Ben Sylvester, head of U.S. equity trading at J.P. Morgan Asset Management, also is also relatively new to his post. But like York's Maier, he hit the ground running. Sylvester already is undertaking a major project to build an Execution Optimizer, or EO, that will further automate the trade workflow by utilizing transaction cost analysis (TCA) and portfolio manager profiling to generate strategies and ultimately execute on them.
Sylvester says that as part of the multimillion-dollar project to automate trade order flow, he specifically is focused on building a strategy server to automate and build sector models. "This project is a massive investment of people, time and infrastructure," he reflects. "This will air traffic control all of the order flow on the buy side and help define who gets what trades and how it gets traded and where instantaneously."
According to Sylvester, how firms use TCA and data analysis should be changing with the times. As the markets, portfolios and traders change, trading strategies should change as well, he adds, and TCA and other data analysis can help improve those strategies. "Transaction cost analysis has always been a 'have to have,' a necessary evil. But it's a gold mine of information [for my trading desk], and I should really learn how to harness it," he says.
Breaking Down Portfolio Managers
In addition to transaction data, the EO will leverage manager profiling data. "I looked at the alpha of each portfolio manager and where that intersected with the trade," Sylvester explains. "Breaking down a portfolio manager from a statistical perspective was important to then be able to build strategies more efficient for that manager." This type of information, he asserts, is empowering for traders, as it helps them understand the short-term alpha generated by a portfolio manager and tailor trading strategies for particular managers.
The technology will rely on a factor model that accounts for J.P. Morgan's 28 different portfolio managers and their inherent trading styles. "With the interfaces we're building, order flow would come down through the OMS into the strategy server, where it decides whether to give the order to a trader or trade the order automatically," Sylvester explains, noting that traders will be able to monitor the decisions made by the strategy server and can interact with the order flow -- for example, if news breaks. "It's considered their order flow, so they can improve the models when need be."
As more of the order flow is handled in the automated environment, Sylvester stresses, traders will have more time to add value to the investment process in other areas. "They'll still be trading the more challenging order flow, so now they can take their heads up and look around and think about what it should look like," he notes. "I hear traders all the time say, 'I would love to be more involved in the investment process, but I do too much trading.' "
In determining the best way to execute a strategy, the EO also analyzes historical order flow -- both from days before and after trades -- to understand positions. "So it takes the origin of an order -- the portfolio manager -- and the strategy server will decide what strategy or strategies associated with that portfolio manager will be most effective and what the time horizon should be," Sylvester describes.
The system might direct the order to a trader and then specify that the order be traded over the next three hours. It helps define the participation rate or level of aggression, and may even dictate the venue in which the strategy should enter the market. Or the EO may execute the order automatically based on those same parameters. "The strategy server will be a factor model based on a variety of different models based on historical transaction data that is constantly updated with real-time transactions," Sylvester says.
Sylvester notes that he and his team worked with Citigroup's Automated Trading Desk (ATD) unit to validate the idea behind the EO initially and now is codeveloping the factor model with the group. "ATD is an automated market maker and they profile stocks, so I approached their director of research with the idea to see if my assumptions were valid and to see if they could enrich the process," he recalls.
The internal strategy server is expected to be rolled out to the U.S. equity trading desk and in production by the end of the second quarter. Sylvester suggests that the system represents the growing trends toward automation and the incorporation of real-time data in the decision-making process -- "using real-time venue analysis to pick your execution destination, for example," he says.
But regulation also will be high on Sylvester's agenda for the foreseeable future. "The financial frontier is being defined by the regulatory environment," he says. "So we need to be mindful [of regulators], because we're all subject to whatever they decide."
Ben Sylvester, Head of U.S. Equity Trading, J.P. Morgan Asset Management
Assets Under Management: $80 billion for the U.S. equity desk, $300-plus billion globally in equities and $1.3 trillion across the entire J.P. Morgan Asset Management organization.
Time in Current Position: 1.5 years.
Technology Used to Trade: J.P. Morgan Asset Management relies on the Longview OMS. For execution, active single-stock traders utilize Bloomberg EMSX and portfolio traders rely on the ITG Triton EMS.
Size, Structure of Trading Desk: The U.S. equity desk includes 10 traders structured around market capitalization -- specifically, a large cap group, a small and mid cap group, and a quantitative trading group. Sylvester oversees equity trading for the Americas, which includes Latin America and Canada. "It had previously been a sector-based trading desk, but we wanted to turn back in and face portfolio managers and products in a more efficient way," he explains. "And it's more trading-friendly instead of being all over the cap scale. Fewer managers facing fewer traders builds better probability sets of positive outcomes."