PREDICTING EXTREME RETURNS AND PORTFOLIO MANAGEMENT IMPLICATIONS

• Author: Kevin Krieger,Greg Stevenson,Andy Fodor,Nathan Mauck
• Published date: 01 December 2013
• DOI: http://doi.org/10.1111/jfir.12020
• Date: 01 December 2013

PREDICTING EXTREME RETURNS AND PORTFOLIO

MANAGEMENT IMPLICATIONS

Andy Fodor

Ohio University

Kevin Krieger

University of West Florida

Nathan Mauck

University of Missouri‐Kansas City

Greg Stevenson

Janus Capital

Abstract

We consider which readily observable characteristics of individual stocks may be used to

forecast subsequent extreme price movements. We believe we are the first to explicitly

consider the predictive influence of option implied volatility in such a framework, which

we find to be an important indicator. However, after controlling for implied volatility

levels, other factors, particularly firm age and size, continue to have additional predictive

power of extreme returns. Furthermore, excluding predicted extreme return stocks leads

to a portfolio that has lower risk (standard deviation of returns and lower beta) without

sacrificing performance.

JEL Classification: G10, G14, G17

I. Introduction

Traditional studies often seek out firm characteristics or tendencies that may be linked to

higher expected returns. Although such studies are numerous, and many other works

examine the forecasting of future variance levels, markedly less research investigates the

question of whether firms experiencing returns of high magnitude (positive or negative)

may be identifiable ex ante. We consider whether firms exhibiting extreme returns have

common characteristics. If so, can extreme return firms of future periods be predicted?

Furthermore, can portfolio risk be reduced by eliminating firms likely to have extreme

future returns without adversely affecting portfolio performance?

Although finance theory suggests a positive relation between risk and return

(Sharpe 1964; Merton 1973, 1980), empirically, the results are mixed. Black (1972) and

The authors would like to thank David Peterson, Bong‐Soo Lee, Eric Higgins, Steve Pruitt, Johnny Chan (the

referee), and Mark Griffiths (the associate editor) as well as seminar participants at the University of Tulsa for their

influential input.

The Journal of Financial Research Vol. 36, No. 4 Pages 471–492 Winter 2013

471

© 2013 The Southern Finance Association and the Southwestern Finance Association

Black, Jensen, and Scholes (1972) provide empirical tests of the capital asset pricing model

(CAPM) and find that the relation between beta and firm returns differs from theory. Frazzini

and Pedersen (2010) find evidence that indicates this result holds in several markets in

addition to stocks, including Treasuries, credit instruments, and futures.

Tests regarding the pricing of idiosyncratic risk are similarly mixed. Ang et al.

(2006, 2009) examine the pricing of idiosyncratic risk in the cross‐section. They find that

portfolios consisting of firms with the highest idiosyncratic risk are characterized by

negative alphas and significantly underperform a portfolio of stocks with the lowest

idiosyncratic risk. Blitz and Van Vilet (2007) study the possibility of actively managing

portfolios to benefit from the inclusionof low‐volatility stocks. Using U.S., European, and

Japanese stocks, they document a 12% spread between low‐and high‐volatility portfolios

and argue that, in practice, portfolio managers should add low‐volatility stocks as a

separate asset class.However, Jiang and Lee (2006), Fu (2009), and Chua, Goh, and Zhang

(2010) find a positive relation between idiosyncratic risk and returns. Huang et al. (2010)

find that once return reversals are controlledfor, the negative relation between returns and

idiosyncratic returns is no longer present. Similarly, Peterson and Smedema (2011) find

that the negative relation between idiosyncratic risk and returns is sensitive to seasonal

factors. Overall, although the empirical relation between risk and return remains unclear,

there is considerable evidence that low‐volatility stocks carry potential advantages.

We invoke a new approach in this article. Rather than attempting to pick

“winners,”we seek to identify extreme return firms ex ante and remove them from a

value‐weighted market position similar to that held by many passive investors. The ability

to identify firms more apt to experience extreme future returns might provide lower risk

portfolios, a possibility we investigate.

Stock volatility level serves as perhaps the most intuitive firm characteristic that

may aid in the predictability of extreme stock returns. Chan, Jha, and Kalimipalli (2009)

demonstrate thatforecasting future volatility may result in economicbenefits when trading.

Poon and Granger (2003) outline numerous publications on the methods by which

volatility can be forecasted. They demonstrate that the majority of research on predicting

volatility favors a stock’s implied volatility from option prices, unsurprising given that

implied volatility reflects active perceptions of future levels of volatility. Furthermore,

Ofek, Richardson, and Whitelaw (2004), Cremers and Weinbaum (2010), and Xing,

Zhang, and Zhao (2010) find that option implied volatility predicts future stock returns.

Information available in companies’financial statements may also indicate extreme

future equity returns. Accounting variables, fundamental analysis, and trading characteristics

help predict returns in past research on firm size and book‐to‐market ratios (Fama and

French, 1992, 1993), trading volume (Campbell, Grossman, and Wang 1993), price

momentum (Jegadeesh and Titman 1993), dividend yields (Hodrick 1992), accruals

(Sloan 1996), and earnings predictions (Lev and Thiagarajan 1993). We consider the impact

of such measures on the likelihood that firms will exhibit extreme future price movements.

To identify potential extreme price movements, Reinganum (1988) and Beneish,

Lee, and Tarpley (2001) use unique, directionally oriented approaches that consider

market‐based signals. Among this group of 12 market‐based signals, Beneish, Lee, and

Tarpley find that relatively younger firms with lower market capitalization, lower stock

prices, higher trading volume, lower sales‐to‐price ratios, lower analyst coverage, and

472 The Journal of Financial Research