Scholarlink

This study seeks to evaluate the factors shaping cryptocurrency market data forecasting by utilizing 2451 daily observations of Bitcoin (BTC) closing prices from November 2017 to July 2024. Employing a hybrid financial technology (Fintech) approach, the model combines a nonlinear spectral method involving the Maximum Overlapping Discrete Wavelet Transform (MODWT) with a Genetic Fuzzy System inspired by Thrift’s technique (GFS.Thrift). Input variables, such as the logarithm volume of Bitcoin (LVL) and logarithm Ethereum closing price (LET), were chosen based on factors like correlation, tolerance, Variance Inflation Factor (VIF), and multiple regressions, sourced from the cryptocurrency market. The adaptive GFS.THRIFT model underwent training on 80% of the dataset, reserving the remaining 20% for performance testing. A comparative analysis between the proposed model (MODWT-GFS.Thrift) against the traditional model, GFS.Thrift, demonstrated that the performance of MODWT-GFS.Thrift was found to be less effective than that of the traditional model.

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

Evaluating Factors Shaping Bitcoin Closing Prices with Fintech Strategies

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Jamil J. Jaber,
  • Younis Ahmed Ghulam,
  • Anwar Al-Gasaymeh,
  • Rania Al Omari,
  • Nawaf N. Hamadneh

摘要

This study seeks to evaluate the factors shaping cryptocurrency market data forecasting by utilizing 2451 daily observations of Bitcoin (BTC) closing prices from November 2017 to July 2024. Employing a hybrid financial technology (Fintech) approach, the model combines a nonlinear spectral method involving the Maximum Overlapping Discrete Wavelet Transform (MODWT) with a Genetic Fuzzy System inspired by Thrift’s technique (GFS.Thrift). Input variables, such as the logarithm volume of Bitcoin (LVL) and logarithm Ethereum closing price (LET), were chosen based on factors like correlation, tolerance, Variance Inflation Factor (VIF), and multiple regressions, sourced from the cryptocurrency market. The adaptive GFS.THRIFT model underwent training on 80% of the dataset, reserving the remaining 20% for performance testing. A comparative analysis between the proposed model (MODWT-GFS.Thrift) against the traditional model, GFS.Thrift, demonstrated that the performance of MODWT-GFS.Thrift was found to be less effective than that of the traditional model.