Deep learning,probability of order execution and trading costs
To deeply explore the impact mechanism of trading costs under microstructure of high-frequency market,we reconstruct the time-varying limit order book using the Shenzhen Stock Exchange tick-by-tick high-frequency data.Based on the survival analysis model and deep learning method,we comprehensively consider the effects of order characteristics and character-istics of order book on the execution probability of limit order,and calculate the trading costs of limit order under different market conditions.Then,we expand on the Maglaras'cost impact model by including the order flow information measurement indicator and deeply discuss the impact of market microstructure variables on trading costs.The empirical results show that,at the micro level,the impact of transaction information on trading costs is limited.Specifically,price fluctuations have a significant effect on trading costs,but its direction varies depending on different stocks.In contrast,the impact of the unexecuted order flow on trading costs is consis-tent and contains more valid information,although the degree of effectiveness depends on the measurement perspectives and order characteristics.In addition,the submission of large orders is a significant driver of trading costs,serving as a signal source for adjusting trading strategies,and it is also a key area of regulatory agencies'attention.
market micro-structurelimit order bookexecution probabilitydeep learningtrading costs
万方数据
维普
