Search

In finance, it is believed that market information, namely, fundamentals and news information, affects stock movements. Such media-aware stock movements essentially comprise a multimodal problem. Two unique challenges arise in processing these multimodal data. First, information from one data mode will interact with information from other data modes. A common strategy is to concatenate various data modes into one compound vector; however, this strategy ignores the interactions among different modes. The second challenge is the heterogeneity of the data in terms of sampling time. Specifically, fundamental data consist of continuous values sampled at fixed time intervals, whereas news information emerges randomly. This heterogeneity can cause valuable information to be partially missing or can distort the feature spaces. In addition, the study of media-aware stock movements in previous work has focused on the one-to-one problem, in which it is assumed that news affects only the performance of the stocks mentioned in the reports. However, news articles also impact related stocks and cause stock co-movements. In this article, we propose a tensor-based event-driven LSTM model to address these challenges. Experiments performed on the China securities market demonstrate the superiority of the proposed approach over state-of-the-art algorithms, including AZFinText, eMAQT, and TeSIA. [ABSTRACT FROM AUTHOR]

There are two unique challenges in current research of price volatility studies in the crude oil futures market, which is how to capture the impact of interactivity between data as much as possible when multi-dimensional data are fused, and how to solve the problem of temporal heterogeneity of research samples existing in multi-dimensional data. We proposed a tensor-based event-driven gated recurrent unit (Event-GRU) deep learning model that helps to solve the above dilemmas. Our model can effectively prevent the loss of interaction relationship of multi-dimensional data by taking the advantage of spatial data representation of tensor, as well as using news occurrence as a trigger to construct the event-driven mechanism based on GRU model, in order to overcome the feature space distortion between discrete news data and continuous trading fundamentals data due to the uneven sample time matching. Regarding the current insufficient research of crude oil prices in China, we addressed the crude oil futures trading data from the Shanghai Futures Exchange and 31,422 news data of the crude oil futures market from April 2018 to August 2020. We used the XGBoost algorithm (eXtreme Gradient Boosting) to evaluate and filter. Besides, we built a tensor-based event-GRU deep learning model to finally analyze the closing price on the day after. This research contributes to improving the accuracy of crude oil futures price analysis compared with using benchmark models such as DT, SVM, BP, LSTM, and GRU, and laying the foundation for the subsequent opening of corresponding quantitative investment strategies. • Price volatility in crude oil futures market is a multidata fusion modeling problem. • Tensor documents the interaction between trading data and news information. • Event-GRU enhances the ability to capture multidimensional data through tensor. [ABSTRACT FROM AUTHOR]