PP-FSPML: Preemptive and Proactive Financial Signal Processing Framework Using Machine Learning

Abstract The current financial market is adopting advanced data-driven technologies where Financial Signal Processing (FSP) plays a crucial role not only in acquisition of enriched knowledge from financial data but also assists in securing the interest of consumer from the uncertain and unpredictable market volatility. Machine Learning (ML) has an integral contribution towards boosting the predictive performance by identifying complex patterns in financial distribution data; however, existing learning approaches doesn’t offer an assurance from market volatility. Hence, this problem is addressed by presenting a Preemptive and Proactive FSP using ML that emphasizes on addressing the problem related to prediction of Financial Asset Resources (FAR) that the existing solution has yet not come up with definitive solution towards securing its investors. The scheme performs statistical analysis of threats associated with input financial data considering market volatility, interest rates, and credit risk associated with financial sector of Asia-Pacific region. It is followed by obtaining statistical features for extreme values. Further, a contextual partitioning on time-series financial data is carried out considering directional trend, iterative trend, and residual. The outcome of the signal processing in proposed model is then subjected to Convolution Neural Network (CNN) and Bidirectional Long Short-Term Memory (BiLSTM) to generate the final predictive outcome. Analyzed on Asian Stock Market Dataset, Nikkei 225 Historical Data, and Hong Kong Stock Market Data, proposed scheme excels 40% increased accuracy and 35% reduced processing time in contrast to existing learning schemes.