Joseph Azanza
Joseph Matthew R. Azanza
Data Scientist for a US-based cloud communications provider,
with expertise in Machine Learning, Artificial Intelligence,
Data Wrangling, Data Storytelling,
Sales Analytics, Sales Operations (Ops),
Marketing Analytics, Marketing Ops,
Business Intelligence, Strategic Initiatives,
Molecular Biology, and Biotechnology,
MS in Data Science
Asian Institute of Management
BS in Molecular Biology and Biotechnology
University of the Philippines Diliman
Predicting the Daily Returns of Stocks in the Philippine Stock Exchange via Natural Language Processing of News Articles
Joseph Matthew Azanza | Eugenio Cedric Corro | Kristoffer Dave Tabong
Asian Institute of Management
Abstract
In this study, we are predicting the movement of daily stock returns in the Philippine Stock Exchange Index (PSEi) using business news articles. We scraped business news articles from Rappler, encompassing news from 2015 to 2020 and we obtained the historical stock prices of the PSEi for the same period. Sentiment scores were assigned to the articles using the gensim.summarization.keywords module and the VADER model. These sentiment scores were then used as input to various machine learning models, with the goal of predicting the movement of stock
returns with three-day delay, as either positive (+1) or negative (-1). Our best machine learning model is a Random Forest Classifier with a test accuracy of 65.46%. Overall, we developed a machine learning model that incorporates Natural Language Processing and the novel concept of “delayed results” via the 3-day delay prediction. End users can leverage our model to augment their decision making of buying, holding, or selling stocks given business news only
Keywords: Natural Language Processing, Sentiment Analysis, Vader, Philippine Stock Market Prediction
Source code can be provided upon request, and upon approval of all project collaborators