Deep Learning for Financial Time Series Prediction
LSTM and Transformer Models for Stock Market Forecasting
Penulis: Prof. Dr. Mada Rahman, S.Kom., M.T., Dr. Rama Pratama, S.Kom., M.T., Dr. Sarah Johnson, Ph.D.
Informasi Konferensi
Konferensi: International Conference on Computational Finance (ICCF)
Tanggal: 2024-07-30
Lokasi: Zurich, Switzerland
Abstrak
Abstract
This paper presents advanced deep learning architectures for financial time series prediction. We compare Long Short-Term Memory (LSTM) networks and Transformer models for stock market forecasting, incorporating multiple market indicators and sentiment analysis.
Model Architectures:
- Multi-layer LSTM with attention mechanisms
- Transformer-based temporal modeling
- Multi-modal input integration (price, volume, news)
- Ensemble prediction strategies
Empirical results demonstrate 68% directional accuracy and 15% improvement over traditional forecasting methods in volatile market conditions.
Kata Kunci
deep learning, financial time series, LSTM, transformers, stock market prediction, ACM