Predicting trend in futures prices time series using a new association rules algorithm

Predicting trend in futures prices time series has always been one of the hottest topics in research, as well as a challenging problem due to time series' volatility. In this paper we propose a new association rules mining algorithm, TB-SCM, which is based on Transaction Bool Matrix and Support Count Matrix, for predicting trend in futures prices to meet this challenge. Here, our algorithm mainly considers the preprocessing and analyzing of data. We develop a novel time series association rules mining prototype system based on the TB-SCM algorithm and C++ STL technology, and investigate the efficiency of the system using West Texas Intermediate (WTI) crude oil futures prices time series listed on the Energy Information Administration (EIA) website, as well as Shanghai Futures Exchange (SHFE) fuel oil futures' closing prices time series on hexun website. The empirical study shows that TB-SCM algorithm out-perform classical Apriori algorithm available in Weka data mining software which is developed by Waikato University in terms of generating time series association rules without redundancy.