Scholarlink

<p>This research proposes a novel approach to forecasting top cryptocurrencies’ prices—Bitcoin, Ethereum, and Binance Coin—by combining copula theory with a Temporal Fusion Network (TFN) into a Hybrid Copula-Temporal Fusion Network (HCTFN). The model can identify intricate relationships among cryptocurrencies and prominent exogenous variables like CO₂ emissions and oil prices, thereby improving the precision of the forecast. Utilizing the advantages of static and dynamic variables, the HCTFN efficiently solves complex temporal patterns and volatility in the market. The outcomes are uniform in their demonstration of the model performing well in cryptocurrencies with low price volatility, like Binance Coin, while Bitcoin volatility is harder to predict. Data preprocessing as well as model optimization also become major areas to enhance dependability in prediction from the study. In spite of computational requirements and possible limitations of data, HCTFN in itself is an enhancement in predicting cryptocurrencies. The model could be advanced further by including dynamic copula models along with other external variables and extended to other digital assets to promote market insights.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Precision Cryptocurrency Forecasting: A Hybrid Copula-Temporal Fusion Approach with Environmental and Economic Insights

  • 期刊论文

  • 发表日期：

    2025年9月20日

查看全文
  • Imran Ali Khan,
  • Sami Ur Rahman

摘要

This research proposes a novel approach to forecasting top cryptocurrencies’ prices—Bitcoin, Ethereum, and Binance Coin—by combining copula theory with a Temporal Fusion Network (TFN) into a Hybrid Copula-Temporal Fusion Network (HCTFN). The model can identify intricate relationships among cryptocurrencies and prominent exogenous variables like CO₂ emissions and oil prices, thereby improving the precision of the forecast. Utilizing the advantages of static and dynamic variables, the HCTFN efficiently solves complex temporal patterns and volatility in the market. The outcomes are uniform in their demonstration of the model performing well in cryptocurrencies with low price volatility, like Binance Coin, while Bitcoin volatility is harder to predict. Data preprocessing as well as model optimization also become major areas to enhance dependability in prediction from the study. In spite of computational requirements and possible limitations of data, HCTFN in itself is an enhancement in predicting cryptocurrencies. The model could be advanced further by including dynamic copula models along with other external variables and extended to other digital assets to promote market insights.