We are excited to introduce a new methodology that enables us to closely examine what has been driving any given stock's returns.
When analyzing an individual stock in your portfolio (or that might have a place in your portfolio), it is often helpful to figure out what has been causing its recent price action. Our new tool helps answer these very questions. Let’s use Amazon (AMZN) as an example. It’s been on a huge bull run, more than enough to fund Bezos’ missions to space for many years to come. Is tech just hot? Is it momentum? Or do people just think they're going to take over the world?
Our new chart allows us break down the returns into alpha vs. various factors. While Technology and Momentum have contributed slightly to Amazon’s recent run up, the underperformance of other factors like Size and Growth have had more of an impact. In total, these charts make it clear that a majority of Amazon’s price action over the past 12 months has been alpha-driven.
Our Methodology for Measuring Alpha
How are we doing this? Up until recently, we have been using explicit factor models. These are factor models in which we take the fundamental or technical data associated with each stock and rank them. As an example, we know each stock's ROE, and we can rank each stock relative to the rest of the universe. We have monitored how each factor is performing by comparing the average return of the 20% of stocks with the highest factor ranking against the 20% of stocks with the lowest factor ranking. The advantages of this approach have been its directness and simplicity. We can look at a company's raw values or rankings today and see how they evolved over time. Any attribution analysis done on portfolios with explicit factors has to be done via time series regression. Worryingly, this approach does not take into account stocks in the middle rankings at all. A stock at the 79th percentile or 41st percentile has no bearing on the factor return. Furthermore, the attribution analysis can be easily skewed by outliers or stocks that are moving for reasons unrelated to their factor weightings and lead to counterintuitive results.
A z-score is a measure of how many standard deviations away from the mean a given number is. To give an example, if the average 12M Return was 10% and the standard deviation of those returns was 10%, then the 12M Return z-score for a stock with a return of 0% would be -1. A stock with a return of 10% would have a 12M z-score of 0. A stock with a 20% return has a 12M z-score of 2. We winsorize the z-scores at positive and negative to limit the impact of outliers, so returns of 30% and above would all be given 12M z-scores of 3. Z-scores for industry and geographic exposures are calculated by finding the beta of the underlying to the time series index representing the relevant sector or geography and then z-scoring the beta relative to the rest of the universe.
These z-scores are placed into a matrix, and through some matrix math and winsorization of outlier returns, (please see our white paper for details), we generate factor betas that on average describe 25% of the daily variance in the top 2000 market cap stocks in the US. If it seems like 25% is low given that each stock has over 60* associated factors, keep in mind that these are longer term factors explaining daily returns and that random betas would only explain ~1% of the daily variance.
To demonstrate how we generate a daily factor return for a factor in this model, let's take a look at the 1 Month Beta example below:
The slope of the red line is a stock’s factor beta calculated in the analysis described in the whitepaper. On the horizontal axis is the z-score for each security. The green dots on the right are all the stocks in the top quintile. The magenta dots represent the z-score and returns of stocks in the bottom quintile. Rather than use the actual return of each stock, the red line is used. This is another important method for avoiding the impact of outliers. Many stocks have returns for reasons unassociated with the relevant factor, and rather than assuming that returns unrelated to the specific factor will average out we utilize this method to get a purer factor return.
To calculate the factor return we first calculate average z-score for the green dots. The spot on the red line associated with that z-score is the upper quintile return. A similar calculation is performed for the magenta dots to calculate the lower quintile performance. The factor return is then simply the upper quintile return minus the lower quintile return.
Another benefit relative to the fundamental approach: Rather than only measuring the impact of stocks in the upper and lower quintiles, the stocks in the middle quintile influence the slope of the line and therefore in the calculation of the factor return.
In addition to giving us a portfolio analysis, this approach allows us to break down each security into factor driven returns (beta) and idiosyncratic returns (alpha).
Let’s review another stock that has been in the news over the past few years, though with mostly negative press. It’s a company that many people have labeled a pyramid scheme: Herbalife.
This is a stock which Bill Ackman has been furiously shorting for years, often directly opposed by other financial heavyweights like Carl Icahn. This stock has missed the bull market over the past 12 months. And while its exposure to Price/Book and Short Interest factors have played against it, it is mostly negative alpha that has been holding this stock back.
With OMEGA POINT you can now check the factor contribution to any stock in your portfolio, or any stock that you are thinking of adding to your portfolio. If you have any questions, please feel free to reach out to us.
The top 2000 US stocks by market capitalization is useful, but we understand that some people benchmark themselves to more specific indexes. We will soon be rolling out the ability to choose a different benchmark and universe of stocks against which to benchmark your portfolio.
The factor return breakdown currently includes all betas. We are going to turn the factor chart into two charts. One chart will be the impact of the benchmark beta, and the other will be the impact of all of the other factors. This will allow for a timeline comparison between alpha, benchmark beta and factor beta.
Click Here to see how OMEGA POINT can help you better manage your portfolio.
The Omega Point Team