Published Paper Details:

Bitcoin is known for its high volatility, making accurate price predictions challenging. This study aims to forecast Bitcoin prices for the upcoming month by incorporating exogenous variables specifically, interest rates and recession probabilities. The primary goal is to examine whether these external factors can improve the accuracy of Bitcoin price predictions. To achieve this, we employed two widely used time series forecasting models: Long Short-Term Memory (LSTM) and Facebook Prophet. We'll be implementing these project with several types of time series forecasting model like Arima model, eda model, lstm, facebook prophet, exponential smoothing, afima. Among all these models we'll be selecting one highly performing model. The research explores how the inclusion of interest rates and recession probabilities influences the performance of these models and compares their prediction results through visualizations and cross-validation techniques. We trained both models using historical Bitcoin price data combined with the selected exogenous variables and evaluated their predictive performance on a separate test dataset. Our findings reveal that the LSTM model outperforms Facebook Prophet in terms of prediction accuracy. While Facebook Prophet is optimized for statistical modeling, LSTM has the advantage of learning complex, non-linear patterns in data due to its deep learning architecture and ability to process sequential information effectively. Moreover, our results demonstrate that incorporating interest rates and recession probabilities significantly improves the forecasting accuracy of both models. This suggests that macroeconomic factors like interest rates and recession risks do have an impact on Bitcoin price movements. By integrating these external variables into the forecasting process, we show that predictive models can achieve better performance, making them more effective for forecasting Bitcoin prices. These findings highlight the importance of considering exogenous factors when predicting the behavior of highly volatile assets like Bitcoin.