Disclosing the Relationship Between Nasdaq and Bitcoin via Artificial Intelligence Approach: Machine Learning and Deep Learning

Abstract

Cryptocurrency has grown in popularity recently, and people try to get revenue from investing in cryptocurrency because the fluctuation is much larger than the stock market. However, because of this characteristic, some people lose a huge amount of money all of a sudden, or some people are addicted to it. Some experts claim that Nasdaq and Bitcoin exhibit a similar trend, showing a positive relationship between the two. Therefore, our research tries to prove that hypothesis with an artificial intelligence approach. The close price of the Nasdaq was used as an input variable, while the close price from Bitcoin and Ethereum was set as the target. Our experiment involves two different steps, which are EDA and machine learning analysis. For the first experiment, correlation matrix and joint plots were plotted, and Nasdaq and Bitcoin showed high correlation. Furthermore, six different machine learning algorithms and two deep learning algorithms were utilized for the analysis, and their accuracy was assessed with RMSE and MAE. RMSE and MAE of two coins based on Nasdaq were 1837.09 and 1066.32 for Bitcoin and 127 and 96.8 for Ethereum price prediction. However, even though the GRU yielded higher RMSE and MAE compared to the machine learning algorithm, visualization indicated that GRU was more effective than the machine learning model. Even though our research reveals a lower RMSE and MAE score compared to similar existing research, our finding is outstanding in that we proved that there exists a positive relationship between Nasdaq and Bitcoin.