Is There a Better Way to Think About Factor Investing?

Topics: ,

Published on December 17, 2019

| 7 min read

Vassilios Papathanakos, PhD, Distinguished Researcher

Share:

The popularity of factor investing appears to be at a crossroad. While investors’ adoption of factors has never been broader, their concerns about the effectiveness of factor investing are increasing. In our opinion, asking the right questions about factor investing has never been more justified.

Factor investing may be outgrowing its usefulness, not just by dramatically diluting the alpha potential it promised investors, but also by materially increasing the uncompensated risks to which they’re exposed. Even investors who do not directly target factors are now subject to these systematic risks.

Questions About Drugs: What condition does this drug treat? Which of the drug’s possibly side effects should I be worried about? Should I take this drug if I’m taking other drugs? Questions About Factors? What is the factor’s alpha potential? Does the factor itsself create risks? How does it interact with other factors?

Overcrowding Risk

Ultimately, the problem is due to overcrowding: the more widely known and employed a factor, the more the factor affects your investments and your trading activity. The marketplace amplifies overcrowding by making factors convenient through simply designed smart beta products, masquerading as “passive” indices. The combination of wide accessibility and simple engineering (typically required to reduce costs), results in a greater potential for overcrowding well-known trades.

This over-the-counter approach to factor investing is particularly risky for factors in terms of performance and valuations. For example, when momentum stocks underperform the market, some momentum portfolios are likely to sell out of them (and buy the new recent winners) further depressing their prices and likely triggering a chain reaction across momentum portfolios. Similarly, if low-volatility stocks attract investors in a risk-off environment, they can become overcrowded and exhibit higher volatility, which adds insult to injury as they offer less downside protection at an increased cost.

Quantitative investors seeking to enhance their exposure to a factor through complex calibrations are still subject to most of these risks. And yet they add to the problem because their trades contribute to the uncertainty about the evolution of the factor’s behavior.

“The rest of the investor community has had to take notice of overcrowding risks, and carefully step aside when hot trades roil the markets.”

This phenomenon threatens more than those who are actively trying to exploit factors. The rest of the investor community must also take notice of overcrowding risks, and carefully step aside when hot trades roil the markets. Overcrowding can pummel factor-agnostic investors: ignorance is not bliss.

Consequently, we believe the main reason for an understanding of factors today should be less about their alpha potential and more about the risks they can exert onto your portfolio, even if you don’t specifically target factor exposures.

Focus on Factor Interactions

A straightforward way to address these risks is to consider them, not in isolation, but in combination with each other in the portfolio. We have all long understood that a single stock cannot be more or less diversifying on its own, but only in relation to the rest of the portfolio.

The same calculus holds for outperformance. As Intech’s founder discovered almost 40 years ago, we can divide portfolio growth into two parts: stock effects and a portfolio effect (Figure 1). The former reflects stocks’ individual growth rates, which tend to be transient and difficult to forecast, while the latter is a simple measure of portfolio diversification. By combining complementary stocks, and thereby improving portfolio diversification, (half the difference between the average stock variance and the portfolio variance), one can increase a portfolio’s potential to outperform.

 

Fig_1_

 

This observation holds the key to managing factor risks as well. For every factor exposure, we need to understand how much of the performance is the result of individual stock effects and how much is portfolio- level collaboration. This approach allows investors to identify the positive contributors and immunize or protect against uncompensated factor risks.

Stock and Portfolio Effects in Action

Figure 2 illustrates the risk management approach for three common factors: momentum, size, and volatility. It plots the monthly return of stocks in the S&P 500 ranked from highest to lowest for these factors. In the case of momentum and volatility, there’s a definite stock effect: avoiding the lowest (most negative) momentum stocks and highest volatility stocks should meaningfully boost the portfolio return.

 

Fig_2_

 

In the case of size, however, we see that there’s no material dependence of the long-term return on a stock’s capitalization. This observation holds for size quite generally, as long as one studies periods that are long enough to include a full market cycle.

How does our picture of the size factor reconcile with the known small-cap effect? Simple: the alpha is due to portfolio-level effects. The performance gain from increased diversification (again, an amount equal to half the difference between the average stock variance and the portfolio variance) is equal to the long-term outperformance observed for portfolios exposed to the size factor (e.g., an equal-weighted portfolio).

By repeating this analysis systematically using real-time market data, a powerful framework emerges for managing factor risk. Decomposing factor performance to stock- and portfolio-level effects allows us to understand positive exposures to the diminishing alpha potential of factors while also dynamically managing for their uncompensated risks. Together, these serve every investor’s long-standing objective: improved risk-adjusted performance.

How Did We Get Here?

We believe factors have moved on from being a compelling partial solution to the problem of discovering alpha, to actually causing a problem of their own – engendering systematic risks. Today, a better use for factors is to help mitigate those risks.

But how did we get here?

Download our paper about factor investing, “Are You Asking The Right Questions About Factor Investing?” to learn more.

Are You Asking the Right Questions About Factor Investing? Understand factor exposure beyond advertised claims of compensated risks. Download Now

 

1. SFernholz, R., & Shay, B. (1982). Stochastic Portfolio Theory and Stock Market Equilibrium. The Journal of Finance, 37(2), 615–624. doi: 10.1111/j.1540-6261.1982. tb03584.x.

2. FACTOR DEFINITIONS: Size is the logarithm of the reconstructed index weight; Momentum is the relative logarithmic total return of the stock with respect to the reconstructed index, over a period starting 252 days and ending 22 days prior to each date (inclusive in both ends); Volatility is the standard deviation of the absolute logarithmic total return of the stock over the previous 252 days. DATA AND METHODOLOGY: The figures in the text are based on an analysis of the stocks in the S&P 500 index over the period January 1966 through September 2019. Similar results hold for other indices, but this dataset is preferable because it offers the longest historical record of daily returns available to us. For each factor, the score is computed for all the stocks in a reconstructed index on a daily basis using the previous historical data. The reconstructed index consists of the stocks in the S&P 500 index at the beginning of each month, with their weights propagated forward on a daily basis using the corresponding total returns (this assumes an appropriate daily reinvestment of the dividends on the ex-date). For each factor, the stocks are ranked by their scores from largest to smallest, and their ranks and absolute logarithmic total returns for the following 21 days are collected over the entire period minus the last 21 days. Finally, the aggregate data are separated into 20 quantile groups by rank; the ranks and the returns in each group are averaged, and the average returns are further normalized to have zero mean across the 20 groups.

All historical data is based on back-testing, which is not actual performance, but is hypothetical. Back-tested data is prepared with the benefit of hindsight and is not a guarantee of future results. In addition, factor investing does not provide any assurance of improved performance or risk reduction. Exposure to such factors may detract from performance in some market environments, perhaps for extended periods. In such circumstances, a strategy may maintain exposure to the targeted investment factors and not adjust to target different factors, which could result in losses. Factor-based investing has additional unique complexities that investors should consider when evaluating expected returns.