Cryptocurrency Price Prediction Using Temporal Decay Attention LSTM: A Novel Approach for Time-Series Forecasting

Kavadi, Krishnaseshu (2025) Cryptocurrency Price Prediction Using Temporal Decay Attention LSTM: A Novel Approach for Time-Series Forecasting. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (907kB) | Preview
Preview	PDF (Configuration Manual) Download (870kB) | Preview

Abstract

Cryptocurrency markets are known for their extreme volatility and sensitivity to both technical and sentiment-driven signals. This study investigates the effectiveness of integrating sentiment features with deep learning models—particularly attention-based architectures, for improved cryptocurrency price forecasting. A comprehensive comparison was conducted between traditional machine learning models (AdaBoost Regressor, Decision Tree, LightGBM) and advanced deep learning models (LSTM, Standard Attention LSTM, Temporal Decay Attention LSTM), using both sentiment-aware and sentiment-agnostic data. Experimental results reveal that deep learning models significantly outperform traditional approaches, with the Temporal Decay Attention LSTM exhibiting the highest predictive accuracy. The model's ability to prioritize recent time steps using a temporal decay attention mechanism enables more responsive and accurate forecasting in fast-changing market conditions. Additionally, the inclusion of sentiment data further enhanced performance in deep architectures, confirming the relevance of behavioral indicators in this domain. These findings suggest that combining temporal attention with sentiment analytics offers a powerful framework for real-time financial prediction, particularly in volatile environments like cryptocurrency markets.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Siddig, Abubakr UNSPECIFIED
Subjects:	H Social Sciences > HG Finance > Money > Digital currency > Cryptocurrencies Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Ciara O'Brien
Date Deposited:	25 Aug 2026 15:06
Last Modified:	25 Aug 2026 15:06
URI:	https://norma.ncirl.ie/id/eprint/9635

Actions (login required)

View Item