In today's financial markets, speed and accuracy are everything. Capitalizing on momentary differences in buy and sell prices can add up to millions of dollars in profits -- or losses, if mishandled.
When Bank of America's Global Markets team decided to reengineer its various trading systems two years ago, it abandoned the traditional Wall Street approach of separate silos for each trading desk. Instead, the bank's IT team began developing a single distributed system that worked across disparate trading environments -- whether selling treasury bonds, foreign exchange products, or farm futures. And it did it faster than anyone thought possible.
The Pioneer Trading System technology team was assembled in mid-2005. By the following January, it had rolled out its first deliverables to a handful of high-volume traders. Today, more than 200 business users in Chicago, London, and New York use the Java-based Pioneer platform, says Nathan Zhang, project lead and managing director of the bank's electronic trading technologies.
Before Pioneer, says Zhang, orders were entered manually into Excel spreadsheets, which then connected directly to each trading exchange's system. But this methodology proved brittle, lacking price verification functionality, which is essential in rapidly moving markets.
"Some price disparities may only be there for 2 seconds, so by the time the order is placed, it could be too late," Zhang says. "With Pioneer, orders get executed in milliseconds instead of seconds, while traders continue using their familiar Excel sheets to connect to the Pioneer trading platform."
Pioneer also gives traders the ability to manage multiple transactions at once, allowing them to build and execute sophisticated trading algorithms. The bottom line: Since implementing Pioneer, trading volume for the bank's busiest trading desks has increased by a factor of 10.
Zhang says a lot of the work involved optimizing systems for speed, using techniques such as multithreading to submit orders via multiple routes, and working with traders to hone the algorithms Pioneer uses to respond instantly to market movements.
One key requirement for the project was speed to market, says Steve Beasty, CIO of Bank of America Securities. Buying certain components instead of building them enabled the bank to get a useful pap in trader's hands within six months.
For the authentication piece, the bank turned to Securent's Entitlement Management Solution. Besides certifying each trader's rights to buy or sell certain financial instruments, Securent also allows the bank to create rules and enforce policies -- such as capping traders' daily volume if the market drops by a specified percentage.
"Without Securent, deploying the Pioneer application would have required a great deal of custom coding of authorization logic, making the application brittle and difficult to adapt to meet new compliance requirements or changing market conditions," notes Howard Ting, senior director of product management at Securent.
The toughest part of projects such as Pioneer is taking the time to do things right while still delivering value quickly, says Beasty.