Time Series Chaotic Neural Oscillatory Networks for Financial Prediction

Abstract

Financial prediction, ranging from stocks, commodities to forex forecast poses real challenges to researchers and quantitative analysts (a.k.a. quants) due to its highly chaotic and unpredictable nature. With the blooming of cryptocurrency trading which provides 24 × 7 online currency trading, the financial markets, especially the worldwide currency markets are more chaotic and fluctuating than before. An effective and reliable financial prediction system is profoundly required. With the integration of quantum price levels (QPL) evaluated by quantum finance technology, this chapter devises a Time Series Chaotic Neural Oscillatory Network (TSCNON) for worldwide financial market prediction to effectively resolve the system over-training and deadlock problems imposed by traditional recurrent neural networks using classical sigmoid-based activation functions. In terms of system implementation, TSCNON coalesces into 2048 trading-day time series financial data and 39 major financial signals as input signals for the real-time prediction of 129 worldwide financial products which includes: 9 major cryptocurrencies, 84 forex, 19 major commodities, and 17 worldwide financial indices. In terms of system performance, past 500-trading day of average system performance attained less than 1% forecast percentage errors.