Analysis of Multi-Market Stock data to find Exchange Dominance

DOI: 10.5176/2251-1997_AF18.155

Authors: Daya Ram Budhathoki, Dipankar Dasgupta, Pankaj Jain, German Hernandez, Mike Nolen

Abstract:

We identify the stock exchanges that dominate liquidity demand and liquidity supply during stressful and normal periods. In particular, we collected financial data (stored in microseconds) from the Securities Information Processor (SIP) which consolidates trading data from 14 exchanges and dozens of alternative trading facilities including dark pools. We used big data framework such as Hadoop and Spark to analyze these (high frequency) financial data to find the anomalies in data pattern and exchange dominance. We applied statistical and data mining techniques for analysis and observed that stock exchanges which dominate on typical days are not equally active on the flash crash day. Other anomalies in multimarket data pattern between normal and the flash crash days are found via heat map analysis.

Keywords: SIP; liquidity; stock; Hadoop; spark; flash crash, data mining