ETHEREUM'UN ERC-20 TOKENLARI ÜZERİNDEKİ ETKİSİ: LSTM VE CNN MODELLERİYLE KARŞILAŞTIRMALI BİR ANALİZ

Abstract

Ethereum, developed by Vitalik Buterin in 2013, has significantly advanced blockchain technology through smart contracts and ERC-20 token standards. This study examines the impact of Ethereum on ERC-20 tokens using Long Short-Term Memory (LSTM) and Convolutional Neural Networks (CNN) models. For this purpose, LSTM and CNN models were trained using Ethereum data and then employed to predict ERC-20 token prices. According to the study's results, the LSTM model achieved high accuracy rates for LINK, MATIC, and UNI tokens but performed poorly in predicting RNDR token prices. The CNN model provided the highest accuracy for LINK tokens and yielded successful results in predicting RNDR token prices. However, the CNN model showed lower performance for MATIC and UNI tokens than the LSTM model. These findings indicate that both LSTM and CNNmodels significantly impact the prediction of Ethereum's ERC-20 token price dynamics. The variability in model performances across tokens highlights the influence of market dynamics and liquidity levels. In light of these differences, the study emphasizes the importance of selecting the model based on the token's characteristics and market conditions.