Graph-based stock correlation and prediction for high-frequency trading systems

•Identify that the traditional factor-based risk control methods in high frequency trading have become the bottleneck for trading systems in Chinese A-share market.•Propose a novel GALSTM model combined with multi-Hawkes process to learn the correlations between stocks and predict the prices of these stocks.•Build the corresponding data process module, portfolio management module and strategy implementation module to collaborate with the proposed GALSTM model.•The proposed system has achieved stable income in real Chinese A-share market for three months.