Stock Prediction Based On Event Graph

Stock market forecasting has always been a classic but challenging problem. We propose a method of stock price prediction based on the financial event graph. First, based on the deep learning method, events are extracted from the news to build the event graph. Second, the event graph is expressed by the TransD translation model and expressed as dense vectors. Last, taking the encoding vector of the event graph as input, we select the logistic regression model as the stock prediction model and get the output of the stock fluctuations of the next day. The accuracy and F1 score obtained by this method exceed the baseline model, which proves the effectiveness of the algorithm proposed by us.