Comparative Analysis of Hourly Data Approach and Strategies of Short-Term Investment in Cryptocurrency Market

The most dynamic cryptocurrency market has been an ideal site for short-term predictions utilizing deep learning (FL) and machine learning (ML). The study presents how predictive performance is compared when based on different hourly data approaches for short-term capital investments into cryptocurrencies. In addition to the algorithms used in Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), along with Convolutional Neural Networks (CNN), this study includes conventional machine learning techniques like Random Forest (RF), Support Vector Machines (SVM), as well as Gradient Boosting (GBM). Using hourly price data from 20182024 from Bitcoin, Ethereum, and Ripple, we form trading signals and conduct simulations of investment strategies. The outcomes indicate that DL models outperform classical algorithms in capturing time dependencies of hourly movements with LSTM achieving the peak in predicted accuracy and riskadjusted returns. In addition, results underscored the importance of high-frequency data and best model selection in creating profitable short-term cryptocurrency investment strategies.

Ҳали таржима қилинмаган