Related Experiment Video
Updated: Sep 7, 2025
06:37
Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention
Published on: December 15, 2023
4.1K
The Empirical Analysis of Bitcoin Price Prediction Based on Deep Learning Integration Method
Shengao Zhang1, Mengze Li2, Chunxiao Yan3
1School of Public Administration, China University of Geosciences, Wuhan 430074, China.
Computational Intelligence and Neuroscience
|June 20, 2022
Summary
This study introduces a deep learning integration method, stacking denoising autoencoders with bootstrap aggregation (SDAE-B), for accurate bitcoin price prediction. The SDAE-B model demonstrates superior performance over traditional methods in forecasting volatile cryptocurrency markets.
Area of Science:
- Computational Finance
- Artificial Intelligence
- Cryptocurrency Markets
Background:
- Bitcoin's increasing recognition is coupled with significant price volatility and market risks, making accurate price prediction challenging.
- Traditional prediction models often struggle to capture the complex, nonlinear dynamics inherent in cryptocurrency markets.
Purpose of the Study:
- To develop and evaluate a novel deep learning integration method, stacking denoising autoencoders with bootstrap aggregation (SDAE-B), for enhanced bitcoin price prediction.
- To compare the predictive accuracy of the SDAE-B model against traditional machine learning methods like LSSVM and BP.
Main Methods:
- The study employs stacking denoising autoencoders (SDAE) to model intricate relationships between bitcoin prices and influencing factors.
- Bootstrap aggregation (Bagging) is utilized to generate diverse training datasets for multiple SDAE models, enhancing robustness.
- Exogenous variables including block size, hash rate, mining difficulty, transaction volume, market capitalization, search engine trends, gold price, dollar index, and major events were incorporated.
Main Results:
- The SDAE-B model achieved a Mean Absolute Percentage Error (MAPE) of 0.016, Root Mean Square Error (RMSE) of 131.643, and Directional Accuracy (DA) of 0.817.
- The SDAE-B method significantly outperformed Least Squares Support Vector Machine (LSSVM) and Backpropagation (BP) neural networks in bitcoin price prediction.
- The model effectively captured the inherent randomness and nonlinear characteristics of bitcoin price movements.
Conclusions:
- The proposed SDAE-B deep learning integration method offers a highly accurate and robust approach for predicting bitcoin prices.
- This advanced technique provides a valuable tool for navigating the risks associated with volatile cryptocurrency markets.
- The findings highlight the potential of sophisticated deep learning architectures in financial forecasting.
