Author(s)

Zedong Cai

Corresponding Author

Zedong Cai

ABSTRACT

As a kind of financial derivative with price discovery function, stock index future has a wide range of applications in risk management. However, the high leverage and short selling mechanism of stock index futures also greatly increased the trading risk. This paper takes CSI 300, SSE 50, and CSI 500 stock index futures and their corresponding indexes as the research objects. Using the relevant data of the past five years, this paper first constructs VAR model, analyzes the impact of futures market sentiment on stock index volatility by impulse response, and then constructs XGBoost model (a machine learning algorithm), uses futures market sentiment for series fitting and prediction. The research finds that: First, futures market sentiment has several periods of impact on the stock index volatility, and the length of different research targets varies; Second, the sentiment many periods before can still provide important information, which cannot be ignored when fitting the volatility of the sequence; Third, the fitted sequence generally underestimates the extreme risk of the real stock market, mainly due to the over-reaction of stock market investors to the transmission of sentiment in the futures market. Finally, based on the research findings, the paper puts forward relevant recommendations.

KEYWORDS

Stock index futures, stock market, sentiment, extreme risk, fit and predict