Zhuangzhuang (Marshall) Chang, BA/MA '17
Marshall participated in Brandeis University's National Science Foundation I-Corps program to launch his artificial intelligence startup.
Freshman year of college in the United States. What’s the first thing you think of? Frat parties, perhaps? Learning to do laundry, or maybe simply getting accustomed to life away from home? For Zhuangzhuang (Marshall) Chang, BA/MA '17, it meant learning everything there is about trading complex financial instruments. This passion has allowed him to, a few years later, start his own company and win his first two innovation competitions with his sights set on starting a hedge fund.
While he was at Brandeis University during his first year of undergrad, one could often find Marshall trading, combining his passions of finance with machine learning. This gave him a leg up with his three, yes three, majors: business, economics, and American studies. It was around that same time that popularity in artificial neural networks began to increase. Artificial neural networks are, according to Marshall, “the driving force behind everything intelligent today.”
This network inspired him to create a trading network that could actually learn. If Facebook uses neural networks to “learn” facial recognition and make it easier to tag your friends, why not use it to trade, a process that also requires decision making? In early 2017, at the same time that he was President and Portfolio Manager of the university’s Investment Club, Marshall founded A.I. Capital Management. It was a slow process to not only develop infrastructure, but also work on the model and write his own system. In total, it took Marshall around 8 months to produce a successful model, employing a controlled process that utilizes reinforcement learning (think Pavlov’s dogs), but instead of strengthening neural connections in the brain, it strengthens numbers, therefore making rewarded actions more likely to occur again.
In that time, he and his team also faced challenges with hardware: they needed computers strong enough to train the models efficiently, so they “basically had to become IT techs to get things running.” They initially used their own hardware, but after receiving funding from winning two spot in two Brandeis programs, were able to upgrade.
First, Marshall was selected for the National Science Foundation (NSF) I-Corps program at Brandeis University, which is one of only four nationally designated I-Corps sites in Massachusetts. Their I-Corps funding of $3,000 helped them validate their idea with potential customers, accelerating their progress. Next up, Marshall attended, and later won, Brandeis’s SparkTank, on February 4. Before pitching to judges, Marshall and his team were paired with Stuart from PitchDNA with whom they worked closely; Stuart was invited by Brandeis’ Office of Technology Licensing (OTL) to help all SparkTank teams with their pitch. He also met with Bozhanka Vitanova, I-Corps Instructor, a number of times, and received additional guidance from Fern Shamis about procurement and Weiwei Jiang, who helped organize peer connections. At the end of the competition, A.I. Capital Management won $10,000 to further their research.
Riding the wave of these two wins, the team is currently able to run at an improved speed and scale. Now, they are planning to test their A.I. in a live market with the goal of starting a hedge fund. Marshall’s next steps are “to contact investors, brokers, and lawyers to get things up and running.” The team also plans to attend more upcoming entrepreneurship competitions as a good way to get their name out there. He wants to apply to be participate in the national I-Corps program now that they’ve already proven themselves to be successful in the Brandeis programs.