Improving Portfolio Selection Using Option-Implied Volatility and Skewness

Abstract

Our objective in this paper is to examine whether one can use option-implied information to improve the selection of mean-variance portfolios with a large number of stocks, and to document which aspects of option-implied information are most useful to improve their out-of-sample performance. Portfolio performance is measured in terms of volatility, Sharpe ratio, and turnover. Our empirical evidence shows that using option-implied volatility helps to reduce portfolio volatility. Using option-implied correlation does not improve any of the metrics. Using option-implied volatility, risk premium, and skewness to adjust expected returns leads to a substantial improvement in the Sharpe ratio, even after prohibiting short sales and accounting for transaction costs.

HUANG XIANGYA, DDS,PHD, Senior Lecturer, Department of Conservative Dentistry & Endodontics, Guanghua School of Stomatology, Hospital of Stomatology, Sun Yat‐Sen University, Guangzhou, China

“…The out-of-sample investment period is from January 2012 to November 2020. Short-selling is not allowed and we only consider the allocation within risky assets (DeMiguel et al, 2013). For robustness, we rebalance the portfolio every month, quarter, and half a year, and a 100-week rolling window is adopted.…”

“…Bliss and Panigirtzoglou (2004) extract the implied risk-neutral probability (Breeden and Litzenberger, 1978) of the underlying asset from option prices and convert it into physical probability via the pricing kernel. There are also findings that option-implied information helps to improve the out-of-sample performance of portfolios (DeMiguel et al, 2013;Kempf et al, 2015;Kostakis et al, 2011). However, it is notable that estimating the physical return distribution of underlying asset from option prices is only applicable for a single risky asset, while the recovery of joint physical return distribution of multiple underlying assets remains an open problem (Carr and Wu, 2020).…”

Computing the Nondominated Surface in Tri-Criterion Portfolio Selection

“…Also, the cases of 3 = 0 highlight, similar to the reward-to-variability case, the poor out-ofsample performance of classical standard Markowitz optimization. Yet, while we stick to a parsimonious mean and covariance estimation, we point out that our tricriterial model can also incorporate improved estimators as those recently suggested by DeMiguel et al (2012) and Disatnik and Katz (2012).…”

Asset Allocation via Machine Learning

“…Because the first order moments of asset returns are notoriously hard to estimate, simple return-agnostic strategies were created to account for this phenomenon. For example, the minimum variance portfolio and equal weight (1-over-N) portfolio are studied in (Demiguel, Plyakha, Uppal & Vilkov, 2013) and (Allen, Lizieri & Satchell, 2012). In addition to the first and second moments, there have been strategies based on higher order moments, namely, skewness and kurtosis, appearing in the work of (Harvey, Liechty, Liechty & Mueller, 2010) and many others.…”