Forecasting stock market volatility: Do realized skewness and kurtosis help?

Time-varying risk aversion and realized gold volatility

“…The MRV w component is significant at the short forecasting horizon, while the MRV m component is significant mainly at the medium and long forecasting horizons. The models supplemented by the realized moments yield evidence partially in line withMei et. al (2017) in that only realized skewness is found to have significant in-sample predictive power, across all forecast horizons.Interestingly, we see that including risk aversion in the model adds predictive value in the case of the short forecasting horizon, as indicated by the significant estimated coefficients.…”

“…In our empirical analysis, we focus on the realized volatility of gold returns that we compute from intraday data. The use of intraday data allows us to control for higher moments including the realized skewness and kurtosis that have been shown to have predictive power in forecasting models in a number of different contexts including gold (Mei et al 2017, Bonato et al 2018, Gkillas et al 2018. We employ the heterogeneous autoregressive RV (HAR-RV) model developed by Corsi (2009) to model and forecast the realized volatility of gold returns as this widelystudied model accounts for several stylized facts such as fat tails and the long-memory property of financial-market volatility, despite the simplicity offered by the model.…”

“…Variants of the HAR-RV model have been widely studied in recent research (see, for example, Haugom et al 2014, Lyócsa andMolnár 2016) as it accounts for several stylized facts such as fat tails and the long-memory property of financial-market volatility, despite the simplicity the model offers. In our empirical analysis, we extend the core HAR-RV model to include measures of realized higher-moments including realized skewness and kurtosis that have been found to significantly improve model performance in the case of stock market indexes (Mei et al 2017). This is a remarkable finding given that, in the case of gold and silver, evidence suggests that it is difficult to beat the HAR-RV model in terms of forecasting performance by using versions of the univariate HAR-RV model extended to include semi-variances and jumps (Lyócsa and Molnár 2016).…”

The predictive power of oil price shocks on realized volatility of oil: A note

“…Not surprisingly, a large literature exists, which we summarize in the literature review section, on the predictability (both in- and out-of-sample) of daily oil-price volatility. Given that intraday data contains rich information and can lead to more accurate forecasts of daily volatility compared to models of conditional volatility based on daily data ( McAleer and Medeiros, 2008 ), an increasing number of studies (for detailed reviews, see Mei et al, 2017 and Qiu et al, 2019 ) have used variations of the Heterogeneous Autoregressive (HAR) model of Corsi (2009) to forecast the realized volatility (RV) of various asset and commodity markets returns by utilizing the realized volatility estimates obtained from intraday data per Andersen and Bollerslev (1998) . Against this backdrop, we aim to contribute to the existing research on oil market volatility based on high-frequency data by forecasting RV of oil returns, computed from 5-min-interval intraday data, using an extended version of the HAR-RV model.…”

Forecasting the spare part demands for mobile phones

“…Hu [7] presented a two-dimensional preventive policy where replacements of objects were determined based on both calendar and usage times. Besides the common short-term prediction, Mei [8] proposed a method for long-term volatility predictions. The research of this paper is conducted based on the business practice of a supply chain management company in Shanghai, China.…”