Elucidating Cryptocurrency with Trading Dashboard

Abstract

With the rise of interest in cryptocurrency in the recent decade, an ocean of financial news data has surfaced in articles, tweets, and even Reddit posts. Due to the sheer volume, it is not practical for the casual trader to read through all these news sources manually. However, only going through one or two sources alone may result in receiving biased information, or no useful information at all. With the current rise in cryptocurrency, accurately predicting market trends becomes highly beneficial to the user, providing a major opportunity for lower-income households to have a higher chance of profiting and living a substantially more comfortable lifestyle. In this study, a developer's API key was obtained for three news sources to scrape financial news from. Then, the TensorFlow Keras model and Gensim model's doc2vec NLP tool were utilized to process the data scraped online. The data is then saved as a .model and .sav file, and a website was constructed using the Flask framework. The website is now deployed and is available for all users. However, because the data obtained was too small to be utilized well, only a weak linear model that could give us a correlation between price and news sentiment was able to be constructed. The dashboard passed its functional and UAT tests with 100%, and via the usability test with SUS, the dashboard is considered to be easy to use. In all, the website summarizes the main details and sentiment of the coins and will benefit users who are just being introduced to the cryptocurrency space.