In 2013, a high-frequency trader named Michael Coscia was arrested in New Jersey for an activity called "spoofing"--essentially manipulating the market by flooding trading systems with future orders he had no intention of completing. He was fined $6 million--with the possibility of jail time. It was the first such prosecution under a new set of financial regulations from the 2010 banking reform law called the Dodd-Frank Act.
That was an aha! moment for David Widerhorn, 28, and it became his reason for founding Neurensic. Based in Chicago, his company uses artificial intelligence to spot anomalies in high-frequency futures trading.
At the time, Widerhorn was running his own successful consulting firm that helped high-frequency trading desks in 30 countries with their quantitative research and programming to facilitate trades. But here, staring him in the face, was an opportunity to use his expertise in machines and artificial intelligence, garnered at the Massachusetts Institute of Technology, in a completely new way.
"The writing was on the wall," Widerhorn says. "We were putting all of this intelligence and science into sophisticated trading strategies, but what about a backbone for risk management and surveillance and clearing?"
The company's core product uses artificial intelligence to understand and form judgments about the massive amounts of data spit out by trading desks, in search of anomalies like Coscia's spoofing activity, as well as other market manipulating tactics known as layering and stuffing, and violations such as collusion through messaging applications.
Today, Neurensic has 45 employees and expects revenue between $20 million and $31 million in 2016. That's not bad for a company that launched officially in 2015, following a year-long pilot. In addition to Widerhorn, co-founders include Zach Watts, the chief innovation officer, Tim Geannopulos, chief operating officer, and Paul Giedraitis and Jay Vohra, both principals.
If "big data" was the buzz word for startups five years ago, today it's "artificial intelligence." The latter picks up where the former leaves off, using advanced overlapping neural networks to make sense of troves of data.
In essence, Neurensic uses computer networks to recognize patterns in human behavior in the massive amounts of data churned out each day by trading desks. As opposed to a static application of rules, artificial intelligence uses spontaneously changing algorithms to adapt and learn about traders, allowing the computer networks to make judgments: What may look risky for one trader isn't for another.
"The algorithms presumably will know the difference between a trader who never shorts, and one who always shorts," says Danielle Tierney, a senior analyst at financial services research firm Aite Group.
But artificial intelligence, while talked about for decades, is still in its infancy, according to numerous investment experts. That's why companies like Neurensic--which are using it to solve very specific problems, like compliance in financial services--probably have a better chance of success in the short run than generalists, says Cack Wilhelm, a partner at Scale Venture Partners. The Silicon Valley venture capital firm has invested about $35 million in A.I. startups over the past few years, including three in the past year alone.
Neursensic is somewhat in the vanguard, in a nascent market for compliance intelligence that is currently worth about $450 million, says Tierney. That's expected to grow to $1 billion in the next 10 years. It's competing against very large companies that offer similar services, including Nice Actimize, Oracle, and even Nasdaq. And there are startup competitors as well, including B-Next and MilleniumIT.
In 2014, in a pilot, Neurensic signed up its first customer, the futures commission merchant (FCM) Advantage Futures, Widerhorn says. Today, it has signed deals with eight more customers, and it has more than two dozen in the pipeline. Although Widerhorn would not name the new customers, he says the roster includes global financial institutions, as well as other large FCMs, and some government regulating entities.
Meanwhile, Neurensic has assembled a high-powered board of advisers that includes Leo Melamed, chairman emeritus of the CME Group, and Melanie Rubio, former principal at Wolverine Trading and a senior data scientist at Braintree. That's helped Neurensic take in $6 million in funding from a dozen individual investors in angel and seed rounds. In the next year, Widerhorn says Neurensic hopes to secure as much as $30 million in venture capital funds that will allow the company to double its employee count as it opens offices in London and Hong Kong, to tackle markets in Europe and Asia.
Widerhorn and his crew are also branching out by creating a new product to help financial institutions maximize profits in operations. "Our technology is there to stabilize the market and to find people who are manipulating and taking advantage of the market," Widerhorn says. "It provides clarity in times of uncertainty."