Sleepless Nights and Public Data


FINRA Sponsors BitCamp Hackathon

When most people think about the weekend, they envision sleeping in, not pulling an all-nighter. But that’s exactly what FINRA employees did at the University of Maryland BitCamp Hackathon this past April. From April 10th-12th, over 1,400 participants travelled to Maryland from all over the country to test their skills. FINRA was one of many companies--including Microsoft, Capital One, and the NSA--that came with problems for students to solve.

FINRA brought three different sets of public data participants could work with: broker profiles from FINRA’s BrokerCheck® tool for investors and stock quote data. With an emphasis on FINRA’s open source projects, participants worked closely with developers like Bryan Robbins to understand the data and work out ways to use it in new and exciting ways.

Many teams saw FINRA’s challenge as an opportunity to use some familiar and freely-available tools in new ways. One team wanted to use Google’s Prediction API for the stock quote data. They used the API to predict future stock prices. Another team wanted an opportunity to use MongoDB as an engine to store and ultimately make sense of the data.


Real world experience and learning

Participants weren’t just learning new coding techniques, but also gaining experience working in groups and solving real world problems. “A lot of time computer science gets a bad rap of being an unsocial discipline,” Bryan said. “At hackathons, like the real world, everyone needs to add value to the team. Teams were able to build something to show off. That can only be a positive thing for a student.”

But FINRA employees weren’t passively watching from the sidelines, they were coaching and mentoring throughout the event. Debugging was a common and continuous challenge and FINRA’s developers spent a much of the event helping participants debugging their solutions. The time crunch wasn’t too stressful and participants were open to feedback and happy to be learning new ways to solve problems.

FINRA employees did more than help with debugging: they also provided impromptu lessons for the students. Mohammed Ibrahim, for instance, used his PhD in machine learning to give one team background on machine learning for a specific algorithm. Other teams needed context for the financial data FINRA provided. Girish Arunagiri, a Systems Analyst, was able to work closely with these teams and provide background to help their work move forward.

Being so close to these teams throughout the weekend became an issue at the end. The judges struggled to separate their close camaraderie with some teams from the end products of all teams. “The mentors got attached,” Bryan said. “We learned we need to separate mentors from judges next time around.”

Winning innovation

In the end, FINRA volunteers were able to choose winners for each category of data. The winners created some unique solutions. One group created Tickr, a stock prediction platform using the Google Prediction API, social media, and previous stock data to guess where stock prices for blue chip stocks would go the next day.

The second winning group developed InvestorWatch. Based on FINRA’s BrokerCheck® data of registered investment professionals, InvestorWatch allows users to compare certain attributes of brokers side by side. By using Angular, MongoDB, and Bluemix, InvestorWatch helps users choose financial advisors quickly and wisely.

The final winner was InvestBit. The InvestBit team saw an opportunity in the dizzying amount of information around stocks and investments. By using Matlab and Python, they were able to join data from external sites with FINRA’s market data to provide an enhanced view.

Lessons learned

Participants weren’t the only ones who learned from the BitCamp Hackathon. FINRA also learned valuable lessons from this hackathon. We want to refine our challenges and make the data clearer for future hackathons. These tweaks would ensure that participants spend less time trying to understand the data and more time creating unique solutions.

In addition, the Hackathon also became a bonding event for employees. “Because FINRA is so large, I hadn’t met everyone before this event,” Bryan admitted. “It was great getting to see so much creativity and passion for helping people.”

These lessons will be helpful this year. This is only the first of many hackathons FINRA plans to sponsor and attend. We’re already looking toward the fall and filling up our schedule, full of weekends of hacking, learning, and very little sleep.