02.22.2013
By Terry Flanagan

Most Innovative Buy-Side Trader

When a portfolio manager hands off a buy or sell order to a trader on Ben Sylvester’s team, that manager may walk away, but his or her influence remains for the execution.

Sylvester, head of U.S. equity trading at J.P. Morgan Asset Management, has implemented a trading methodology in which the characteristics and investment style of the portfolio manager are important considerations. The idea is to intersect the trading and investing processes in order to achieve the best possible execution.

“Instead of looking at orders throughout the day and trading the market, we look at the orders on an individual basis, and trade the manager,” Sylvester told Markets Media. “For example, if a certain manager trades Cisco decidedly different than another manager, this is a huge information component.”

The 12-person trading staff at J.P. Morgan Asset Management includes eight traders, two quant analysts, a business manager and Sylvester. The trading desk oversees equity trading in the Americas, including equities, futures, options, and preferred securities, and trading is on behalf of all fiduciary elements within the asset manager. J.P. Morgan manages about $1.4 trillion for institutional investors worldwide.

“Instead of looking at orders throughout the day and trading the market, we look at the orders on an individual basis, and trade the manager.”

Sylvester joined J.P. Morgan in 2009 after trading for about 7 years at Babson Capital Management in Boston, and he has made his mark on how the trading desk operates. “Thematically, it’s about data, visualization, and systematic trading, or automation,” Sylvester said.

“We started with what we call manager profiling, which is much like a forensic exercise in that we look at the fingerprints of the manager,” Sylvester said. “This includes the date he sends an order, order size, what does the order look like, what is the stock-price action before the order is sent, etc. This is where trading intersects with the investment process, and we get an indication of the approach this manager takes to investing.”

A portfolio manager may be momentum-oriented, or trade in down markets looking for mean reversion, or inclined to trade in flat markets. Rather than pack all the data on spreadsheets, Sylvester’s team visualizes the styles. “If you picture a line going up to a 45- degree angle, or a smile curve (which would be a mean-reversion trader), or a flat line, visualization gives you the sense that these are three different types of managers,” he said.

Visual articulation of data is a key expediter, according to Sylvester, “because it’s almost instantaneous, and once you can visualize it you can do something about it.”

“We pick up on the data and do the modeling exercise, but we’re not trying to influence the process. We’re just trying to see what the intersection looks like,” Sylvester said. The methodology “allows us to create specific trading strategies either to a product or to a manager.”

Top-10 Rankings

Sylvester is doing something right. J.P. Morgan Asset Management’s U.S. Small-Cap Growth strategy graded out among the 10 best managers in overall execution process for 2012, according to Zeno Consulting Group, which analyzes institutional managers’ trading processes and costs. JPMorgan’s Large-Cap Value and Mid-Cap Growth ranked in the top 10 for best utilization of brokers, and its Mid-Cap Growth, Small-Cap Growth and Small-Cap Value styles ranked in the top 10 for commissions.

Sylvester says the probability considerations on the trading desk can be compared with playing poker. A hand of cards is to the player what the portfolio manager is to the trader, and the other players’ behavior and cards is to the player what the market is to the trader.

“If you play poker, you’re dealing with two variables — your own cards, and the other players. I very much view our trading in the same fashion,” Sylvester said. “We have a probability set of our managers’ behavior — there are varying degrees of that probability, much like your poker hand represents, but it’s stable and you can model it.”

“And then much like your opponents in poker, the market is unstable, but there are some things that you can pick up on, for example related to momentum or earnings,” he continued. “Granted you don’t know what somebody else’s hand is, but you can get a sense of how it will play out.”

Once a probability set is modeled and visualized in line with the trading desk’s experience with the portfolio manager, then systematic, customized trading strategies are created. This is the third theme to Sylvester’s trading methodology.

“If you play poker, you’re dealing with two variables—your own cards, and the other players. I very much view our trading in the same fashion.”

When implementing automation and systematic trading, questions need to be asked about how it will be set up, for example at what trade speed. “You can get a sense of trade urgency from the data and then calibrate to the specific circumstances,” Sylvester said. “If there’s no sense of urgency when I trade a portfolio, I can be relaxed about it. But we might have another manager who’s very event-driven that needs a heightened sense of urgency. When we think about how a trader caps trading costs, rather than making it up as you go along, this is a qualitative and systematic approach.”

J.P. Morgan built its own customized algorithm. “We had to have something different than the normal generic package from a broker,” Sylvester said.

Regarding performance measurement, “we align ourselves with the portfolio manager,” Sylvester said. “For managers it’s all about absolute return, and for us it’s about absolute lowest cost. We look at it on a historical basis and also monthly; it’s measurable and pretty straightforward.”

The J.P. Morgan Asset Management trading desk has evolved beyond just traders. “We have multiple CFAs and Masters degrees,” Sylvester said. “It used to be all traders but now there is more of a quantitative element.”

Sylvester said the framework of data, visualization, and systematic trading is fully complete, but it will always be a work in progress. The next step forward, which is underway, is developing the ‘next-generation’ trader workstation, which among other things will include an element of machine learning, or artificial intelligence.

“This will be powered by our data and built with a very high-end, futuristic visualization capability where we redraw the screen to look dramatically different from anything you’ve seen before,” Sylvester said. “Most importantly, it brings all the attributes back to the active trader. Instead of having to search through reams of data and look back on managers’ profiles and previous trading strategies, it’s all embedded in the workstation.”

“That’s the next frontier,” Sylvester added. “We’ve got the components in place and working right now, but how do we bring them all together and take the active trader to another level?”

Related articles