Investigating the informativeness of technical indicators and news sentiment in financial market price prediction

被引:53
作者
Farimani, Saeede Anbaee [1 ]
Jahan, Majid Vafaei [1 ]
Fard, Amin Milani [2 ]
Tabbakh, Seyed Reza Kamel [1 ]
机构
[1] Islamic Azad Univ, Dept Comp Engn, Mashhad Branch, Mashhad, Iran
[2] New York Inst Technol Vancouver, Dept Comp Sci, Vancouver, Canada
关键词
Market prediction; Transformer-based language models; Financial sentiment analysis; Information gain; FinBERT; SOCIAL MEDIA; STOCK; HEADLINES; MODEL;
D O I
10.1016/j.knosys.2022.108742
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Real-time market prediction tool tracking public opinion in specialized newsgroups and informative market data persuades investors of financial markets. Previous works mainly used lexicon-based sentiment analysis for financial markets prediction, while recently proposed transformer-based sentiment analysis promise good results for cross-domain sentiment analysis. This work considers temporal relationships between consecutive snapshots of informative market data and mood time series for market price prediction. We calculate the sentiment mood time series via the probability distribution of news embedding generated through a BERT-based transformer language model fine-tuned for financial domain sentiment analysis. We then use a deep recurrent neural network for feature extraction followed by a dense layer for price regression. We implemented our approach as an open-source API for real-time price regression. We build a corpus of financial news related to currency pairs in foreign exchange and Cryptocurrency markets. We further augment our model with informative technical indicators and news sentiment scores aligned based on news release timestamp. Results of our experiments show significant error reduction compared to the baselines. Our Financial News and Financial Sentiment Analysis RESTFul APIs are available for public use.(c) 2022 Elsevier B.V. All rights reserved.
引用
收藏
页数:11
相关论文
共 83 条
[1]  
Agarwal B, 2020, ALGO INTELL SY, P1, DOI 10.1007/978-981-15-1216-2
[2]  
Alamatian Z., 2021, P 34 CANADIAN C ARTI, DOI [10.21428/594757db.8d600287, DOI 10.21428/594757DB.8D600287]
[3]  
Alonso M.N. i, 2021, ARXIV211100526
[4]  
[Anonymous], 2021, ROBONEWS
[5]  
[Anonymous], 2020, TA PYTHON PACKAGE
[6]  
[Anonymous], 2021, FINBERT
[7]  
Araci D., 2019, ARXIV PREPRINT ARXIV
[8]   Using frame-based resources for sentiment analysis within the financial domain [J].
Atzeni, Mattia ;
Dridi, Amna ;
Recupero, Diego Reforgiato .
PROGRESS IN ARTIFICIAL INTELLIGENCE, 2018, 7 (04) :273-294
[9]   Twitter mood predicts the stock market [J].
Bollen, Johan ;
Mao, Huina ;
Zeng, Xiaojun .
JOURNAL OF COMPUTATIONAL SCIENCE, 2011, 2 (01) :1-8
[10]   Augmented Textual Features-Based Stock Market Prediction [J].
Bouktif, Salah ;
Fiaz, Ali ;
Awad, Mamoun .
IEEE ACCESS, 2020, 8 (08) :40269-40282