Framework for Assessment of Bitcoin Price Prediction Using Ensembling Machine Learning Approach
Amisha Shimpi, Rohan Gambhir, Abhishek Diwate, Sahil Udawant, Nazema Y. Siddiqui, P. William
Abstract
As crypto market knowledge is not widely available to new investors; this problem may lead to wrong investments and financial losses. Inadequate understanding of the elements that have a significant impact on the price of bitcoin might potentially result in poor investing decisions. Now, there are more than $230 billion worth of openly traded cryptocurrencies on the market. The main purpose of Bitcoin, the most valuable cryptocurrency, which also has the best price predictability, is to act as a digital store of wealth. Bitcoin is a digital currency that is used all over the world for advanced payments or mostly for speculation. For instance, Bitcoin is decentralized because no one owns it. Bitcoin exchanges are easy since they are not tied to any one country. This research study employs a variety of machine learning approaches to predict Bitcoin values. Accurate price prediction is crucial for making wise investing decisions nowadays because of its high volatility feature. The price of bitcoin is initially divided into daily and high-frequency categories in this study. A collection of high-dimensional characteristics and fundamental trading features are used, respectively, for its daily and price forecast. Next, we see that while sophisticated machine learning algorithms like SVM accurately forecast pricing, statistical approaches like linear regression do not. The significance of sample dimensions in machine learning techniques is acknowledged in this study on predicting Bitcoin values.