Decision Tree for Bitcoin Price Prediction Based on Market Factors

Abstract

The volatile nature of Bitcoin poses significant challenges for accurate price prediction, which is critical for informed decision-making by investors and policymakers. This study explores the application of decision tree algorithms to predict Bitcoin prices using a dataset comprising historical data on Bitcoin prices, market capitalization, and trading volumes. The research emphasizes feature engineering techniques, including derived metrics such as rolling averages and volatility indices, and integrates ensemble methods like Random Forest and Gradient Boosting to enhance predictive performance. The decision tree model achieved an accuracy of 53%, demonstrating its capability to capture general trends in Bitcoin price movements, particularly during high volatility periods. The study highlights the importance of key features such as the Relative Strength Index (RSI) and Moving Averages (MA14) while identifying limitations in predicting price decreases. Recommendations for future research include integrating external data sources, such as sentiment analysis and macroeconomic indicators, and exploring advanced modeling techniques to improve robustness and accuracy. This research contributes to the growing field of cryptocurrency price prediction by providing interpretable and actionable insights into market dynamics. The findings offer valuable tools for analysts and investors navigating the complexities of the cryptocurrency market.