基于连续混合正态分布的期货定价偏差研究  

Study on the Bias of Futures Pricing Based on Continuous Mixed Normal Distribution

作　　者：宋玉平 陈志兰 SONG Yu-ping;CHEN Zhi-lan(Business School,Shanghai Normal University,Shanghai 200234,China)

机构地区：[1]上海师范大学商学院

出　　处：《统计与信息论坛》2020年第1期23-29,共7页Journal of Statistics and Information

基　　金：国家自然科学基金青年项目“非平稳高频金融数据的大样本性质及应用”(11901397);全国统计科学研究重点项目“非平稳高频金融数据的统计推断及实证研究”(2018LZ05);上海师范大学第十期重点学科数量经济学(310-AC7031-19-004221)

摘　　要：挖掘期货理论价格和实际价格之间的关系有助于提高期货市场定价效率、发挥期货价格发现功能。基于持有成本定价模型计算期货定价偏差,利用连续混合正态分布模型对定价偏差的分布进行拟合,先采用基于牛顿迭代的极大似然估计法对未知参数进行估计,再进一步利用模拟退火算法对牛顿迭代的结果进行优化。结果发现,模拟退火算法可以有效提高估计精度,连续混合正态分布模型能够更好地拟合期货定价偏差分布。Mining the relationship between the theoretical price and the actual price of futures will help improve the pricing efficiency of futures market and give play to the discovery function of futures price.The deviation of futures pricing is calculated based on the hold-cost pricing model and the deviation is fitted by the continuous mixed normal distribution model.The unknown parameters are estimated by the maximum likelihood estimation method based on Newton iteration,and then the results of Newton iteration are optimized by the simulated annealing algorithm.The results show that the simulated annealing algorithm can effectively improve the estimation accuracy,and the continuous mixed normal distribution model can fit the deviation distribution of futures pricing better.

关 键 词：期货 持有成本模型 定价偏差 连续混合正态分布 极大似然估计量 牛顿迭代法 模拟退火算法 

分 类 号：O212[理学—概率论与数理统计] F222[理学—数学]