Empirical Issues with Level III Limit Order Book Data

• Activity Number: 683
• Type: Contributed
• Date/Time: Thursday, August 13, 2015 : 10:30 AM to 12:20 PM
• Sponsor: Business and Economic Statistics Section
• Author(s): Raja Velu* and Krzysztof Herman
• Companies: Syracuse University and Syracuse University
• Keywords: High-frequency data ; Aggregation Issues ; Algorithmic Trading

Abstract

Recent regulatory and technological changes have resulted in an explosion of activity on electronic exchanges. This, combined with the ever increasing speed of the markets, resulted in a new breed of high frequency market data, known as Level III Data, which records (almost) every event occurring on an exchange. Such dataset allows for an unprecedented view of market dynamics but also gives rise to a number of new theoretical and computational challenges. Several questions need to be re-addressed, including such fundamental ones as whether all of the data needs to be analyzed or whether some kind of optimal sampling or aggregation approach can be used instead, or whether it is possible to find nearly-sufficient statistics that can describe the changes occurring in the market while guaranteeing a significant degree of data reduction. In our research we will outline how to make the data more usable. Moreover, we will investigate whether classical methods meant for data from low frequency perform well with such high frequency events and whether they are appropriate in the first place.