Cognitive Cyber Crimes in the Era of Artificial Intelligence

27Cognitive AI Systems for Real-Time Market Anomaly Detection and Asset Pricing in Financial Markets

Kuntal Dutta*, Kriti Bhaswar Singh, Devbrath and Neha Aggarwal

Department of Commerce and Financial Studies, Central University of Jharkhand, Jharkhand, India

Abstract

The rapid evolution of financial markets has intensified the complexity of asset pricing relationships, demanding intelligent real-time models for cross-sectional return prediction and risk premium estimation. This study presents a Cognitive artificial intelligence (AI) framework for fraud detection that integrates transformer-based temporal pattern learning with explainable reinforcement meta-learning to capture both sequential anomalies and adaptive market manipulations. The framework is evaluated on the NASDAQ ITCH Order Book Dataset, encompassing more than 450 million order events with labeled instances of spoofing, layering, and pump-and-dump schemes. Experimental results demonstrate superior performance, achieving an F1 score of 96.8%, Matthews correlation coefficient of 0.947, and detection latency of 21 milliseconds per event, significantly outperforming conventional long short-term memory and convolutional neural network–based baselines. Moreover, the system attained an area under precision–recall curve of 0.975 and reduced false positives by 18.6%, highlighting its robustness in high-frequency trading environments. The findings underscore the potential of cognitive AI systems to provide real-time, interpretable, ...