Predicting the future with big data

Citi's Ron Papka talks us through how predictive algorithms are taking over the trading floor
Technology at a bank

"There is a huge appetite for 'Big Data' right now," explains Ron Papka, Global Head of Velocity Analytics at Citi. "That demand comes from various industries such as finance, retail, marketing, and internet search giants. Millions of data points are generated on a daily basis, and with Velocity, our capital markets intelligence client platform, our main aim is to successfully interpret data for the needs of our clients."

Improved trading opportunities

"Our aim is to collect and disseminate data centrally," explains Ron. "We transform data - synthesising it into something useful for our clients. If you can provide a sales person on the trading floor with visualisation tools that allow them to distill a whole universe of data into one page on their screen, they're more able to approach clients with meaningful and relevant trade opportunities."

However, the impact of big data isn't limited to the process of deciding what trades to make. It also plays a key role in helping clients manage risks. "Clients need to be aware of sensitivities to interest rate changes and other changes in the macro environment," Ron tells us. By using big data, clients are better equipped to not only identify potential future risks in advance but also to monitor those risks if and when they develop.

Developing with data

Developers find innovative ways to collate data and analyse it, in a way which can then be used to determine whether data from a particular source can be used to consistently predict market movements. The task of producing accurate predictive algorithms presents both opportunity and challenge for the financial services industry.

"Technological innovation has revolutionized the way we do business," says Ron. "However, people are still sceptical of predictive algorithms that traders often use in combination with historical data."

With Citi unlikely to put a predictive algorithm into use without evidence of its ability to forecast future market movements, the testing phase for an algorithm is crucial. "You have to put your scientist hat on and do as objective a set of experiments on the algorithm as possible," says Ron.

Citi uses millions of data points from a wide variety of different sources to generate predictive algorithms. "It's surprising and exciting when you find an approach that works," says Ron. "If you find a business use for it, even better."

Ron's role in Citi Velocity Analytics requires frequent interaction with different parts of the bank: "I get to work with traders, with quants (quantitative analysts), with risk colleagues, sales people, and compliance." Ron adds that the graduates who do best in this line of work are those keen to learn about finance, and to engage with the bank as a whole. "Developers would need advanced programming languages such as Java, C++ and C#; however, many graduate opportunities exist that don't require a programming background."

A future of data

Citi's developers are constantly looking for new and better algorithms for handling big data - and in Ron's view, "the firms who don't innovate won't survive in the future." That's unlikely to be a problem for Citi, who regularly give their developers free rein to develop anything they want. "At Citi, ingenuity is at the heart of everything we do, and developers often build solutions and products that end up being implemented by the business."

Ron is convinced the amount of data and the consequential demand on algorithms will increase in the future. But even though electronic trading systems and high speed trading are getting more efficient and intelligent, Ron explains machines haven't taken over the trading floor entirely yet.

"Although the majority of trades are done through algorithms now, traditional trading still takes place via phone. Clients still want the option of speaking to a human being!"

While the world might not be ready for a fully-automated trading system just yet, one thing is certain: that huge appetite for big data is showing no sign of being satisfied.

Continue learning below