Multiscale time series analysis of the Taiwan stock market: A wavelet-based approach

In this paper we propose the multiscale time series model (MSTM), which is a hybrid model combining wavelet analysis with time series models. The basic advantage of MSTM is that, while investigating complex adaptive systems, such as economic systems or financial markets, the signals (prices) obtained from the system are an aggregation of many heterogeneous agents' behaviors, and consequently a global model cannot efficiently handle such signals. By using the MSTM, the signals (prices) can be examined locally and different patterns can be revealed under different scales.