Correlation analysis between Reddit sentiments and Ether (ETH) price action

Abstract

Cryptocurrencies (Crypto) have seen a tremendous growth in recent years. According to CNBC, the adoption of Cryptocurrencies around the world was up by 881% in the year 2020 [1] Several factors contributed to Crypto’s rise, among them is due to a rising number of businesses across a variety of industries embracing cryptocurrencies. These businesses allow their customers to pay with crypto in exchange for their goods and services, officially acknowledging cryptocurrencies as an alternative form of payment [2] Furthermore, Cryptocurrencies are now being recognized as an alternative form of investment, with Morgan Stanley leading the way in offering their clients access to Bitcoin funds [3] As a result, interest in the study of cryptocurrencies and its’ market behaviour are ever growing.

This final year project’s undertaking is to contribute into this field of research. The purpose of this project is to determine if there exists a correlation between Social Media Sentiments and price action of cryptocurrencies. Specifically, this project attempts to determine if there exists a correlation between Social Media Sentiments captured from Reddit and the price action of Ether (ETH) coin (A type of Cryptocurrency). Data were collected with a few codes written in Python programming language and the use of some Application Programming Interfaces (API). These are, Python Reddit API Wrapper (PRAW) and Google News API. Three different Time Periods were identified from the past year to narrow down the investigation on Correlation analysis between Reddit Sentiments and Ether price action. Correlation Significance tests were then conducted for each Time period to prove that the findings did not happen by chance and holds true with the larger population.

The investigation findings show that there is no correlation between Reddit Sentiments and the price action of Ether, based on the sample correlation coefficients obtained and Significance Testing results. Future researchers could try to expand their Data sources and explore other Sentiment Analyser tools for the Sentiment analysis in order to improve this study. Another suggestion in improving the study is to improve filtering of collected Data before pre-processing.