Correlations and flow of information between the New York Times and stock markets

被引:13
|
作者
Garcia-Medina, Andres [1 ]
Sandoval Junior, Leonidas [3 ]
Urrutia Banuelos, Efrain [2 ]
Martinez-Arguello, A. M. [4 ]
机构
[1] CONACYT, Ctr Invest Matemat AC, CIMAT Unidad Monterrey PITT, Monterrey, Nuevo Leon, Mexico
[2] Univ Sonora, Phys Res Dept, Hermosillo 83000, Sonora, Mexico
[3] Insper, Inst Ensino & Pesquisa, Rua Quata 300, BR-04546240 Sao Paulo, SP, Brazil
[4] Benemerita Univ Autonoma Puebla, Inst Fis, Apartado Postal J-48, Puebla 72570, Mexico
关键词
Random matrix theory; Transfer entropy; Sentiment analysis; Behavioral finance; RANDOM-MATRIX THEORY; CROSS-CORRELATIONS; TRANSFER ENTROPY; SENTIMENT; DISTRIBUTIONS; NETWORK;
D O I
10.1016/j.physa.2018.02.154
中图分类号
O4 [物理学];
学科分类号
0702 ;
摘要
We use Random Matrix Theory (RMT) and information theory to analyze the correlations and flow of information between 64,939 news from The New York Times and 40 world financial indices during 10 months along the period 2015-2016. The set of news is quantified and transformed into daily polarity time series using tools from sentiment analysis. The results show that a common factor influences the world indices and news, which even share the same dynamics. Furthermore, the global correlation structure is found to be preserved when adding white noise, what indicates that correlations are not due to sample size effects. Likewise, we find a considerable amount of information flowing from news to world indices for some specific delay. This is of practical interest for trading purposes. Our results suggest a deep relationship between news and world indices, and show a situation where news drive world market movements, giving a new evidence to support behavioral finance as the current economic paradigm. (C) 2018 Elsevier B.V. All rights reserved.
引用
收藏
页码:403 / 415
页数:13
相关论文
共 50 条