Other Factors

In general, the more reliable models start with a theory and then are empirically tested (or more formally, its “null hypothesis” is rejected or not). Otherwise, something akin to the “Super Bowl Model” can result. With 91% accuracy, this model predicted the direction of the stock market based on the winner of the Super Bowl. If the winner was an old NFL vs AFL team–at least for the first 22 games. After those impressive initial results, the “models” predictive ability has been zero. Yet, we should not be blind to the data. Our forefathers may have lacked a theory regarding lightening before they observed it was still best to take cover in a thunderstorm.


Momentum is an example of a variable shown to be a source of additional incremental return and is not just robust throughout history (even for a paper testing for momentum in the St. Petersburg Stock Exchange from January 1865 to July 1914!), but also across markets and subsequent “out of sample” testing (although less so recently).  In other words, despite a rather flimsy theory centered around poor portfolio manager behavior, the data tests are undeniable.  Plus, momentum seems to be perform best when value strategies lag, thus at the very least dampening portfolio volatility. But how rigorous should a taxable investor adhere to the rules of building a momentum tilted portfolio?

Our answer is not so much as to generate taxable gains or pay a sizeable management fee, but certainly enough to be willing to take a loss on a stock (or a fund) that is in the bottom quartile of performance over the semi-recent past and to take momentum into account when building an otherwise value-centric portfolio. To limit taxable gains, we often stick to tax-efficient Exchange Traded Funds (ETFs) focussed on momentum.


Utilizing pure mathematics, one can relatively easily construct a portfolio that provides an optimal expected return relative to it’s volatility, “mean-variance efficient” in academic parlance. But those portfolios are highly susceptible to your inputs, and although historical risk and return measurements are plentiful, expected future returns and associated risk measures are and always will be largely guess work. Even then, correlations are forever changing. Instead of giving up on minimizing volatility, though, we strive to make sure that in the face of uncertainty, our attempts are both sound and adequate.

We skin the low volatility conundrum in multiple ways. We bias low volatility stocks with lower correlations to the market portfolio. More, we overlay a layer of common sense into the equation and limit the weights of any individual stock and the weights of any one industry — likewise with “risk” factors like Value, Size and Momentum. In short, we do not allow the quest of low volatility in an imperfect world to replace our pursuit for maximum tax efficiency.


Although we don’t directly tilt our stock portfolios to low liquidity stocks, we are cognitive of the research and believe we achieve the some of the same bias by trading in closed end funds (see section of Closed-End Funds) and illiquid bonds (especially pre-crisis private label residential mortgage backed bonds and preferred shares. In short, we look to get paid to provide liquidity, but realize the cost and so temper our exposure to these asset classes.


The important point for all of our biases, though, bares repeating: Our goal is not to find the “mean-variance” optimal basket of stocks – that is a fool’s errand. Even if markets were completely efficient, optimization is impossible when the factors we use are themselves proxies for some unknown risk. Rather, while building a portfolio of either stocks or ETFs to gain exposure to many of these hoped for return enhancing attributes, we also thoughtfully try and maximize the real benefits of tax efficiency. Holding a large but more equally weighted basket of stocks is consistent with that approach (see section under Tax Loss Harvesting).