Price prediction of PFP NFT based on the sentiments of users in posts on social media

With the introduction of blockchain technology and the emergence of non-fungible tokens (NFTs), users can prove ownership of digital content by cryptographically tokenizing the content they create, and it becomes possible to trade digital content. As user-generated digital content is frequently traded online, many scholars have analyzed the factors of user transactions, but there is a limitation that they have not been able to analyze the direct relationship between the sentiments of users and price. Therefore, this study uses multi-layer perceptron so as to analyze the factors that affect the price of profile picture (PFP) NFTs by using not only collectable market indicators and technical indicators but also sentiment indicators. As a result, it was found that PFP NFTs are closely correlated with various indicators, and a model was developed to accurately predict the price fluctuations of PFP NFTs using these indicators. The empirical results demonstrate that the proposed MLP model achieved prediction accuracies of 81.49% for BAYC and 93.39% for Cryptopunks. Furthermore, stock indices were found to exert a positive influence on NFT prices, whereas increases in cryptocurrency values, interest rates, and discussion volume acted as negative determinants. By contrast, the interaction of positive and objective sentiment contributed positively to price formation.