Abstract: This paper proposes an improved quantum-behaved particle swarm optimization using the notion of species for establishing the ARCH model for stock return, and then forecastes subsequent trend. The experimental results show quantum-behaved particle swarm optimization is better at solving this problem than PSO and GA.

Key words: Auto-Regressive Conditional Heteroskedasticity(ARCH)model, Quantum-Behaved Particle Swarm Optimization(QPSO)algorithm, Particle Swarm Optimization(PSO)algorithm, heteroskedasticity, genetic algorithm

摘要： 介绍一种利用量子行为粒子群算法(QPSO)建立上证指数收益的 ARCH模型，利用不同的算法精确地估计模型中的参数，验证QPSO算法的优越性。利用得到的估计模型对指数收益进行预测，得到大致跟随指数实际走势的预测值。试验结果表明，QPSO算法比粒子群算法、遗传算法能更好地解决此类问题。

关键词: ARCH模型, QPSO算法, PSO算法, 异方差, 遗传算法

CLC Number: 

• TP391