Open Platform

As traders clamor for trading algorithms, order management systems vendors need to be able to handle the demand.

Over the past few years, regulatory bodies have introduced market structure changes in an effort to reduce spreads, reduce execution costs and increase fair competition. For the most part, they have been successful, but the changes have produced side effects, such as fragmentation in the marketplace and smaller trade sizes. The fact that average trade sizes have decreased has given rise to a different type of trading counter-party—statistical arbitrage players that use fast connectivity to rapidly disseminate small, high-frequency amounts of shares and baskets of stocks into the marketplace. On the New York Stock Exchange (NYSE), for example, program trading—of which statistical arbitrage algorithms are a part—is now more than 40 percent of the marketplace.

As a result of these marketplace and regulatory changes, asset managers are emphasizing that their traders must help to increase their overall fund performance through better executions. In tandem, asset managers are continually looking for ways to increase traders' productivity and the capabilities of trading desks. These changes, among others, have contributed to the growing popularity of broker-provided algorithms. Generally designed to minimize shortfall against a specified implementation strategy, augment a trader's own skills and improve trader productivity, the benefits of algorithms include better executions and increased performance control for traders.

The order management system (OMS) has emerged as the primary—and logical—front end for providing traders with access to broker-supplied algorithms. When integrated into the OMS trading blotter, algorithms are easy for the trader to implement. Furthermore, these algorithms allow the trader to take continuous advantage of OMS benefits, such as centralized order management, pre-trade compliance and a seamless trading workflow. The trader can run a potential trade through compliance, apply a particular algorithmic strategy directly from the OMS blotter, set specific trade parameters, and electronically route the order via FIX to one or more execution venues. Fills then flow back into the OMS. The process is automatic and seamless—no double entry, no switching between software systems, no changes to current workflow.

Sell-side brokers originally developed algorithms as proprietary tools for use by their internal trading desks. Later, when brokers began offering these "black box" trading techniques to the buy side as a value-added, service differentiator, several attempted to generate interest by supplying their own front end. The brokers quickly realized that traders did not want more, but instead, fewer screens on their desktops. As a result, the OMS emerged as the practical launch pad to buy-side access. The leading OMS vendors listened carefully and took the first step of adding new algorithmic screens in their trading blotters. However, as more brokers enter the algorithmic space and introduce additional features—like pre-trade and post-trade transaction cost analysis tools—it may be a significant challenge for the OMS to keep pace with the associated integration demands.

For example, the next step after providing direct access to algorithms in the OMS is providing a real-time feedback loop that will allow the buy side to see how a selected algorithmic strategy is performing from the OMS. In short, it offers a view into the implementation shortfall, if any exists. Traders often add instructions to algorithmic strategies, indicating, for example, how aggressive the trader wants the algorithm to be and/or how soon should it complete the trade. If traders are trying to beat volume weighted average price (VWAP), they want to know in real-time just how far the algorithm is away from the target. If they want the trade completed by the close of business, they will want to watch throughout the trading day and know how close it is to completion. Although algorithmic brokers provide a variety of these monitoring and analysis tools today, they are available through broker-supplied front ends—separate from the OMS. In the future, the OMS—the conception point of the trade—will be the place for monitoring algorithmic trades.

Despite the current popularity of algorithms, some perceive little difference between one broker's set of algorithms and those of another. Areas where brokers try to differentiate include tools for pre-trade and post-execution analysis. On the pre-trade level, some brokers provide analytics that break down a trade before it happens so that the trader can evaluate what the trade trajectory might look like and determine that VWAP, for example, is the appropriate strategy. Post-execution, many brokers provide a scorecard that evaluates each algorithm's effectiveness. The report looks at, among other things, what was targeted vs. what was executed, the weighted spread of the name, the average volatility of the name, how long it took to trade the order, and how much volatility the trade demanded while the algorithm was out interacting in the marketplace.

Again, the downside is that these tools are currently in a separate front end taking up desktop real estate. Once integrated, these tools will allow traders to monitor and evaluate their algorithmic executions right from a single full-service OMS blotter.

From the OMS perspective, it is a significant challenge to conduct custom integration development for a large number of proprietary, broker-supplied algorithms and tools. Not only is it a question of volume, but also of style. Integrating one way with one broker and another way with another broker is inefficient. Therefore, it is imperative that the OMS architecture is open, flexible and able to accommodate multiple, disparate systems easily and quickly.

Furthermore, not only must the OMS dedicate resources to the initial integration; it is important that the vendor is ready to accept ongoing parameter modifications or the addition and/or deletion of algorithms from each algorithmic broker. That said, in today's competitive landscape, clients will want to take advantage of new functionality immediately—they cannot wait for the OMS' next major release, nor will they want to endure reinstallation or an upgrade. Therefore, OMS vendors must be able to integrate and provide updates on the fly.

Although it may take a year or more before the majority of asset managers are trading algorithms through an OMS, it seems to be an irreversible trend. In the meantime, OMS vendors are continuing to integrate more algorithmic tools and this movement is indicative of a higher-level trend. Buy-side asset managers are asking all of their vendors and brokers to work together to minimize the number of applications they must access on the desktop. They are willing to take advantage of new tools, but want them integrated into existing screens whenever possible. As OMS vendors work to integrate brokers and their tools at algorithmic speed, it is important that their architectures are flexible and open. An OMS that supports many integration options including Web services-, messaging- and file-based, will have a clear advantage. And this may translate into a competitive edge for its clients.

Michael Hayes is director of strategic alliances for Charles River Development, based in Burlington, Mass.