การวิเคราะห์อัตราผลตอบแทนตราสารทุนในมิติเวลาและความถี่ / ณรงค์ฤทธิ์ ทรัพย์ทิพยรัตนา = Analysis of stock returns in time and frequency dimensions / Narongrit Suptippayarattana
This thesis studied chaos in behavior of stock returns which traded in stock Exchange of Thailand by using daily sectoral indices and SET index from August 16th, 2001 to August 1st, 2003 and monthly indices from December 1992 to June 2003.The study used time-frequency analysis which can be divided into three approaches; discrete fourier transformation, discrete wavelet packet transformation by hear wavelet function and discrete wavelet packet transformation by daubechies4 wavelet function. The result of study shows that discrete fourier transformation cannot separate chaos from all index returns. On the contrary, discrete wavelet packet transformation by hear wavelet function and daubechies 4 wavelet function can separate chaos from all index returns and there are positive relations between chaos and index returns at 99 percent significant level. In addition, chaos can explain volatility of index returns about 20 to 85 percent, depending upon the type of index. Furthermore, chaos which separated by hear wavelet function can also forecast SET index return which relates positively to real SET index return at 99 percent significant level.