A hybrid BPNN-weighted GREY-C3LSP prediction (BWGC) is performed very well for resolving the overshoot phenomenon significantly, but it definitely deteriorates the predictive accuracy once the volatility clustering occurs. Thus, we propose a new scheme to incorporate non-linear generalized autoregressive conditional heteroscedasticity (NGARCH) into BWGC prediction, which is optimally regularized by quantum minimization (QM), in such a way that the composite model, BWGC/NGARCH, can overcome the problem of volatility clustering substantially. In comparison with forecasting performance of several notable prediction methods, the proposed one can get the best predictive accuracy in two typical experiments.
|Number of pages||12|
|Journal||International Journal of Innovative Computing, Information and Control|
|Publication status||Published - 2007 Oct|
All Science Journal Classification (ASJC) codes
- Theoretical Computer Science
- Information Systems
- Computational Theory and Mathematics