This In Practice piece gives a practitioner’s perspective on the article “Optimal Tilts: Combining Persistent Characteristic Portfolios,” by Malcolm Baker, Ryan Taliaferro, and Terence Burnham, published in the Fourth Quarter 2017 issue of the Financial Analysts Journal.
What’s the Investment Issue?
Investors employ factor-based analysis to generate stock returns by combining their chosen factors in a single portfolio or by assembling multiple single-factor portfolios. Such investors attempt to create the optimal mix of factors—defined by the authors as a stock’s deviation from the return predicted by the capital asset pricing model—to generate the best risk-adjusted returns.
Finding the best way to compare factors is critical to investor decision making. The authors identify two types of factors: one type that is long-lasting and persistent and another that is less persistent and entails high-frequency trading and higher costs.
Comparing the gross returns of the two types, the authors argue, is akin to comparing apples with pears. Investors need to consider how implementable each factor is by considering the costs—principally, trading and liquidity costs—when comparing and weighting factors.How Do the Authors Tackle the Issue?
In order to compare apples with apples, the authors divide the seven most-recognised factors—which have been identified in the academic literature as having the highest risk-adjusted returns—into “tilts” and “trades.” They define tilts as factor-based strategies that are easily implementable because they have relatively low annual turnover or allocate to large-cap stocks, which are very liquid and thus readily traded. For example, the small-size factor requires minimal trading, because small stocks this year are likely to remain small next year.
The tilts selected for this study are low beta (better known as “contrarian investing”), value (using fundamental accounting measures), small size (investment in small-cap stocks), and high profits (growth investing).
Trades, on the other hand, require much more frequent rebalancing. The trades in this study are defined as low growth, momentum, and high-frequency reversal. The authors categorise the low-growth factor as a trade because its annual turnover is high, at around 70%, and because it relies on small stocks. As a result, trading costs are high and liquidity is relatively low.
Finally, the authors add fixed-income tilts designed to capture duration and credit risk.
The authors create long–short portfolios using each factor, based on data 1968–2014, and compare them from the points of view of both risk and return budgets.What Are the Findings?
Based on the costs of implementation and the difficulties of computing these costs, the authors focus solely on the optimal mix of the four “tilts.”
Risk–return is highest when the tilts are more or less equally weighted, the authors find—that is, allocating a 26% share to small size, 24% to a low-beta tilt, 23% to high profits, and 20% to value. The remaining portfolio is allocated to the two bond market factors: duration and credit. This optimal blend produces an average return per month of 3.3%, with an annual Sharpe ratio of 1.0.
The large weighting to the low-beta factor contrasts with findings of previous studies, which claim that low-risk strategies are subsumed by the value and high-profit factors. In fact, through the apples-to-apples comparison in this study, it is shown that the low-beta tilt is not subsumed by other tilts. The authors find that much of the risk and half of the return of low beta are unexplained, putting it on a par with the value and profitability factors as a driver of stock returns.What Are the Implications for Investors and Investment Professionals?
The authors show that by excluding “trade” factors, it is possible to create an optimal mix of tilts that delivers a high Sharpe ratio. That mix is, roughly speaking, an equal allocation of the funds in the equity portfolio to each of the four tilts.
According to the authors, the common practice of mixing and matching tilts and trades is illogical. For example, creating a portfolio of small-cap value stocks produces statistically positive alpha. In contrast, a value strategy based on large stocks would produce a portfolio with statistically negative alpha. Similarly, long–short portfolios of small stocks would produce higher alpha than long-only small-stock portfolios.
When implementing this strategy, allocations to the tilt portfolios, to the index, and to cash depend on the desired level of risk. By focusing on tilts and excluding trades, the authors have designed the strategy to be most relevant to large-scale equity investors, such as pension funds, endowments, and sovereign wealth funds.