LSTM and Prophet Model Fusion Framework for Bitcoin Candlestick Visualization and Price Prediction Forecasting

Bitcoin is a decentralized digital currency that has climbed substantially in recent years, reflecting the fact about the amount of rise in its price. Even though investors and traders pay a lot of attention to Bitcoin, there is very little we know about its actual nature. Bitcoin's value is incredibly fickle. No doubt, there are many advantages of trading and investing in Bitcoin, but there are also many risks due to the inconstant nature of its price. Price forecasting is necessary to reduce the danger posed by Bitcoin's price volatility. The prior studies favored using only past (or historical) data and neglected real-time data. Ultimately, the results had some inaccuracies. To combat this issue and make a more accurate prediction, this work suggests integrating both past and live data. The objective of this paper is to visualize the bitcoin candlesticks (i.e., Open, High, Low, and Closing prices) and forecast the future price using historical and real-time data. The dataset used for the current study is taken from Tiingo. The specific reason for choosing Tiingo is that its APIs are built to be performant, and consistent and support extensive filters to speed up our development time. We specifically used a fusion of two deep learning models, the Stacked LSTM (Long Short-term memory) model and Prophet model. The outcomes of this study assist traders and bitcoin consumers to make educated choices regarding bitcoin investments. Moreover, this study is a crucial tool for researchers to comprehend the behaviors of the bitcoin market. Therefore, the appropriate question, to begin with, is: Can Deep Learning accurately forecast bitcoin prices?