新兴市场股票交易行为建模理论与实证研究- - 基于整数值金融时间序列模型新视角

In stock trading activities, integer-valued time series are encountered frequently. For such data, the process follows a discrete distribution, such as the number of weekly rising or falling the limit, daily large transactions, weekend effect and so on. The continuous time series models don't extract their discrete characteristics and do not analyze the impact of these discrete events on the stock volume or price. However, theoretical and empirical studies of the integer-valued financial time series are narrower and not deep enough. It is necessary to conduct in-depth analysis and study of the trading behavior of integer-valued characteristics in the Chinese stock market. And this will make active and far-reaching sense for the setting and healthiness of the mechanisms for price formation in Chinese stock market, and improving the efficiency of the market...The project's research priorities and the key issues to be solved as follows:.(1) Unit root test of DF and ADF of integer-valued time series models; (2) Study the volatility models of integer-valued financial time series, establish the INGARCH models with the mean and variance separately, explore INARCH models with the covariates or the negative coefficient; (3) Forcasting research of INMA models; (4) Analyze integer-valued trading data in Chinese stock market, Find the suitable models for integer-valued financial time series to depict this type of variables, Empirical study trading mechanism and market structure of Chinese stock market, Build new financial risk indicators systems of unusual stock transactions including integer- valued variables.

股票交易活动中存在大量整数取值的时序数据，如周涨(跌)停次数、日大宗交易笔数、周日效应发生次数等。但是常用的连续时间序列模型并没有提取到它们的离散特征以及分析这些离散事件对股票量价的影响，而且目前整数值金融时间序列的理论与实证研究的范围较窄、深度也不够，因此有必要对中国股市的"整数值交易特征行为"进行深入研究。这对于建立和健全中国股票市场的价格决定机制，提高市场的效率, 都具有重要意义。.本项目的研究重点和拟解决的关键问题是：⑴整数值DF、ADF单位根检验；⑵研究整数值金融时间序列波动模型，建立均值与方差有别的INGARCH模型，探讨带协变量、系数可负的INARCH模型；⑶INMA模型预测研究；⑷分析中国股票市场整数取值的交易数据，寻找刻画这类变量的整数值金融时间序列模型，实证研究中国股市的交易机制与市场结构特征，构建含有整数值变量的股票异常交易新预警指标体系。

股票交易活动中存在大量整数取值的时序数据，如周涨(跌)停次数、日大宗交易笔数、周日效应发生次数等。但是常用的连续时间序列模型并没有提取到它们的离散特征以及分析这些离散事件对股票量价的影响，而且目前整数值金融时间序列的理论与实证研究的范围较窄、深度也不够，因此有必要对中国股市的“整数值交易特征行为”进行深入研究。这对于建立和健全中国股票市场的价格决定机制，具有重要意义。. 本项目的研究重点和拟解决的关键问题是：(1)具有自相关性的整数值非平稳过程性质；(2) 扩展整数值时间序列模型类型，建立了组合INMA模型、具有结构变化的INMA模型；(3)使用整数值时间序列模型做预测研究；(4) 分析中国股票市场整数取值的交易数据，寻找刻画这类变量的整数值金融时间序列模型.