In this research, we want to use the strength of attempts to classify the abnormal volatility and want to use the classifications to observe the relationship between the abnormal k-lines and the trend of the Taiwan weighted stock price index. Moreover, we use these observations to construct two logistic regression models to predict the trend of the Taiwan weighted stock price index when the abnormal volatility occurred, and expect the results can be favorable to investors while investing in Taiwan stock market. We use 4,602 daily data from 1990 to 2006 and found that no matter the k-lines are red or black, the short-term predicions are better than the long-term ones.