Previous [ 1] [ 2] [ 3] [ 4] [ 5] [ 6] [ 7] [ 8] [ 9] [ 10] [ 11] [ 12] [ 13] [ 14] [ 15] [ 16] [ 17] [ 18] [ 19] [ 20] [ 21]


Journal of Information Science and Engineering, Vol. 30 No. 3, pp. 653-667 (May 2014)

Stock Trend Prediction by Using K-Means and AprioriAll Algorithm for Sequential Chart Pattern Mining*

1Department of Computer Science and Information Engineering
National Taiwan University of Science and Technology
Taipei, 106 Taiwan
E-mail: {wgb; M9915908; hmlee}
2Institute of Information Science
Academia Sinica
Taipei, 115 Taiwan

In this paper we present a model to predict the stock trend based on a combination of sequential chart pattern, K-means and AprioriAll algorithm. The stock price sequence is truncated to charts by sliding window, then the charts are clustered by K-means algorithm to form chart patterns. Therefore, the chart sequences are converted to chart pattern sequences, and frequent patterns in the sequences can be extracted by AprioriAll algorithm. The existence of frequent patterns implies that some specific market behaviors often appear accompanied, thus the corresponding trend can be predicted. Experiment results show that the proposed system can produce better index return with fewer trades. Its annualized return is also better than award winning mutual funds. Therefore, the proposed method makes profits on the real market, even in a long-term usage.

Keywords: Haar wavelet, K-means, AprioriAll, sequential chart pattern, stock trend prediction

Full Text () Retrieve PDF document (201405_07.pdf)

Received February 28, 2013; accepted June 15, 2013.
Communicated by Hung-Yu Kao, Tzung-Pei Hong, Takahira Yamaguchi, Yau-Hwang Kuo, and Vincent Shin- Mu Tseng.
* This research is supported in part by the National Science Council of Taiwan under grants number NSC 101-2218-E-011-008 and NSC 99-2221-E-011-075-MY3.