TY - GEN
T1 - Stock Price Trend Prediction Using LSTM and Sentiment Analysis on News Headlines
AU - Li, Jung Bin
AU - Lin, Szu Yin
AU - Leu, Fang Yie
AU - Chu, Yen Chu
N1 - Publisher Copyright:
© 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.
PY - 2023
Y1 - 2023
N2 - To simulate the trading behavior of investors in the stock market, this study adopts parameters including technical, fundamental, and chip to build a LSTM model, and also observes the ability of news sentiment to predict stock prices. Influential stocks such as TSMC, Fulgent Sun, and HTC are chosen as the target of our experiment. Four common natural language processing packages are used to label news sentiment. Then the combined sentiment labels along with the LSTM model are used for backtesting. The results of the study found that FinBERT's ability to predict the price trend outperforms other methods, with an accuracy of 41.6%. In addition, combining news sentiment labels with the LSTM model generally leads to better outcome than using either the news label or the LSTM model alone. However, in certain extreme cases, traditional technical indicators or even buy-and-hold strategy have better performances.
AB - To simulate the trading behavior of investors in the stock market, this study adopts parameters including technical, fundamental, and chip to build a LSTM model, and also observes the ability of news sentiment to predict stock prices. Influential stocks such as TSMC, Fulgent Sun, and HTC are chosen as the target of our experiment. Four common natural language processing packages are used to label news sentiment. Then the combined sentiment labels along with the LSTM model are used for backtesting. The results of the study found that FinBERT's ability to predict the price trend outperforms other methods, with an accuracy of 41.6%. In addition, combining news sentiment labels with the LSTM model generally leads to better outcome than using either the news label or the LSTM model alone. However, in certain extreme cases, traditional technical indicators or even buy-and-hold strategy have better performances.
UR - http://www.scopus.com/inward/record.url?scp=85141706050&partnerID=8YFLogxK
U2 - 10.1007/978-3-031-20029-8_27
DO - 10.1007/978-3-031-20029-8_27
M3 - Conference contribution
AN - SCOPUS:85141706050
SN - 9783031200281
T3 - Lecture Notes in Networks and Systems
SP - 282
EP - 291
BT - Advances on Broad-Band Wireless Computing, Communication and Applications - Proceedings of the 17th International Conference on Broad-Band Wireless Computing, Communication and Applications BWCCA-2022
A2 - Barolli, Leonard
PB - Springer Science and Business Media Deutschland GmbH
T2 - 17th International Conference on Broadband Wireless Computing, Communication and Applications, BWCCA 2022, held in conjunction with the 17th International Conference on P2P, Parallel, Grid, Cloud and Internet Computing, 3PGCIC 2022
Y2 - 27 October 2022 through 29 October 2022
ER -