Effect Of Social Media Sentiment On The Price Of Bitcoin

Abstract

Bitcoin started in 2009 as a peer-to-peer value transfer system. It quickly gained significant value against the dollar and became a traded asset. However, because of its age and its origin in the dark corners of the Internet, Bitcoin does not have formal communication channels where investors discuss its value. For example, a stock relies on its company for official announcement about the development of the company; investors could use this information to give valuation to and make decision to sell or buy the asset. Bitcoin does not have a formal channel because there is no entity behind it. For that reasons, it is necessary to study the channels that people use to communicate about Bitcoin. I studied the conversations on Twitter and Reddit in the days leading up to and the days following the point of all-time-high price of Bitcoin. I preferred Twitter because its data are automatically labeled as either bullish or bearish, but a restriction on the amount of data I could get access turned to Reddit instead. To study the sentiment hidden in the messages of Reddit, I wrote customary code to extract the most common words in the period of study. I planned on using them to construct a model that could automatically assign a sentiment to any Bitcoin conversation. Unfortunately, I discovered that conversations on Reddit are extremely specialized for Reddit users. Words on this website usually confer a different meaning from its regular dictionary meaning. To complete the model to truly extract the sentiment, there needs to be human effort put into manually labelling what sentiment the vocabulary on Reddit means. That next step would complete this thesis and give a more definitive answer to how informal and untraditional channels influence the price of an equally untraditional asset