Advanced

How 140 Characters can be related to the Stock Market Movements: Sentiment Analysis of Twitter

Shehadeh, Mohammad LU and Khritantsev, Maksim LU (2018) NEKN02 20181
Department of Economics
Abstract
Stock market movements forecast based on sentiment analysis is certainly a field worth investigating. Being able to build future investment strategies based on forecasted stock returns would be of tremendous importance for individual investors and high-frequency trading firms. This thesis aims to closely investigate the impact that tweet sentiments in Twitter have on stock market movements; it intends to empirically test the prediction power of public sentiments of tweets with respect to stock price returns of S&P 500 companies. In the analysis of the study, this paper uses a sample consisting of 181 776 tweets, collected between December 26th, 2017 and March 15th, 2018. The results of the dissertation present significant evidence of... (More)
Stock market movements forecast based on sentiment analysis is certainly a field worth investigating. Being able to build future investment strategies based on forecasted stock returns would be of tremendous importance for individual investors and high-frequency trading firms. This thesis aims to closely investigate the impact that tweet sentiments in Twitter have on stock market movements; it intends to empirically test the prediction power of public sentiments of tweets with respect to stock price returns of S&P 500 companies. In the analysis of the study, this paper uses a sample consisting of 181 776 tweets, collected between December 26th, 2017 and March 15th, 2018. The results of the dissertation present significant evidence of dependence between tweet sentiments and stock returns, using lexicon-based panel data regression. The closing results show, using complex algorithmic machine learning techniques, that random forest and SVM-Gaussian are the optimal models in the prediction of stock returns based on tweets under unigram, bigram and trigram methods. (Less)
Please use this url to cite or link to this publication:
author
Shehadeh, Mohammad LU and Khritantsev, Maksim LU
supervisor
organization
course
NEKN02 20181
year
type
H1 - Master's Degree (One Year)
subject
keywords
stock returns, tweet sentiments, lexicon based, panel data, regression analysis, machine learning, support vector machines, random forest
language
English
id
8949187
date added to LUP
2018-07-02 15:38:16
date last changed
2018-07-02 15:38:16
@misc{8949187,
  abstract     = {Stock market movements forecast based on sentiment analysis is certainly a field worth investigating. Being able to build future investment strategies based on forecasted stock returns would be of tremendous importance for individual investors and high-frequency trading firms. This thesis aims to closely investigate the impact that tweet sentiments in Twitter have on stock market movements; it intends to empirically test the prediction power of public sentiments of tweets with respect to stock price returns of S&P 500 companies. In the analysis of the study, this paper uses a sample consisting of 181 776 tweets, collected between December 26th, 2017 and March 15th, 2018. The results of the dissertation present significant evidence of dependence between tweet sentiments and stock returns, using lexicon-based panel data regression. The closing results show, using complex algorithmic machine learning techniques, that random forest and SVM-Gaussian are the optimal models in the prediction of stock returns based on tweets under unigram, bigram and trigram methods.},
  author       = {Shehadeh, Mohammad and Khritantsev, Maksim},
  keyword      = {stock returns,tweet sentiments,lexicon based,panel data,regression analysis,machine learning,support vector machines,random forest},
  language     = {eng},
  note         = {Student Paper},
  title        = {How 140 Characters can be related to the Stock Market Movements: Sentiment Analysis of Twitter},
  year         = {2018},
}