Two Sigma, a $24bn hedge fund in New York’s Soho, has robots roaming the halls. Seriously. But they aren’t doing much. The Wall Street Journal says they’re restricted to tasks such as playing shuffleboard or air hockey.
By definition mechanical devices, robots move far too slowly for the world of trading, which has been built on analytics and trade decisions whose operations are measured in microseconds. It turns out that David Siegel, a co-founder of Two Sigma with a computer science Ph.D. from MIT, simply likes robots.
When it comes to trading, the robots encroaching on the jobs of human beings are algorithms rather than mechanical devices. Alex Tabb, partner and chief operating officer of the TABB Group, says you won’t get metal men trading, but you will get more and more resources committed to the development of smarter algorithms.
“Firms are bringing on more development people, data scientists, mathematicians and engineers. Speed is still very important but smart is even more important,” he says. “What we are seeing with algos is trying to become smarter, not just faster. They have to take in more data, analyze and compute quicker. They are trying to get to a knowledge state rather than just a data state. You need to derive knowledge from data and that then takes you to decision making.”
Big data is at the forefront of automated finance in 2015. Andrew Sheppard, a big data consultant on Wall Street and part-time finance professor at Baruch College in New York City, says the tools for analyzing big data are getting much better and more of the analysis is being demanded in real-time. As algorithms improve, Sheppard sounds a warning to traders: he expects trading will be conducted by machines running 24×7 with very little role for people. “Humans might supervise the machines, but even that is questionable,” he tells us.
Dr. Rupak Chatterjee, deputy director of the Financial Engineering Division at Stevens Institute of Technology in Hoboken, NJ is more optimistic about jobs in the industry. Someone has to create, and then test, the trading strategies before they are executed by computers, he said. Banks will need skilled financial engineers and quants with good database and data visualization skills: “They will have to know the standard Wall Street products and techniques, but those who bring in big data skills will hold an advantage over colleagues.”
Big data, Java, Python, Hadoop, statistics: the trading skills of the future
Chatterjee encourages his students to widen their skills so they can handle big data and analytics in trading, in commercial and retail banking, or even in health care. The skills are highly valued in multiple fields. Students should also learn a computing language or two such as Java and Python. Familiarity with non-relational databases like (NoSQL) and the Hadoop cluster environment is also extremely valuable, he added.
At a more senior level, banks need trading supervisors who understand exactly how algorithmic models work and the impact that choosing one above another will have upon P&L. Jim Jockle, chief marketing officer and Joe Saporito, global head of direct sales at Numerix – a company that develops software to analyze financial derivatives – say trading is no longer an island with its own P&L. A trading business has to show how it is complying with regulations, and how transactions impact the rest of the organization. Decisions like model choice are not left to the trader alone, Jockle said. Supervisors need to see how a model helps the organization, not just the trader’s bonus.
Bob Gach, capital markets lead for Accenture Strategy, said some trading roles are moving toward management of an investment bank’s balance sheet. Simple trading is being automated, but a whole category of trades need to be understood in terms of collateral, capital charges or whether they are a hedge (so-called XVA trading). On a more simple agency trading level, Gach said human traders are needed to work with customers in areas like trade finance, managing receivables, managing credit risks and doing swaps for alternative cash flows.
Jockle said that he would urge his son to develop a better understanding of statistics and better critical thinking.
“The successful traders will be those who can ask questions beyond the pre-packaged visualization presented, whether getting into underlying analytics, different Monte Carlo paths to find more opportunity, or standing on top and being able to query data better than someone sitting next to them.”
Most traders can use at most 10% of Excel, said Saporito. He said would-be traders should try to learn 20% – that will put them way ahead of most other traders.
Complexity is the way forward, experts agreed. Trading equities has limited upside; firms will be working across assets and in multiple currencies.
“The days of FX traders doing two currencies — that will be something we can look back on and say remember when?” said Jockle.
Tabb sees new traders learning to program at the same time they are learning to trade. “The ability to code and tweak these algos is going to be very important, especially for those market making firms who need speed and smarts to survive,” he explained. “The skills necessary to become a really proficient trader are changing, and it involves a strong base in technology.”
Mani Mahjouri, CIO of Tradeworx, said new traders should be familiar with technology, especially computer programming. Which program is best? Any is fine, said Mahjouri: “Computer languages change – the core value is being able to program in one language and then it is easier to pick up a new language.”