Momentum, by name, means the tendency to keep the original status. In the realm of the stock market, if the stock price is up today, and if the momentum carries on, you can predict an even higher performance the next day, and vice versa.
Momentum ETFs ought to be widely used by investors contingent on different regimes. Under some extreme circumstances such as the financial crisis in 2008, all stocks clustered and plunged, betting on any momentum strategies is equivalent to investment suicide. On the other hand, in this nascent Donald Trump era, the market has displayed somewhat up-trending momentum. NASDAQ, SP500 are hitting record highs, and more investors are pouring in money to ride the wave.
There is a handful of momentum ETFs: MTUM, PDP, ONEO, DAWS, PIE… available in the market, providing handy tools for the investors. The AuM of the above five ETFs is $2,101.7 Million, $1425.3 Million, $451.7 Million, $183.7 Million and $147.5 Million respectively.
Let’s explore MTUM, PDP and ONEO’s underlying indexes to understand and hence, be able to choose the most suitable Momentum ETF.
iShares Edge MSCI USA Momentum Factor ETF, MTUM, was launched in April 2013, with $2.83 billion assets already tied to it in June 2017, is benchmarked against the MSCI USA Momentum Index. It aims to track the performance of High Momentum companies. They are characterized in the literature as companies with high price performance in the recent history, up to 12-months.
The applicable universe includes all the existing constituents of an underlying MSCI Parent Index (herein, a “Parent Index”). The Momentum value for each security is calculated by combining recent 12-month and 6-month local price performance of the security. They are adjusted by risk and then normalized by the standard deviation to derive a Z score.
Risk-adjusted Price Momentum = Price Momentum / σi
Where σi = Annualized Standard Deviation of weekly local price returns over the period of 3 years.
Z = 6-month Momentum Z-score*0.5 + 12-month Momentum Z-score*0.5
The Momentum Z-score is winsorized at three times threshold, i.e. the Z-scores above three are capped at three and Z-scores below -3 are capped at -3.
Then within the applicable universe, all component stocks are ranked in a descending order based on their Z scores. The index has a fixed number of constituents, so the stock will be cut out when it hits beyond the lowest rank.
The MSCI Momentum Indexes are rebalanced on a semi-annual basis, a natural question here is why they are only rebalanced semi-annually; won’t the momentum dissipate given such a long time horizon? Certainly, it can be defended by arguing momentum needs an ample time to carry out, therefore, a less frequent rebalancing schedule may be more appropriate.
PowerShares DWA Momentum Portfolio, PDP, started in March 2007 with $1.4 billion AuM in June 2017, is tagged to Dorsey Wright Technical Leaders Index (DWTL), one of the index family consisting of more than 30 indexes with various market segments, regions and sectors. As to the methodology of DWTL, the very first step, setting the initial universe, is the NASDAQ US Benchmark index (NQUSB), then they are ranked by market capitalization to select the top 1000, minimum three-month average daily dollar trading volume of $1 million is required, the security must be classified as either a common stock or shares of beneficial interest of REIT. All securities in each Index universe are further ranked using a proprietary relative strength (momentum) measure. Each security’s score is based on intermediate and long-term price movements relative to a representative market benchmark. The Indexes are rebalanced and reconstituted at the end of each calendar quarter. A minimum of 30 securities is selected for each Index. The Index weights are determined by the scores of each security in their respective Index. Securities with higher scores receiving larger weights in the Index.
The calculation of the relative strength (momentum) score is pursuant to Dorsey, Wright & Associates, and LLC’s proprietary methodology. Public documents seem unavailable. However, the methodology is very likely similar to MSCI’s momentum index described above.
Next, we compare the performance and holdings between these two ETFs using Nasdaq Composite as a benchmark.
Both funds are highly correlated to the market, represented by Nasdaq Composite in the below two charts, however, they also underperformed.
MTUM came to market much later than PDP (2013 vs. 2007), but has since double the AuM of PDP. The reasons, very likely, are two: firstly, the high expense ratio 0.64% of PDP put a damp on investors comparing it to the 0.15% offered by BlackRock’s MTUM; secondly, MTUM’s a clear, transparent strategy compared to the black-box, proprietary mechanism claimed by PDP.
Take a deep look at their holdings, MTUM has 120, while PDP contains 100 constituents, the funds hold only 22 companies in common, which is only about a 20% overlapping rate. And the discrepancy between the two is a clear indication to investors that even both names indicate that they are “momentum” funds, their construction is vastly different.
Below is a performance comparison.
The third momentum ETF, SPDR Russell 1000 Momentum Focus, ONEO, is part of a suite of ETFs (along with ONEV and ONEY) that scores the members of the Russell 1000 index on three “core” factors (value, quality and small size) and one “focus” factor—in this case, momentum – Russell 1000 Momentum Focused Factor Index is used to create ONEO.
Launched in December 2015 by State Street, ONEO has attracted $472 million AuM since. The one-year return is high at about 20%.
The indexer is Russell, hence, the eligible starting universe is FTSE All-World Index, with a caveat that companies that manufacture or provide controversial weapon (CW), specific parts for anti-personnel mines, cluster munitions, chemical and biological weapons will be reviewed semi-annually and excluded. Note the base universe is different from PDP and MTUM, both of which are of U. S. market stocks.
Momentum factor here is defined as the cumulative total local return, calculated over the period that starts twelve months prior to the effective date, and ends the Monday following the third Friday of the previous month. A Z-score is created for normalization cross-sectional. Z-Scores that are greater (less) than three (minus three) are truncated to a value of three (minus three). Post-truncation, individual Z-Scores are renormalized by the re-application of the formula:
𝑍𝑍𝐽𝐽,𝑖𝑖 = (𝐹𝐹𝐽𝐽,𝑖𝑖 − 𝜇𝜇𝐽𝐽)/𝜎.