Factor investing has many benefits, but the strategy is not risk free. In developing our indices, we sought to recognize the risks up front and incorporate an implementation route to minimize them. All too often within the factor investing community, risks tend to be understated while diversification benefits tend to be overstated (“Ignored Risks of Factor Investing” by Kalesnik and Linnainmaa, 2018).  For example, many investors focus on the average value-add and underestimate how a factor’s tracking error can translate into underperformance over time periods long enough to test the patience and determination of most investors. In addition, the performance of factors tends to deviate from a normal distribution, thus leading to a higher chance of outlier outcomes than may otherwise be expected.

Three additional, but often-overlooked, risks include valuations, costs, and data mining.

Valuations. With the growing popularity of factor-based smart beta strategies, several of these strategies have become expensive relative to their historical valuations. This is an important consideration in selecting factor-based strategies for investment. Our sister company, Research Affiliates, has shown that current valuations in a factor context are predictive of future returns (“How Can ‘Smart Beta’ Go Horribly Wrong?” by Rob Arnott, Noah Beck, Vitali Kalesnik, and John West, 2016). When making decisions about factor-based smart beta strategies, investors should consider current valuations relative to history in order to manage return expectations.

Costs. Whereas most investors solely focus on explicit costs such as licensing fees and total expense ratios, we find that all transaction costs—including implicit costs such as market impact—are crucial considerations in strategy construction as these can severely undermine investment outcomes. Many factor-based strategies are derived from academic studies alone. As a result, many factor and smart beta strategies are not designed with implementation in mind and therefore are expensive to implement. A careful assessment of the transaction costs of specific factor strategies is important when making strategy selections for a portfolio.

Data Mining. The field of factor investing exposes investors to an outsized risk of being sold strategies on the basis of severely data-mined backtests, leading to an upward bias in performance expectations and future disappointment. Indeed, in the worst of cases, historically simulated results may suggest the existence of value-add in a strategy or factor that possesses no efficacy in reality. In addition to approaching backtests with caution from the perspective of transaction costs arising in real-world implementation, investors are encouraged to ask whether the investment strategy is predicated on sound investing principles and theory, irrespective of backtest results. For a more detailed analysis of how to avoid data mining, please refer to “A Backtesting Protocol in the Era of Machine Learning” by Arnott, Harvey and Markowitz (November 2018).   

Clearly factor investing is not a free lunch, but that does not mean investors should avoid it altogether. Factor investing combined with smart beta designs, such as the Fundamental Index™ weighting methodology, has the potential to improve a portfolio’s long-term risk-adjusted returns, especially for patient investors who understand the risks. The ability to fully capture potential returns is improved by investing in strategies that are transparent, use robust factors, have reasonable management fees, and offer superior implementation characteristics.