Over the past several years, we have witnessed the rapid development and launch of many “smart-beta factor strategies.” We’ve looked under the hood at many of them and found the methodologies to be very complex and the initial goal hard to decipher. Some of them appear more like factor-soup than an actual common sense strategy. So to answer, “How did we build it?” We built the index with the client in mind, and replicated a strategy we were already using in client accounts.
We took the factors which we have already discussed: minimum volatility, value, and momentum, and simply looked at how they performed compared to each other. The end methodology will either own two factors at one time, equally weighted 50% / 50%; or all three with a weighting of 40% / 40% / 20%. Our methodology will look to eliminate or underweight the best performing factor over a specified trailing time period.
Why eliminate the best performing factor? This hits at the heart of our philosophy: mean-reversion. Factors are cyclical, and factor performance can become stretched. I’m sure we can all point to times when the momentum factor became too far stretched to the upside after the investment herd bought in. What happened? It mean-reverted after spending time at the top and came crashing down. The same can be said of value as well; its performance can become stretched. Our goal is to own the factors that are overly stretched on the downside and avoid the ones that are overly stretched on the upside. This provides the opportunity for the underperforming factors to appreciate and hopefully avoid the factor that is about to fall. While not foolproof, it is a common sense strategy that attempts to follow a basic investment principal, buy low and sell high. That is exactly what tried to build.
While thinking through the construction of the strategy, we also knew something we did not want to do. We did not want to re-optimize the portfolio to look like the market-cap weighted index. This means we did not overlay a sector-neutral strategy, place a limit on individual security position size, or other complex constraints. The goal was to let the factor selection and the mean-reversion of the factors drive the return. We feel the simple solution can be an optimal solution.
Yes, we could have applied some of these constraints and tried to over-engineer the product, as many ‘smart-beta’ products are built. This over-engineering could have lowered the tracking error, however it would have also lowered the alpha. If the goal is to outperform something over time, we know it becomes more difficult to achieve with additional constraints. The strategy owns at a minimum two factors at any point in time. By always owning at least two, the strategy has avoided sector concentration that may occur from only owning one single factor.