Multi-Layer Coupled Hidden Markov Model for Cross-Market Behavior Analysis and Trend Forecasting

摘要: The frequent global financial crisis indicates the increasing importance and challenge to analyze forecast future trends of stock market for investors trading agents. Especially with globalization world economy integration international markets, complex relationships between markets from different countries should be considered in forecasting trends, involving multi-layered, interactive, evolutionary, heterogeneous variables couplings variable sets countries. A variety methods have been proposed implemented but there is very limited work reported on predicting a market’s movement based analyzing hidden coupling various This involves analysis hierarchical coupled behaviors their across multiple nonlinear dynamics. To address this critical issue, paper proposes new approach Multi-layer Coupled Hidden Markov Model (MCHMM) Hierarchical Cross-market Behavior Analysis (HCBA), namely exploring country (Layer-1 coupling) (Layer-2 coupling), movements. Toward capturing behaviors, Multi-layered built infer movements target by its price return probabilities. experimental results 11 years data two types (stock currency market) 13 show that our outperforms other four benchmarks technical business perspectives.