Creating a Level Playing Field for Investors

Allison Bishop battles predatory traders with machine learning weaponry.

Allison Bishop
Allison Bishop (Photo by Timothy Lee Photographers)

For Computer Science Professor Allison Bishop, fintech is an enabling technology, one that potentially provides the key to expanding access to financial services. “Fintech lets us look at the ways we can use technology to maintain fair, efficient, and transparent marketplaces and resource allocation,” she said.

Bishop took a leave from SEAS in fall 2016 when she was invited to apply her statistical modeling skills at IEX, a young Manhattan based stock exchange that aims to level the playing field for investors. She also teaches an evening SEAS course, in cryptography, but explained that she values the opportunity to take a pause from academia: “As a theoretical person, and one who’s interested in societal good, I think it’s important to see how models play out in practice.”

The mission of IEX appealed to Bishop’s idealism. Its founders, profiled in Michael Lewis’s book Flash Boys, initially set out to understand the complex technology used by high-frequency traders to gain the fraction of a second that gave them the edge over the rest of the market—and then they were motivated to blunt that advantage. Their first innovation, in 2014, was a so-called speed bump, a proprietary feature that delays access to actively trading stock prices by a crucial 350 microseconds to help protect longterm investors like pension funds and other institutional investors from predatory strategies on that time scale.

Allison Bishop
A sample categorical breakdown of the roughly 10 million stock ticks in a day and the current performance of Bishop’s model at a two-millisecond timescale. (Courtesy of Allison Bishop)

IEX is currently developing an additional tool to arm its target investor base: predictive modeling, which is where Bishop is applying her specific expertise in statistical modeling and supervised machine learning. The goal of her research is to enhance a tool that can predict changes to the stock price, as sophisticated predatory traders do in order to strike before the price deteriorates. Her algorithms are aimed at refining the IEX Signal, a model that observes waves of pricing, essentially observing the dominos as the orders at the best price begin to fall, before the final tile is toppled.

Bishop likens the IEX Signal to a yellow traffic signal that warns when the price of a stock is about to change and protects investors from paying too much or from getting outwitted by traders who sneak in and grab the stock fractions of a second before the price changes. “I like working on models that anticipate adversarial behavior and can systematically defend against it,” Bishop said, “so this work is philosophically related to my Columbia research in cryptography.”

By Marilyn Harris