return of 2008% after transaction cost. In comparison the benchmark earns a total return of 387%. Impressively enough, all winner portfolios manage to earn a total return substantially larger than the benchmark. Most of the zero-‐cost portfolios also perform well compared to the benchmark. The 9/6-‐zero-‐cost portfolio earns a total return of 962%, which makes it the best zero-‐cost strategy. Almost every zero-‐cost strategy accumulates a total return higher than the benchmark; only the 3/3-‐zero-‐cost strategy earns less over the course of the analysis period.
By now it’s more than evident that some of the strategies produce great results. However, it is still not clear how the strategies developed over time. Did the total return accumulate steadily, or did the high average monthly returns mainly occur due to extreme positive outliers? To answer these questions, we will look at the return as it accumulated in event time.
In figure 9.2 the total return on three selected strategies based on the winner portfolios is shown in event time together with the benchmark. The three strategies are selected because they have different formation period lengths, which previously proved to partly explain differences in returns, and because these strategies were top performers. The figure clearly illustrates the impressive performance of the strategies, especially in relation to the benchmark. Moreover, the strategies seem to be performing particularly well in upward trending markets, like in the mid 2000’s and from 2012 and onwards. In these periods the strategies noticeably distance themselves from the benchmark. However, when the market drops, so does the return on the strategies. In the months of the subprime crisis the benchmark dropped by around 65%. Surprisingly though, 2 out of the 3 strategies only dropped by around 45%, indicating that the momentum strategies don’t necessarily loose as much as the average market when things go down. With that said, the 12/3-‐strategy dropped by around 60%. Furthermore, all the selected winner strategies dropped to the same level during the subprime crisis, which is puzzling.
Figure 9.2: Return over time of winner portfolios (Index: 1 = Strategy start)
The figure shows the development in total return for the winner portfolios for 3 distinct strategies. The returns are based on an equal weighting scheme, with partial monthly rebalancing and reinvestment of profits and losses.
Furthermore, the returns have been adjusted to account for transaction costs. The benchmark shows the return of a buy and hold strategy holding the Danish OMXC index.
The zero-‐cost portfolios depictured in figure 9.3 showed that, in the bull years leading up to the subprime crisis, the zero-‐cost strategies perform worse than the
benchmark. This relatively bad performance is due to the loser portfolios creating impressively good returns from 2005 to 2006, decreasing the profitability of the zero-‐cost strategies. When the market drops, the return on the zero-‐cost portfolios doesn’t follow along immediately. This tendency can be seen in the figure all the way up until march 2009; 7 months after the subprime crisis struck the stock market. At this point, the zero-‐cost strategies fall with 40%–60% the
following months, which makes the total return on the 9/3 and 12/3-‐strategy turn negative (below index 1). However, as the market begins to recover, so does the zero-‐cost strategies, although each at a different pace. Throughout the period of analysis, the 6/6-‐strategy performs the best for
a long stretch of time, but in the last year of the analysis, the two other strategies generate some impressive returns and even end up surpassing the 6/6-‐strategy’s performance.
Figure 9.3: Return over time on zero-‐cost portfolios (Index: 1 = Strategy start)
The figure shows the development in total return for the zero-‐cost portfolios for 3 distinct strategies. The returns are based on an equal weighting scheme, with partial monthly rebalancing and reinvestment of profits and losses.
Furthermore, the returns have been adjusted to account for transaction costs. The benchmark shows the return of a buy and hold strategy holding the Danish OMXC index
Consequently, the success of the momentum strategies cannot be attributed to steady and continuous positive return rates throughout the analysis period. The analysis shows that the strategies have periods with positive returns, some with negative returns, and some periods where the returns stagnate. The strategies’ biggest and most obvious strength is that they manage to generate some impressively high returns when evaluated over a long-‐term horizon.
Furthermore, the strategies seem to be performing just as well, and maybe even slightly better, compared to the general market when things go down.