From the Jan-Feb 2006 issue of Advanced Trading
In their never-ending quest to please their customers, brokers increasingly are blurring the lines between the traditional full-service model and the role of technology vendor. The infusion of algorithms into the marketplace has made brokers' electronic execution businesses even more vendor-like, as they become firms that sell products as well as advice.
The newest wave of algorithmic customization aims to streamline the process of algorithm creation and alteration, and put more power in the hands of the buy-side trader. The question is, How much power do buy-side traders want, and how much of the execution business do they want to leave in their brokers' hands? In 2006, brokers will attempt to find the sweet spot between algorithm provider and execution provider.
JPMorgan Securities currently is at work on a project dubbed AlgoFlex, which would allow clients and internal users to achieve "dynamic configurability" by enabling them to change algorithms on the fly, says Carl Carrie, vice president of JPMorgan Securities, who heads product development for algorithmic trading as part of the firm's electronic execution services (EES) initiative. According to Carrie, the service should be available sometime in the first half of 2006.
"We are working feverishly to develop technologies to let us rapidly redevelop algorithms on the fly for both internal and external use," Carrie says. JPMorgan Securities also is exploring ways to offer clients dynamic advice through the Web to accompany the algorithms they are building; the company is working on an internal platform called Algorithmic Workbench through which it hopes to provide the functionality.
Not to be outdone, in late 2006, Goldman Sachs' algorithmic trading division will release Building Blocks - a set of algorithm customization tools - to all of its RediPlus trading system customers, representing about 600 firms and 5,600 individual accounts, relates Jana Hale, head of algorithmic trading at Goldman Sachs. The idea is that traders will be able to toggle various settings and parameters common to all of their algorithms, and in a sense create their own customized trading strategies.
Although this may create a picture of traders forsaking conversations with their brokers in favor of joystick trading and making finite, real-time adjustments to their algorithms all day, Hale is quick to caution that this is not the case. "Building Blocks still involves a virtual questionnaire, and we would have a consultation with clients to understand their trading styles," Hale says. "Most clients say they want their algorithms to go slower or faster, or to hide more order flow in reserve. But many do not really understand the types of tools that do this, and, frankly, they don't want to be bothered with learning them."
Do You Want Advice With That?
As such, full-service brokers tend to emphasize algorithms as part of the overall full-service offering, and as an adjunct to the fundamental trading advice traders expect. To do otherwise is to potentially alienate clients. "If someone really wants to put a Lego tool out there without the advice, I don't see clients wanting to explore that on their own," Hale says. "It is like asking them to install their own computer every day."
Agency-type brokers have a bit more flexibility in offering stand-alone products, as there is less concern from clients that the algorithms they deploy - which are sometimes versions of statistical arbitrage algorithms that the broker uses for its own trading accounts - will be spied upon by proprietary trading desks. BNY Brokerage, the agency brokerage division of the Bank of New York, offers clients a third-party vendor platform, FlexTrade. The platform allows clients to devise their own trading models, which they then can execute through Sonic, BNY's direct-market-access (DMA) platform, relates Derek Morris, cohead of program trading at BNY Brokerage.
Some clients prefer to send orders directly through Sonic and bypass the BNY desk; others require significantly more attention, Morris explains. But in no case do his clients call for on-the-fly customization as an algorithm is running its course, he asserts. "The problem with [dynamic customization] is paralyzing clients with infinite choice," Morris says. "We try to give them some consultation along the way."
Clients usually are loath simply to tweak a setting and press the button, and when they do, they are sometimes very sorry, Morris adds.
"I can't tell you how many times a week someone puts in an order and they have no idea what happened," Morris says. "I don't think people want to turn something on and leave it on autopilot. Maybe they don't want to climb on the wing and lift the aileron, but they want to pilot the ship."