The study aims to construct an effective early warning system (EWS) to predict the crisis triggered turbulence in Chinese bond market by integrating the volatility regime switching model, SWARCH, to improve the crisis classifying precision, and the stylized predictive model, Attention-BiLSTM of attention mechanism based deep neural networks, to resolve the predicting hysteresis. The model versatility and comparability are investigated and testified by applying multiple prominent EWS models to bonds with different credit rating levels. The hybrid EWS also specifies the leading factors relating to the bond credit rating, that will practically instruct governors and market participants to focus on either the national economy associated or the corporate finance concerned factors according to the bond varying credit risks to make more effective predictions.
This study constructs an integrated early warning system (EWS) that identifies and predicts stock market turbulence. Based on switching ARCH (SWARCH) filtering probabilities of the high volatility regime, the proposed EWS first classifies stock market crises according to an indicator function with thresholds dynamically selected by the two-peak method. A hybrid algorithm is then developed in the framework of a long short-term memory (LSTM) network to make daily predictions that alert turmoils. In the empirical evaluation based on ten-year Chinese stock data, the proposed EWS yields satisfying results with the test-set accuracy of $96.6\%$ and on average $2.4$ days of the forewarned period. The model's stability and practical value in real-time decision-making are also proven by the cross-validation and back-testing.
The study efforts to explore and extend the crisis predictability by synthetically reviewing and comparing a full mixture of early warning models into two constitutions: crisis identifications and predictive models. Given empirical results on Chinese currency and stock markets, three-strata findings are concluded as (i) the SWARCH model conditional on an elastic thresholding methodology can most accurately classify crisis observations and greatly contribute to boosting the predicting precision, (ii) stylized machine learning models are preferred given higher precision in predicting and greater benefit in practicing, (iii) leading factors sign the crisis in a diversified way for different types of markets and varied prediction periods.