Feb. 8, 2019 — Until recently, the thought of using supercomputers, the world’s most powerful computational machines, to study global financial markets was relatively unheard of. Today, Mao Ye, a Finance researcher from the University of Illinois at Urbana-Champaign is seeking to change that, and perhaps even inform regulatory decision-making in the future, all with the help of XSEDE’s computational and consulting expertise.
“I’m doing research on market microstructure, which requires massive amounts of data, and, as a consequence, analyzing even just two hours worth of data takes a very long time,” explained Ye.
In an industry that was once dominated by telephones and human beings on the floor of the New York Stock Exchange, for example, robust computation has totally revolutionized the movement of investments on a global scale. The rapid modernization of the digitization of financial markets has left regulators with mountains of data to comb through and analyze.
“Modern financial markets generate vast quantities of data,” explained Ye. “As the data environment has become increasingly ‘big’ and analyses increasingly computerized, the information that different market participants extract and use has grown more varied and diverse.”
To examine how modern computing could be used to analyze these huge collections of data, Ye required outside expertise to aid in his understanding of high-performance computing, and how to navigate the administrative channels to make this research a reality.
“With the unique challenge of data this size, some of my colleagues recommended that I contact TeraGrid [XSEDE’s predecessor, which ended in 2011]. Through the application process, I bumped into [XSEDE Principal Investigator] John Towns, who helped me find computing resources and helped me to analyze the data.”
Thanks to XSEDE, approaching the use of advanced computing systems to dive through these massive sets of data wasn’t nearly as daunting as it could be; it was exciting. Thanks to guidance, in part from XSEDE’s ECSS team, Ye’s novel research thoughts were transformed from an idea to one of the most innovative and unique present uses of supercomputing.
“The work that Mao is conducting is truly impressive and of high impact in financial economics,” said Towns. “His work is opening new research opportunities in the discipline by bringing the application advanced computing resources and services to this community to tackle these immense challenges that fundamentally affect the markets.”
This partnership between Ye and TeraGrid/XSEDE was not just a flash in the pan. Nearly a decade in the making, Ye is still working with XSEDE to utilize computational resources, relying on ECSS expertise to do this efficiently, and establishing a series of workshops meant to educate interested parties.
Over time, Towns and Ye haven’t been the only ones who saw value in this unique financial research. This September, Ye was awarded a $422,288 grant from the National Science Foundation for his project BIGDATA: IA: Collaborative Research: Understanding the Financial Market Ecosystem, to dive deeper into the intersection of Computer Science and Economics by interpreting and implementing big data techniques to understand financial data.
“This award will not only fund my own research, but it will also organize six workshops with the National Bureau of Economic Research (NBER),” said Ye. “These conferences will be related to big data and finance, with four of them in collaboration with XSEDE and the National Center for Supercomputing Applications (NCSA).”
“The goal is to jump-start a new community at the intersection of big data, finance, and public policy.”
While Ye’s research has created a significant starting point for using high-performance computing to analyze market data, there is still a substantial amount to be discovered about this unique and emerging relationship. In time, we may see research in this confluence of disciplines inform financial regulations
As finance and computers become more intertwined, the potential for growing pains is amplified, explains Ye. “Now, in financial markets, 70-80% of trade orders are coming from very fast computers. Trading is dominated by machines. There is an entire spectrum of novel and innovative trading, from people using bots and artificial intelligence.”
“But with this comes a blind spot,” cautions Ye, “You can make money off of using machines this way, but public policy often lags behind. If you want to analyze these machine-based traders, you also need to have their combined processing power, which is only accessible through high-performance machines.”
Through access to XSEDE-allocated computational resources (and the expertise to know how to utilize them), Ye and others are offered the rare opportunity to peer into an industry whose regulations are in their infancy, and in turn, may inform how to regulate machine-based trading in the future.
“What does the next generation of regulation look like? Should we update existing regulations or should we destroy the old regulations for humans? That’s the first order of importance to the fairness and transparency of financial markets,” concludes Ye. “XSEDE will give me the computing power to fulfill my dream of pursuing this line of research.”
BIGDATA: IA is supported by award no. 1838183 via the National Science Foundation, with supercomputer time and ECSS consultation provided through XSEDE Allocation grant SES120001.
Source: Boswell Hutson, NCSA