A Quantitative Investment System Based On Data Mining Is The Design And Implementation

Quantitative investment started in the late1970s and has made considerable progress in the past30years. According to data compiled by Bloomberg, in global market, the number of quantitative funds increased to1184in2008from21in1998. An increasing number of investors are moving into the quantitative investing field. And, naturally, quantitative investing is becoming one of the most important styles of investment.At present an investment behavior of financial products can be regarded as quantitative investing only if quantitative techniques are used in more than one stage in the process of investing. These quantitative techniques include modeling, pattern using, computing, programming and automating. Accordingly, there exist various terms about quantitative investment such as financial products quantitative selection, beat trading time quantitative analysis, algorithmic trading, automated arbitrage trading, and so on.Data mining is a better way to discover the patterns from the financial products transactions than traditional techniques such as sampling. The process of financial product investing mainly consists of investigating information from market and financial products, studying historical and current transactions of these products, predicating future transactions, making investment decisions and analyzing performance of transactions. Aiming at these stages, data mining techniques are employed in this paper in order to implement quantitative investing strategies. These techniques involve historical trading rules analysis, transaction sequential pattern database construction, trading price prediction and ordering trading strategies implement, etc.In this paper, a software framework of the quantitative investing system using data mining techniques is presented. This framework realizes main quantitation stages in the process of financial products investment systematically, which covering financial products transaction data integration and mining, transaction sequential pattern mining and pattern database management, trading price prediction, investing decision making, and automated ordering trading strategies implement, etc. Based on atomic patterns of transaction sequences, a Top-K atomic patterns mining algorithm, a pattern-based financial products clustering algorithm and an automated ordering trading algorithm are presented in this paper.