The effect of financial news on BIST stock prices: a machine learning approach

Download
2018
Kanmaz, Medet
This thesis examines the relationship between the price data of companies in different sectors in the Borsa Istanbul (BIST) stock exchange and the verbal data revealed in the financial news related to these companies. In this work, sentiment analysis, natural language processing and the effect of financial news on individual stock performances are studied with a simple and novel method. Sentiment analysis is created by automatically labelling the news for companies publicly traded in BIST as positive or negative on the basis of the daily performance of stocks with different methods in machine learning. These algorithms determine the polarity in financial news with an accuracy of around 70%. As a result of this study, it was seen that positive or negative news had a positive / negative effect on the related stock prices. Whether the outcome of this algorithm provides incentives to make a profit in the market or not is also questioned. On the other hand, it is shown that it is hard to gain profit from this public information unless there is insider information.

Suggestions

Cross sectional determinants of Turkish stock market returns
Çeliker, Umut; Rhoades, Seza; Department of Business Administration (2004)
This thesis analyzes the relationship between stock returns and firm-specific characteristics including market beta, size, book-to-market ratio, leverage, earnings yield, net sales-to-price ratio and prior return performance in Istanbul Stock Exchange during the period 1993-2003. Moreover, the predictability of some macroeconomic variables based on the stock market return behavior is investigated.
A Note on Cross Section of Stock Returns on the Istanbul Stock Exchange
Kayaçetin, Nuri Volkan; Güner, Zehra Nuray (Boğaziçi Üniversitesi, 2007-01-01)
This paper investigates the explanatory powers of firm size, book-to-market, sales-to-price, debt-toequity ratio, and dividend yield on the cross-section of returns on the Istanbul Stock Exchange (ISE). Our results indicate that each of these variables (except dividend yield), commands a significant return premium when included in a simple regression model with the highest premium being associated with sales-to-price ratio. Our results suggest that sales-to-price ratio and debt-to-equity ratio have higher e...
An Examination of betas for Borsa İstanbul
Koca, Onur; Oran, Adil; Department of Business Administration (2016)
This study aims to investigate the Betas, which is called as systematic risk and introduced by Capital Asset Pricing Model (CAPM), of stocks traded in Borsa Istanbul (BIST). Issues about Beta have been examined for many years and mainly focus on its estimation and stability. These topics set up the core of this thesis. The closing prices of 203 eligible stocks between 2005 and 2015 are used in the work and data is collected from Thomson Reuters. The estimation of Beta is performed in four different methods,...
Volatility indexes and an implementation of the Turkish BIST 30 index
Karakurt, Caner; Uğur, Ömür; Department of Financial Mathematics (2018)
In 1993, by representing of CBOE Vix, global financial markets met volatility indexes. In 2003, methodology of the CBOE Vix is updated and it took the form which used today. Day after day, volatility indexes have attracted more and more investors and financial institutions, and soon volatility indexes have succeeded in becoming one of the most followed financial indicators. Following these developments, many countries have introduced their implied volatility indexes by using CBOE Vix methodology or its vari...
A Test of multi-index asset pricing models: the case of Istanbul Stock Exchange
Kalaç, Sırrı Selim; Danışoğlu, Seza; Department of Business Administration (2012)
This study employs widely excepted asset pricing models to test their explanatory power in the context of Istanbul Stock Exchange listed companies between 1990 and 2010. The risk factors, beta, size, book-to-market equity, and momentum are used to form portfolios and their factor loadings are estimated. The results of this study are mostly in line with the previous academic research, and some unique attributes of the return generation mechanism of Istanbul Stock Exchange are reported.
Citation Formats
M. Kanmaz, “The effect of financial news on BIST stock prices: a machine learning approach,” M.S. - Master of Science, Middle East Technical University, 2018.