دانلود مقاله پیش بینی نوسانات بازار سهام با تعداد زیادی پیش بینی کننده
خلاصه
1. مقدمه
2 روش شناسی
3 داده
4 نتایج تجربی
5 تجزیه و تحلیل پسوند
6. نتیجه گیری
منابع
Abstract
1 Introduction
2 Methodology
3 Data
4 Empirical results
5 Extension analysis
6 Conclusions
References
چکیده
این مقاله اثربخشی پیش‌بینی‌کننده‌ها، از جمله نه شاخص عدم قطعیت سیاست اقتصادی، چهار شاخص احساس بازار و دو شاخص استرس مالی را در پیش‌بینی نوسانات تحقق یافته شاخص S&P 500 بررسی می‌کند. ما از چارچوب MIDAS-RV استفاده می کنیم و مدل MIDAS-LASSO و پسوند تغییر رژیم آن (یعنی MS-MIDAS-LASSO) را می سازیم. اول، در میان همه پیش‌بینی‌کننده‌های در نظر گرفته شده، شاخص‌های عدم قطعیت سیاست اقتصادی (به ویژه شاخص نوسانات بازار سهام) و شاخص نوسان CBOE قابل توجه‌ترین پیش‌بینی‌کننده‌ها هستند. اگرچه شاخص نوسانات CBOE بهترین توانایی پیش بینی برای نوسانات بازار سهام را دارد، توانایی پیش بینی آن در طول همه گیری COVID-19 ضعیف شده است و شاخص نوسانات بازار سهام در این دوره بهترین است. دوم، مدل MS-MIDAS-LASSO بهترین عملکرد پیش بینی را در مقایسه با سایر مدل های رقیب دارد. عملکرد پیش‌بینی برتر این مدل حتی زمانی که بین دوره‌های با نوسان بالا و کم نوسان تمایز قائل می‌شود، قوی است. در نهایت، دقت پیش‌بینی مدل MS-MIDAS-LASSO حتی از استراتژی سنتی LASSO و گسترش تغییر رژیم آن نیز بهتر عمل می‌کند. علاوه بر این، عملکرد پیش‌بینی برتر این مدل با شیوع اپیدمی COVID-19 تغییر نکرده است.
توجه! این متن ترجمه ماشینی بوده و توسط مترجمین ای ترجمه، ترجمه نشده است.
Abstract
This paper explores the effectiveness of predictors, including nine economic policy uncertainty indicators, four market sentiment indicators and two financial stress indices, in predicting the realized volatility of the S&P 500 index. We employ the MIDAS-RV framework and construct the MIDAS-LASSO model and its regime switching extension (namely, MS-MIDAS-LASSO). First, among all considered predictors, the economic policy uncertainty indices (especially the equity market volatility index) and the CBOE volatility index are the most noteworthy predictors. Although the CBOE volatility index has the best predictive ability for stock market volatility, its predictive ability has weakened during the COVID-19 epidemic, and the equity market volatility index is best during this period. Second, the MS-MIDAS-LASSO model has the best predictive performance compared to other competing models. The superior forecasting performance of this model is robust, even when distinguishing between high- and low-volatility periods. Finally, the prediction accuracy of the MS-MIDAS-LASSO model even outperforms the traditional LASSO strategy and its regime switching extension. Furthermore, the superior predictive performance of this model has not changed with the outbreak of the COVID-19 epidemic.
Introduction
Financial market volatility is not only a key factor in assessing the risk of financial assets (see, e.g., Wang et al., 2015; Bee et al., 2016; Clements & Liao, 2017; Li & Wei, 2018; Ji et al., 2021) but also an important parameter in the pricing of financial derivatives and asset allocation (Graham & Harvey, 1996; Zhu & Ling, 2015), as well as an important factor in portfolio management (Ayub et al., 2015; Cederburg et al., 2020). Therefore, accurately predicting the volatility of the stock market is particularly important for investors when deciding on the sizes and timing of their investments.
Despite the many benefits of accurately predicting stock market volatility, improving predictive accuracy has proven challenging. First, the diversity of potential predictors poses a great challenge to the selection of predictors. For example, studies have determined the predictive abilities of macroeconomic and financial variables for stock market volatility (Nonejad, 2017; Paye, 2012). In recent years, with the introduction of economic policy uncertainty indices, some studies have also determined the predictive power of some economic policy uncertainty indicators (e.g., the economic policy uncertainty index, trade policy uncertainty index, monetary policy uncertainty index, equity market volatility index and geopolitical risk index) relative to stock market volatility (Alqahtani et al., 2020a; Gupta & Wohar, 2019; Li et al., 2020a; Olasehinde-Williams, 2021; Paye, 2012; Yu et al., 2018). In addition, studies have found evidence that the stock market can be driven by investors’ psychology (Daniel et al., 2002; Tseng, 2006) and determined the predictive power of some market sentiment indicators (see, e.g., Gupta et al., 2014; Perez-Liston et al., 2014; Oliveira et al., 2017; Jin et al., 2020; Liang et al., 2020a; Wang et al., 2020a). Other indicators, such as financial stress indices, have also been proven to have potential forecasting ability (Gupta et al., 2014; Singh, 2016; Sum, 2014). The predictive abilities of these potential predictors tend to change with changes in various external factors, making it more difficult to find stable predictors.
Conclusions
In this paper, we explore the effectiveness of 9 economic policy uncertainty indices, 4 market sentiment indicators and 2 financial stress indices in predicting the realized volatility of the S&P 500 index. We use several MIDAS-RV-X models to determine the predictive power of individual predictors and further construct PCA-, PLS- and LASSO-based MIDAS-RV extensions, a MIDAS-LASSO model with regime switching and various combination forecasts to find the optimal forecasting strategy. The out-of-sample forecasting performance of our forecasting models of interest is evaluated mainly by the MCS test and out-of-sample R2 statistics.
The empirical results first show that the economic policy uncertainty indicators (especially the EMV) and the VIX are more likely to produce better forecasting accuracy than the three market sentiment indicators and two financial stress indices. Among these predictors, the VIX has the strongest predictive power, followed by the EMV. Second, the combination forecasts and MIDAS-LASSO models usually have statistically significant relative forecasting performance. Across all forecasting models of interest, the MIDAS-LASSO model with regime switching has the best forecasting performance, followed by the MIDAS-LASSO model, indicating that LASSO technology can capture more valuable information from a large set of predictors than other information integration methods. The forecasting accuracy of MIDAS-LASSO with regime switching is also superior to that of the traditional HAR-LASSO method and its regime switching extension. When distinguishing between high- and low-volatility regimes, we find that the VIX is the best predictor under both scenarios. In addition, the economic policy uncertainty indicators other than GEPU either help predict high stock market volatility or low stock market volatility, but the market sentiment indices other than the VIX and the financial stress index (OFRFSI) exhibit statistically significant relative predictive performance only during low-volatility regimes. Across all considered forecasting methods, the MS-MIDAS-LASSO model still exhibits the best predictive performance during both high- and low-volatility regimes, but the MIDAS-LASSO model can outperform other forecasting methods except for MS-MIDAS-LASSO only during high-volatility regimes. We finally consider the impact of the COVID-19 epidemic. During the COVID-19 epidemic, the predictive power of the VIX diminished, and it was no longer the best out-of-sample predictor, while the predictive power of the EMV increased, making it the best predictor during this special period. However, the MIDAS-LASSO model with regime switching still exhibited the best forecasting performance across all considered forecasting models during the COVID-19 epidemic.