Yazar "Horo, Omar Hannan" seçeneğine göre listele

Listeleniyor 1 - 1 / 1
  • Küçük Resim
    Öğe
    Prediction of S&P500 stock market movement using Twitter sentiment analysis
    (İstanbul Bilgi Üniversitesi, 2021) Horo, Omar Hannan; Özyıldırım, Cenktan
    ABSTRACT: Twitter is amongst the biggest social networking services, offering its services to over 350 million active users as of 2021. For over the past few years, Twitter has been the most used social network platform as a tool of interaction between investors and stock market given its convenience and ability to move information from one part to another in a very short time. For over years, econometrics had always turn to the basic Ordinary Least Regression (OLS) to analyze the relationship between two variables and how much impact one variable could have another. However, during OLS analysis, the assumption is that the squared values of all errors terms at any particular point is constant (Homoskedasticity). Therefore, for a given set of data in a time series with different error terms (Heteroskedasticity) may result to a slight sense of false analysis and forecasting precision, hence leading us to go with ARCH and GARCH models, where heteroskedasticity is basically considered as variance to be accounted for and modeled. The aim of this study was to determine whether there is a significance impact of sentiment to the stock return in the time series. Despite several related studies, most papers mostly used Ordinary Least Squares with an assumption of linearity. Despite the original OLS showing significant influence of sentiments in our model, ARCH and GARCH models concluded that sentiment was not a significant predictor in the conditional variance.