CBISI-LSTM Deep Learning Model for Short-term Cross-border Capital Flow Prediction

Abstract

With the drastic fluctuation of the international financial market in recent years, the cross-border capital flow between Shanghai and Hong Kong has become increasingly active. The lack of effective and timely tracking monitoring and scientific management of cross-border capital flow in the capital market will seriously affect the overall financial security of China's economy. This paper constructs the cross-border investor sentiment index CBISI based on principal component analysis and analyzes the impact of cross-border investor sentiment and cross-border capital flows by constructing the VAR model. In addition, CBISI is used as part of the input variable of LSTM to forecast the cross-border capital flow (NF). The findings of the study indicate that changes in cross-border investor sentiment will have a significant short-term impact on cross-border capital flows, and the addition of CBISI will improve the accuracy of cross-border flow estimates.

Keywords

• cross-border funds
• data mining
• investor sentiment
• operation management
• LSTM