Application of improving fuzzy time series to predict TAIFEX

Taiwan Futures Exchange (TAIFEX) is the world´s 23rd futures exchange to trade SSFs, which is also the 21st financial product offered by the exchange. In this paper, the TAIFEX is predicted based on improving fuzzy time series model. Nature-ratio lengths of intervals technique is employed to partition the universal of discourse of linguistic variable and an improving high-order heuristic function is also used to select the defuzzified value of fuzzified prediction variable to improve prediction accuracy. The experiment result shows that our proposing model has higher prediction accuracy than those of some conventional prediction methods.