Incorporating Fine-grained Events in Stock Movement Prediction

Abstract

Considering event structure information has proven helpful in text-based stock movement prediction. However, existing works mainly adopt the coarse-grained events, which loses the specific semantic information of diverse event types. In this work, we propose to incorporate the fine-grained events in stock movement prediction. Firstly, we propose a professional finance event dictionary built by domain experts and use it to extract fine-grained events automatically from finance news. Then we design a neural model to combine finance news with fine-grained event structure and stock trade data to predict the stock movement. Besides, in order to improve the generalizability of the proposed method, we design an advanced model that uses the extracted finegrained events as the distant supervised label to train a multi-task framework of event extraction and stock prediction. The experimental results show that our method outperforms all the baselines and has good generalizability.

News sensitive stock market prediction: literature review and suggestions

“…So textual data is represented as structured events which contain entity-relation information ( Ding et al, 2014 ). The use of domain knowledge along with event extraction method refines this process and provides opportunity to use less training data with interpretable and traceable results ( Hogenboom et al, 2016 ; Chen et al, 2019 ). Moreover, event embedding gives the distributed representation of structured events which significantly reduces the issue of high dimensionality.…”

“…The promising feature of deep learning techniques is that they can extract features from data through learning. The work of Ding et al (2015) , Ding et al (2016) , Vargas, De Lima & Evsukoff (2017) , Chen et al (2019) , Deng et al (2019) and Li, Wu & Wang (2020) shows the strength of deep learning for prediction along with event base textual representation and sentiment analysis based features ( Picasso et al, 2019 ; Li, Wu & Wang, 2020 ). Solutions of the challenges in implementing news sensitive stock prediction are discussed in this section and are summarized in Table 2 .…”

“…Market Full-size  DOI: 10.7717/peerj-cs.490/fig- 1 data is preprocessed to remove inconsistencies and noise to achieve better prediction accuracy. It is also processed in order to select features and derive new features from existing one to align with textual features (Chen et al, 2019;Picasso et al, 2019;Li, Wu & Wang, 2020;Liu et al, 2020). Financial time series data encompasses dynamic and nonlinear historical data.…”

“…Event extraction based on expert knowledge with small data corpus improves search performance of information extraction technique ( Borsje, Hogenboom & Frasincar, 2010 ; Hogenboom et al, 2013 ). Chen et al (2019) extracted fine grained events automatically using a finance event dictionary built by domain experts. They proposed a professional financial event dictionary that contains all main financial events along their roles, and trigger words.…”

Stock Embeddings Acquired from News Articles and Price History, and an Application to Portfolio Optimization

“…Recent studies (Ding et al, 2015;Hu et al, 2018;Chen et al, 2019;Yang et al, 2018) have indeed reported that news articles can be leveraged to improve the accuracy of predicting stock price movements. These previous works have used deep learning techniques.…”