Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Comparative Study of Bitcoin Price Prediction

Ali Mohammadjafari

TL;DR

The paper evaluates Bitcoin price forecasting using two neural network architectures, LSTM and GRU, with 5-fold cross-validation and L2 regularization. It utilizes Bitcoin historical data from Yahoo Finance (2015–present) with multiple price-related features to train and compare the models. The results show that GRU achieves higher predictive accuracy, with a mean squared error of 4.67 versus 6.25 for LSTM, and is about 30% faster, highlighting GRU's suitability for financial time series. The work underscores the value of regularization in reducing overfitting and provides guidance for model selection in volatile markets.

Abstract

Prediction of stock prices has been a crucial and challenging task, especially in the case of highly volatile digital currencies such as Bitcoin. This research examineS the potential of using neural network models, namely LSTMs and GRUs, to forecast Bitcoin's price movements. We employ five-fold cross-validation to enhance generalization and utilize L2 regularization to reduce overfitting and noise. Our study demonstrates that the GRUs models offer better accuracy than LSTMs model for predicting Bitcoin's price. Specifically, the GRU model has an MSE of 4.67, while the LSTM model has an MSE of 6.25 when compared to the actual prices in the test set data. This finding indicates that GRU models are better equipped to process sequential data with long-term dependencies, a characteristic of financial time series data such as Bitcoin prices. In summary, our results provide valuable insights into the potential of neural network models for accurate Bitcoin price prediction and emphasize the importance of employing appropriate regularization techniques to enhance model performance.

Comparative Study of Bitcoin Price Prediction

TL;DR

The paper evaluates Bitcoin price forecasting using two neural network architectures, LSTM and GRU, with 5-fold cross-validation and L2 regularization. It utilizes Bitcoin historical data from Yahoo Finance (2015–present) with multiple price-related features to train and compare the models. The results show that GRU achieves higher predictive accuracy, with a mean squared error of 4.67 versus 6.25 for LSTM, and is about 30% faster, highlighting GRU's suitability for financial time series. The work underscores the value of regularization in reducing overfitting and provides guidance for model selection in volatile markets.

Abstract

Prediction of stock prices has been a crucial and challenging task, especially in the case of highly volatile digital currencies such as Bitcoin. This research examineS the potential of using neural network models, namely LSTMs and GRUs, to forecast Bitcoin's price movements. We employ five-fold cross-validation to enhance generalization and utilize L2 regularization to reduce overfitting and noise. Our study demonstrates that the GRUs models offer better accuracy than LSTMs model for predicting Bitcoin's price. Specifically, the GRU model has an MSE of 4.67, while the LSTM model has an MSE of 6.25 when compared to the actual prices in the test set data. This finding indicates that GRU models are better equipped to process sequential data with long-term dependencies, a characteristic of financial time series data such as Bitcoin prices. In summary, our results provide valuable insights into the potential of neural network models for accurate Bitcoin price prediction and emphasize the importance of employing appropriate regularization techniques to enhance model performance.
Paper Structure (19 sections, 8 figures, 1 table)

This paper contains 19 sections, 8 figures, 1 table.

Table of Contents

  1. Introduction
  2. Motivation and Overview
  3. Goal and Contributions
  4. Related Works
  5. Price prediction
  6. Bitcoin Price prediction
  7. Methodology
  8. Problem Statement
  9. Initial Experiment
  10. Data Processing
  11. Data Collection and Preprocessing
  12. K-Fold Cross Validation
  13. Model Training and Validation
  14. LSTM
  15. GRU
  16. ...and 4 more sections

Figures (8)

  • Figure 1: Bitcoin daily prices on Bitstamp (USD) from 29 November 2011 to 27 March 2023. The upper line shows log prices, whereas the lower line shows plain prices. Some recurring patterns seem to exist when considering the log value of the Bitcoin price.
  • Figure 2: Display of the Data collected
  • Figure 3: schematic of 5-fold cross validation
  • Figure 4: The structure of a long short-term memory (LSTM) cell.
  • Figure 5: The structure of GRU cell.
  • ...and 3 more figures