Current Proceedings on Technology
Yazarlar: Jui- Chung Hung
Konular:-
Anahtar Kelimeler:Adaptive neuron fuzzy inference systems (ANFIS),Fuzzy system,Generalized autoregressive conditional heteroskedasticity (GARCH) model,Volatility,Particle swarm optimization (PSO)
Özet: In general, stock market volatility consists of characteristics that are clustering and asymmetric with respect to both good news and bad news. Therefore, we have proposed a two-stage method to address this issue. The procedure that we used is as follows: first, we use fuzzy switch to analyze asymmetric of stock market volatility and adopt the threshold values to discriminate between two regimes of positive and negative fluctuation. Second, we use the two regimes of first stage to set up adaptive neuron fuzzy inference systems (ANFIS) - Generalized autoregressive conditional heteroskedasticity (GARCH) model, the purpose of which was to reduce the clustering complexity. The fuzzy switch ANFIS-GARCH model joins the parameters of membership functions and GARCH models make this problem highly nonlinear and complicated. This paper presents an iterative algorithm based on particle swarm optimization (PSO) to estimate parameters of the membership functions. The effectiveness of the approach is demonstrated on stock market data from the Taiwan Stock Exchange weighted stock index (Taiwan), DAX 30 (Germany), and Nikkei 225 index (Japan). From the simulation results we have determined that both the estimation of in-sample and forecasting of out-of-sample volatility performance are significantly improved when the model considers both the clustering and asymmetric effect.