Automated Decision-Making System for Decentralized Stock Trading using Blockchain Smart Contracts and Quaternion Generative Adversarial Networks

Abstract

In the field of financial technology, the creation of an automated decision system for the decentralized execution of stock market operations is a noteworthy development. The challenges of the generic approach, such as transparency, security, and efficiency in stock trading, are effectively solved with the help of predicting analytics combined with smart contracts based on blockchain technology. In this research, a Blockchain Smart Contract Adoption with Quaternion Generative Adversarial Networks, (BSCA-QGAN), is presented for decentralized stock market prediction over Yahoo Finance data. The first pre-processing is data cleaning and normalization of data collected and then feature extraction is done by the Spike Driven Transformer (SDT) which is the integration of Transformers and Spiking Neural Networks. Also, the Quaternion Generative Adversarial Networks (QGANs) forecast the stock prices whereby time series with built-in quaternion math is modelled. The QGAN's hyperparameters are optimized using the White Shark Optimizer (WSO). An automated decision system for decentralized stock markets is integrated with blockchain smart contracts. The system achieved 99.72% prediction accuracy and minimal memory usage of 10MB.