Optimizing Trades Real-Time

After two years of development and testing, Robeco Investments is going live with a new trading tool. Director of trading Mark Kuzminskas’ latest technology project marries data from portfolio managers’ trading history with real-time stock analytics for the desk. The goal of the project was to optimize trade scheduling based on the PM’s tendencies in order placement, using real-time analytics.

“We wanted the system to help us achieve enhanced execution performance for orders that not only flow through the optimizer, but also those that remain in the hands of the traders,” Kuzminskas said.

He has spent countless hours trying to increase efficiency in today’s low-volume trading environment. This new execution optimization tool, which launched just last month, is key to best execution, he said.

“The bottom line is, we are now pairing up historical PM trading profiles with real-time tick-by-tick data to optimize trade execution,” Kuzminskas said. By employing this technology, he added, some order flow can be automatically routed, freeing him and other traders to handle the more difficult orders.

Here’s how it works: The system looks at each PM’s alpha profile and the software analyzes it. Then when an order comes from the PM, the optimizer looks at the PM’s historic behavior and projects a trade execution schedule given what’s going on in the marketplace. At this point, the trader has the option to let the optimizer handle the order, or he can pull it out of the system and trade it on its own. 

If Kuzminskas opts to work an order, he still can select what kind of order type will be used, either market or limit, as well as venue and broker, just like the optimizer could if it were working the order. Today, he still looks for a natural contra to fill his orders, often using electronic blotter-scraping tools from Liquidnet, ITG or BlockCross to get the size he wants. Or he will take the high-touch natural option. If neither is readily available, then he puts the orders into broker algorithms while searching for blocks.

“We’re using the brokers as conduits to get to a market center,” he said. “Instead of using the brokers to route our algorithms, we’re using trade-cost analysis and market analytics to determine the routing ourselves, and access these markets at our discretion.”

The choice to work with SJ Levinson & Sons was straightforward, Kuzminskas said, as the vendor already had a “very good” TCA product that could be leveraged to help develop real-time self-directed trading solutions. This, he added, was more efficient than developing a similar system from scratch, which was at one time considered and briefly pursued.

“This application will draw orders from all our silos-program orders, high-touch single stock, as well as self-directed electronic trading,” he said. “It enables us to reorganize our order flow so traders can focus on the orders that have more hair.”