Researchers at the University of Michigan and the Georgia Institute of Technology will develop innovative approaches to detecting and deterring the computerized manipulation of financial markets under a $1M grant from the National Science Foundations’s Big Data program.
Market manipulation, also known as price manipulation, is the deliberate use of misleading information with the intent of deceiving investors about the supply or demand of a security for financial gain or other advantage. Increasingly, manipulators attack market integrity through complex computer-controlled attacks.
Tactics used by market manipulators include “spoofing,” in which traders place orders with the intent to cancel before execution, “banging the close,” where traders impact closing prices to affect derivative holdings, and propagating false information through social media or other channels to move markets. The potency of these tactics are increased dramatically through the use of increasingly sophisticated attack algorithms and deployment platforms.
“These attacks are difficult to detect and defend against because the underlying actions have legitimate purposes as well as nefarious ones,” said project director and principal investigator Michael Wellman, the Lynn A. Conway Collegiate Professor of Computer Science and Engineering at Michigan.
At the same time, “It is important that we develop ways to detect these attacks, and we’ll be applying machine learning to that task,” said principal investigator Tucker Balch of Georgia Tech. Both Wellman and Balch are computer scientists who specialize in artificial intelligence and algorithmic trading.
The research team also includes principal investigators Uday Rajan, Professor and Head of Finance Area, Ross School of Business at Michigan and Michael Barr, Dean of the Ford School of Public Policy and Professor of Law at Michigan, who bring expertise and experience in market structure and financial regulation to the project’s effort to develop practical measures to combat electronic market manipulators.
Their approach will monitor trading current activity data to develop a model that is calibrated to normal trading activity and will discover and extract “signatures” of manipulation. This will be paired with a system of model-based techniques for characterizing manipulation strategies. Together, this data will generate a series of successful manipulation strategies, which can be introduced into modeled trading environments to reveal signatures of spoofing activity, which the researchers will in turn use to construct surveillance and audit algorithms.
Methods produced under this project, in conjunction with guidance on market design and regulation policy, are expected to contribute to reducing the threat from increasingly capable market manipulators.
The project is entitled “Detecting Financial Market Manipulation: An Integrated Data- and Model-Driven Approach.” The project is funded for $1M over three years.