日内交易量预测的局部波动模型及其实证研究  

A local volatility model for intraday volume prediction and its empirical study

作　　者：季怡轩 徐紫怡 程雪 Yixuan Ji;Ziyi Xu;Xue Cheng

机构地区：[1]北京大学数学科学学院,北京100871

出　　处：《中国科学：数学》2021年第11期1743-1768,共26页Scientia Sinica：Mathematica

基　　金：国家自然科学基金(批准号:11971040和11871010)资助项目。

摘　　要：本文在辨析3种典型日内交易量预测模型—加和模型、乘积模型和分解优化模型—的理论差异并使用中国市场数据实证检验的基础上,提出一种新的日内交易量预测模型:局部波动模型.在交易量预测和成交量加权平均价(volume weighted average price,VWAP)策略层面,局部波动模型的稳定性均优于经典的基准方法—历史滚动均值.该模型运算速度快且可实现动态预测,预测精度方面表现良好,仅不及精度最高但速度最慢的乘积模型,且其稳健性优于乘积模型,介于乘积模型和分解优化模型之间.该模型在大盘风格数据上表现较好,且在处理频率较高的数据以及交易量波动较低的数据上具有优势.Based on clarifying the theoretical points of three intraday trading volume forecasting models,the additive model,the product model and the decomposition-optimization model,and conducting the empirical tests using intraday volume data from China stock market,a new intraday trading volume prediction model,the so-called local volatility model was proposed.In terms of transaction volume forecasting and VWAP(volume weighted average price)strategy,the local volatility model is more stable than the historical average estimation.Compared to its fast calculation speed and compatibleness with dynamic forecasting,it has satisfactory prediction accuracy.Its robustness lies between the product model and the decomposition-optimization model,which is better than the production model.Moreover,the model excels at large-cap data and the data with high frequency and low transaction volume fluctuation.

关 键 词：日内交易量预测 VWAP策略 分解组合方法 局部波动 

分 类 号：F224[经济管理—国民经济] F832.51