This paper completes an empirical research with HMM and stock indexforecasting rfom the financial engineeirng angle. By using stock technicalanalysis indexes this paper try to find optimal observation vector group ofHMM and give a trading strategy on IF01based on the given HMM models.This paper wants to find a new approach for stock index future trading fromtheoretical and empiircal angles.HMM is composed of two nested stochastic processes. In theoreticalresearch part of this paper, I discuss the method for stock index forecasting andgive the method of daily ofrecast and muti-day weighted forecast. Atferestimate the amount of work I choose K-mean clusteirng to choosing the initialvalue.In empirical research part, this paper descirbes the defect of traditionalstock techniques analysis and proposes to use technical analysis indexes inHMM model input observation vector group. This paper use PC A to get adimension reduction on excessive indexes groups.At the end of this part, thispaper find a ideal HMM model with small MAPE and high winning ratio.In the part of trading strategy, this paper presents a trading strategy for IF01based on HMM giving daily predictive signals and calculates the simulatedtrading income of candidate HMM models.At last,by compairson of twodifferent models with high cumulative, I give the feasible conditions on stockindex futures trading using HMM. |