SBT-QL : a stock time series query language based on specialized binary tree representation

Abstract

Stock time series query is one of the fundamental components In technical analysis. Technical analysts would like to query either a whole time series or a segment of a time series in different resolutions for long-term or short-term investigations. Moreover, unnecessary data points should be filtered during the query process for easy analysis. In this paper, query approaches for both the whole stock time series and a subsequence of the stock time series based on the SB-Tree representation scheme are proposed. An approximate approach is proposed to improve the performance of the subsequence query process. Moreover, the local and global pruning methods are suggested to filter the unimportant data points. Besides studying these various query methods, an user-oriented query language, called SBT-QL, is proposed for: using easily, implementing efficiently and providing the ability to specify the query conditions to limit the retrieved data. The efficiency and effectiveness of the proposed approach are shown by the experiments.