中国证券市场动态VWAP策略研究

发布时间：2018-05-31 18:13

本文选题：动态VWAP策略 + 日内交易量预测　；参考：《上海交通大学》2013年硕士论文

【摘要】：伴随着算法交易的兴起与发展，对于算法的研究逐渐成为金融学的研究热点，作为最基本的算法交易策略——VWAP（交易量加权平均价格）已然成为机构投资者使用最多的算法之一。通过按照市场成交量的分布将股票交易母单拆分为若干子单，VWAP策略能够降低冲击成本，从而减少执行成本，这种执行策略对于机构投资者而言是极其重要的。VWAP策略的核心是对日内交易量分布的预测，它的准确度对于策略的执行起着关键作用，因而如何更准确地预测日内交易量分布便成为提高VWAP算法执行效果的重要一步。目前市场中现有的VWAP算法以静态的历史VWAP方法为主，而且文献中研究样本也多运用了低频数据，本文将以动态的VWAP策略为研究目标，在历史VWAP的基础上利用市场上可得到的即时信息对交易量进行动态修正，以A股市场日内五分钟间隔的成交量高频数据为样本，进行动态建模，并将VWAP策略的交易量与实际市场交易量的跟踪误差作为衡量标准之一，得出动态VWAP策略以平均56.7%的概率战胜了历史VWAP策略；在策略的稳健性检验方面，将动态VWAP策略的表现对日内可能影响策略实施结果的若干股票特征进行多元统计回归发现，动态VWAP策略不会由于日内高频数据特征的差别而产生“区别对待”，它具有较广泛的适用性，尽管对于低换手率和高日均成交量的股票而言有更少的跟踪误差。在动态VWAP策略实现了对历史VWAP算法的优化之后，本文又将拆单之后的下单方式进行了详细的研究，提出了基于限价指令的最优变现策略。目前国内大部分对算法交易的研究是针对于市价的交易策略，，对限价的涉及较少，可能是由于指令簿的信息不全，而本文对成交概率进行指数函数的假设后，通过推导给出了限价交易最优变现报价策略的数值解，并进行了数值模拟，利用蒙特卡洛方法描绘出了平均交易曲线。该方法还考虑了无法及时变现的流动性风险问题，将其融入模型的限制条件中，并在适当的情况下将限价指令转换为市价指令降低执行风险。总之，本文研究了集VWAP拆单和限价指令下单为一体的多步骤变现策略，利用中国证券市场的高频数据，建模并实证，给出了一种适用于机构投资者优化的交易变现策略，这种策略的执行可以避免大单对市场造成的价格波动，对于降低市场冲击成本，提高整个证券市场的流动性具有重要意义。
[Abstract]:With the rise and development of algorithmic transactions, the research of algorithms has gradually become a hot research topic in finance. As the most basic algorithm trading strategy, VWAP (transaction volume weighted average price) has become one of the most widely used algorithms for institutional investors. According to the distribution of market turnover, the VWAP strategy can reduce the impact cost and reduce the execution cost by splitting the stock trading mother order into several sub-orders. The core of the .VWAP strategy, which is extremely important to institutional investors, is the prediction of intraday trading volume distribution, and its accuracy plays a key role in the execution of the strategy. Therefore, how to predict the intraday trading volume distribution more accurately is an important step to improve the performance of VWAP algorithm. At present, the existing VWAP algorithm in the market is based on the static historical VWAP method, and most of the research samples in the literature use low-frequency data. This paper will take the dynamic VWAP strategy as the research goal. On the basis of historical VWAP, using the instant information available in the market to dynamically modify the trading volume, taking the high-frequency data of trading volume between five minutes in a A-share market as a sample, the dynamic modeling is carried out. Taking the tracking error between the trading volume of VWAP strategy and the actual market trading volume as one of the measurement criteria, it is concluded that the dynamic VWAP strategy overcomes the historical VWAP strategy with an average probability of 56.7%, and the robustness of the strategy is tested. By applying the performance of dynamic VWAP strategy to a number of stock characteristics that may influence the implementation of the strategy, it is found that the dynamic VWAP strategy will not be "treated differently" because of the differences in high-frequency data features in the day. It has a wide range of applicability, although there are less tracking errors for stocks with low turnover and high daily turnover. After the dynamic VWAP strategy has realized the optimization of the historical VWAP algorithm, this paper makes a detailed study on the way of issuing orders after disassembly, and puts forward the optimal realization strategy based on the price limit order. At present, most of the research on algorithmic trading in China is aimed at the trading strategy of market price, but the limited price is less involved, which may be due to incomplete information in the instruction book. However, after the assumption of the exponential function of the transaction probability in this paper, In this paper, the numerical solution of the optimal realizable quotation strategy for the limited price trading is derived, and the numerical simulation is carried out, and the average transaction curve is described by using the Monte Carlo method. The method also considers the liquidity risk problem which can not be realized in time, and integrates it into the restrictive conditions of the model, and converts the price limit order into the market price order to reduce the execution risk under the appropriate circumstances. In a word, this paper studies a multi-step liquidity strategy which integrates VWAP disassembly and price limit orders. Using the high-frequency data of China's securities market, the paper presents a transaction liquidity strategy which is suitable for institutional investors to optimize, using the high-frequency data of China's securities market. The implementation of this strategy can avoid the price fluctuation caused by the large single market, and it is of great significance to reduce the impact cost of the market and improve the liquidity of the whole securities market.
【学位授予单位】：上海交通大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：F832.51;F224

【相似文献】