On May 1st, four new factor stock models went live on the models page. Three of the four new models focus on single factors (value, low volatility, and quality) while the fourth model is a multifactor stock model.
On May 1st, four new factor stock models went live on the models page. Three of the four new models focus on single factors (value, low volatility, and quality) while the fourth model is a multifactor stock model. Factor exposure has continued to grow in popularity over the years especially in the ETF space. However, many clients prefer individual stock positions in their portfolios or would like more concentrated exposure than holding an ETF can typically offer. The three individual factor models can be mixed and matched with existing allocations, providing flexibility in how they are used within different clients’ portfolios. Momentum and relative strength still drive the security selection process for each model, while broader factor rankings are used to determine which names to include in the models’ universes. To walk through the process, we’ll cover the steps the models took to get to their final holdings as of 5/1.
Step 1: Universe Selection
For the three individual factor models, the first step is defining a universe of stocks that will go into each model. All three factor models begin with a universe of the 1000 largest stocks in the US, primarily large and mid-cap stocks. From there, the initial universe of 1000 stocks are ranked based on factor’s specific criteria, ensuring that each model’s starting universe is “calibrated” properly. After each stock is assigned gets a ranking, the top quintile is selected to be the respective model’s final universe. For example, the universe for the NDW Quality Factor Model is made up of the top quintile of stocks based on a quality ranking including debt-to-equity, return-on-equity, and earnings variance. The universes for each factor model are updated at the end of each calendar year to reflect changes in the factor rankings of the 1000 largest stocks in the US. An important point for each of these factor models is that the first distillation of the universe is factor specific, meaning that each model is designed to keep exposure to a single factor. Relative strength is then used to pick out the strongest names from a universe already vetted for the highest quality names, for example.
Step 2: Stock Selection and Model Rules
The three individual factor models all use almost the same exact buy and sell rules and have the same evaluation frequency. The table below gives an overview of each factor model’s rule set. Trades are triggered when a current holding within the model falls sufficiently out of favor and is then replaced by the highest-ranking stock that is not currently a holding. There is one caveat for the NDW Value Model, which uses a positive trend overlay in addition to matrix rank to buy and sell holdings. If a current holding falls into a negative trend, then the NDW Value Model will sell that holding. The model will then select the highest-ranking stock in the matrix that trades in a positive trend to replace the old holding. This helps the model avoid “value traps” by using a trend overlay.
Step 3: Multifactor Approach
Step 3 is optional as investors can mix and match the individual factor models as they wish. However, the fourth model, a multifactor stock model, puts everything together for you. As we see with momentum itself as a factor, there are periods where factors rotate in and out of favor. By constantly holding exposure to different factors, the risk of holding a single factor is minimized, all while using relative strength to select the strongest stocks from each factor. This helps us maintain an emotionless process and removes some of the risk that we (or our clients) “choose” the wrong factor. The NDW Multifactor Model is made up of four factor submodels that each have five holdings for a maximum holding count of 20. Three of these submodels cover those factors covered earlier: quality, low volatility, and value. The submodels use the same ruleset as their ten-holding counterparts except for the holding count. The fourth submodel aims to encompass the growth factor and is a five-holding version of the NDW NDX Model that’s been available since 2019.
If you have any questions regarding the factor models, please reach out to joseph.tuzzolo@nasdaq.com.
