Financial crisis forecasting via coupled market state analysis

Financial crisis forecasting has been a long-standing challenge that often involves couplings between indicators of multiple markets. Such couplings include implicit relations that might not be effectively detected from raw market observations. However, most methods for crisis forecasting rely directly on market observations and might not detect the hidden interactions between markets. To this end, the authors explore coupled market state analysis (CMSA), assuming that the observations of markets are governed by a collection of intra- and intercoupled hidden market states. Accordingly, they built a forecaster based on these coupled market states instead of observations.