Language Expression and Economic Value: An Empirical Study on Stock Index Prediction Based on Multi-Feature Emotional Analysis of Financial Discourse

Language Expression and Economic Value: An Empirical Study on Stock Index Prediction Based on Multi-Feature Emotional Analysis of Financial Discourse

Geyang Hu (Central University of Finance and Economics, China) and Yifang Liu (Central University of Finance and Economics, China)
Copyright: © 2023 |Pages: 15
DOI: 10.4018/JOEUC.321538
Article PDF Download
Open access articles are freely available for download

Abstract

As the stock market becomes increasingly diverse and complex, an accurate and feasible prediction of stock index has become an urgent demand for stock investors. As an important driver of changes in the stock market, financial discourse can guide investors' emotions, thus affecting the trading of stocks and the development of the stock market. Therefore, the prediction of stock index from the perspective of financial discourse emotion has gradually become a hotspot of research. Through the analysis of the existing literature, it is found that the models used in the current relevant research are not ideal, the prediction is not accurate, and there are problems such as single method, few selection indicators, narrow analysis area, etc. To solve the above problems, this study proposes an LSTM model integrating multiple feature emotional indexes, constructs the TextCNN emotional index and the metaphorical power index to quantitatively analyze the emotional expression and semantic use of news text, and then integrates the indexes into the LSTM neural network model to predict the Shanghai Stock Exchange Index.
Article Preview
Top

Introduction

With economic development, the financial market has gradually become a core part of the national economic system, and the pricing trend of the stock market reflects the national macroeconomic policy (Xu & Tian, 2021). Therefore, predicting and adjusting to the economic situation has become a hot research topic for scholars today. In the early research, scholars did not pay attention to the role of language factors in economic prediction and mainly used the traditional linear time series model for prediction. However, traditional methods have strict assumptions about data distribution, so good results cannot be achieved (Shen et al., 2021). In recent years, with the rapid development of artificial intelligence, data constructed by machine learning algorithms meet the nonlinear relationship (Gu et al., 2020). However, due to the dimension disaster, the prediction ability is reduced, so the selection of time series characteristic variables is a problem to be solved in a model application (Shen et al., 2021).

As an important driver of changes in the stock market, financial discourse can guide investor emotions, thus affecting the trading of stocks and the development of the stock market. At the same time, text information in the financial news contains normative data, high value, and much effective information (Xu & Tian, 2021), so it can be used as an important source of information for stock market prediction and can also effectively explain the irrational factors of stock market changes. Research combining linguistic text analysis with financial market prediction originated from abroad. Cowles (1933) was the first to combine text analysis with financial forecasting. He divided the articles in the Wall Street Journal into three categories—bullish, bearish, and uncertain—then applied them to the return prediction of the Dow Jones Industrial Average and achieved effective results in economic prediction. With the development of linguistics, research on language value has become more mature. Through the analysis and transformation of language factors in language texts, language data is generated and introduced into the study of financial market conditions. Among them, most studies mainly focus on emotional expression and semantic use.

Emotional expression refers to the evaluation of the emotion and attitude expressed by the language expressionist on a certain thing, which can be divided into positive emotion, neutral emotion, and negative emotion. It is the core content of the emotional analysis of the text, aiming to obtain the emotional category and attitude of people to things from the expressed language text (Ma Yuanyuan, Yu, 2023). Prosky et al. (2017) used the Google natural language processing API software to analyze the sentiment of financial news text and combined it with the LSTM model to realize stock market trend prediction. Yan (2021) carried out an emotional analysis of the stock market text in the stock bar and forum and used it as input for the ARIMA-LSTM combination prediction model, which improved the accuracy of stock price prediction. Li et al. (2021) used the bidirectional time convolution network to extract the features of language text and carry out emotional analysis to obtain positive or negative emotional classification and applied it to stock market prediction.

As an important way of language expression, metaphor is the language act of using one thing to analogy another; it also represents the psychological activities of the language expression. Metaphor is usually more emotional than literal expression and has important analytical value (Mohammad, 2016). Through metaphor analysis of language, researchers can explore the economic environment and ideology behind the use of language, and thus predict the economic trend (Melissa, 2015). López and Llopis (2010) classified metaphorical discourse in financial news into two categories—positive metaphor and negative metaphor—and carried out a quantitative analysis. It was found that this indicator can reflect the social and economic environment, thus realizing economic prediction. De Landtsheer (2015) proposed the metaphor power index (MPI), which linked metaphors in language with economic indicators. Research found that MPI can reflect the economic situation of the current environment, and the value of MPI was negatively correlated with the economic situation.

Complete Article List

Search this Journal:
Reset
Volume 37: 1 Issue (2025)
Volume 36: 1 Issue (2024)
Volume 35: 3 Issues (2023)
Volume 34: 10 Issues (2022)
Volume 33: 6 Issues (2021)
Volume 32: 4 Issues (2020)
Volume 31: 4 Issues (2019)
Volume 30: 4 Issues (2018)
Volume 29: 4 Issues (2017)
Volume 28: 4 Issues (2016)
Volume 27: 4 Issues (2015)
Volume 26: 4 Issues (2014)
Volume 25: 4 Issues (2013)
Volume 24: 4 Issues (2012)
Volume 23: 4 Issues (2011)
Volume 22: 4 Issues (2010)
Volume 21: 4 Issues (2009)
Volume 20: 4 Issues (2008)
Volume 19: 4 Issues (2007)
Volume 18: 4 Issues (2006)
Volume 17: 4 Issues (2005)
Volume 16: 4 Issues (2004)
Volume 15: 4 Issues (2003)
Volume 14: 4 Issues (2002)
Volume 13: 4 Issues (2001)
Volume 12: 4 Issues (2000)
Volume 11: 4 Issues (1999)
Volume 10: 4 Issues (1998)
Volume 9: 4 Issues (1997)
Volume 8: 4 Issues (1996)
Volume 7: 4 Issues (1995)
Volume 6: 4 Issues (1994)
Volume 5: 4 Issues (1993)
Volume 4: 4 Issues (1992)
Volume 3: 4 Issues (1991)
Volume 2: 4 Issues (1990)
Volume 1: 3 Issues (1989)
View Complete Journal Contents Listing