Social Text-driven Hybrid Deep Sequential Stock Prediction Model

In addition to only considering stocks' price series,utilizing short and instant texts from social medias like Twitter has potential to yield better stock market prediction.While some previous approaches have explored this direction,their results are still far from satisfactory due to their reliance on performance of sentiment analysis and limited capabilities of learning direct relations between target stock trends and their daily social texts.In order to solve these problems,this paper first proposes a coupled Long-short Term Memory network to replace the original stock price features using the labeled Latent Dirichlet Allocation Model to extract more fine-grained feature sequence representations as inputs,establishes a bi-layered recurrent neural network to model the social text data and further integrates the data of the two different modal data in the upper layer.On the basis of the coupled Long-short Term Memory network,this paper propose a novel Cross-modal attention based Hybrid Recurrent Neural Network(CH-RNN).Specifically,CH-RNN consists of two essential modules.One adopts a two-stage attention recurrent neural network to gain stock trend representations for different stocks.The other utilizes recurrent neural network to model daily aggregated social texts.These two modules interact seamlessly by the following two manners: 1)daily representations of target stock trends from the first module are leveraged to select trend-related social texts through a cross-modal attention mechanism,and 2)representations of text sequences and trend series are further integrated.This paper crawls the tweet text data and the stock price data of Yahoo Finance with a time span of one year.Through a comprehensive experiment on the real dataset,it can be proved that the coupled Long-short Term Memory network combines with the two modal data and further merges the two parts with the neural network,resulting in a relatively obvious prediction improvement.In the Cross-modal attention based Hybrid Recurrent Neural Network,it is proved that the reasonable representation and fusion of multi-modal data can effectively filter the social text noise that is not helpful for prediction,which can produce good prediction results and return on investment.