Waters: What are traders demanding when it comes to algorithms? Are there specific new features they want? Are they available now?
Ary Khatchikian, president and CTO, Portware: First and foremost, traders want access to a variety of different algorithms from different brokers. Over the last few years, we have seen clients take an increasingly active role in analyzing broker performance, and as a result, firms are making more informed decisions when it comes to directing order flow. Given that firms today have relationships with numerous broker-dealers and agency brokers, they are looking for ways to access all of these liquidity points efficiently. As such, we have seen a dramatic increase in the number of buy-side firms adopting true broker-neutral trading platforms that provide a single point of access to all algorithmic destinations.
In terms of algorithms themselves, market fragmentation in the US and Europe after the Markets in Financial Instruments Directive (MiFID) took effect is creating challenges for firms trying to execute orders efficiently and pursue best execution. In response, traders are looking for advanced algorithms that can help them probe multiple liquidity pools, including dark books, ECNs, alternative trading systems (ATSes), exchanges, and other market centers, simultaneously. Traders are also looking for algorithmic solutions for a wider variety of asset classes, including futures, foreign exchange (FX) and options. Many of these algorithms are available now, and we have seen increased interest from clients in these and other advanced trading strategies.
Waters: Are you seeing a move by buy-side firms toward developing algorithmic trading strategies in-house, as opposed to using those provided by brokers? If so, does this speak to greater sophistication among buy-side firms today?
Khatchikian: We have definitely seen an uptick in the number of firms developing and running proprietary trading strategies. This includes both alpha-generating strategies and proprietary execution algorithms, the latter developed as alternatives to broker offerings.
Few people would be surprised that alpha-generating strategies are on the rise, given the proliferation of quantitative hedge funds and the steady growth in electronic trading among non-equity asset classes. Perhaps more surprising has been the steady increase in the number of firms writing their own execution algorithms strategies designed to take pre-determined orders and execute them as efficiently as possible. One of the driving forces behind this trend has been the increased use of pre- and post-trade analytics. Again, firms today are paying much closer attention to their brokers' performance. Those who feel they can do better and have the resources to develop their own trading strategies are certainly doing so.
With that being said, the vast majority of buy-side firms still rely on their brokers for algorithmic execution. These firms lack the kind of internal expertise, financial resources and historical trade data that sell-side firms can bring to bear in the development of execution-based algorithms.
Waters: Is complex event processing (CEP) gaining ground when it comes to algorithmic trading? Is it being widely adopted? Does the CEP and algo trading combination have any inherent challenges or shortcomings?
Khatchikian: Firms today are faced with a dizzying array of so-called complex event processing engines, or CEPs. Marketed as high-frequency, low-latency analytical engines that can process data in real time and form the backbone of firms' trading systems, CEPs have gained attention recently as firms seek to deploy customized algorithmic trading solutions. Through numerous "bake-offs," CEP vendors tout their ability to receive and process an ever-increasing number of messages per second.
These solutions, however, are severely limited in their ability to provide firms with a true algorithmic trading engine. Beyond basic event processing, CEPs offer little else. Simply put, they only provide a single component of a firm's automated trading infrastructure. Everything else must be customized by the client firm. This includes the market data handlers that connect the platform to various feed providers and normalize all incoming data; connections to a front end, since no stand-alone CEP provides a front end; state management; and certified connections to all electronic market centers, ATSes, ECNs, exchanges, crossing networks, and broker destinations. In addition, the language on which these engines are built is proprietary. As a result, client firms must invest time and resources to learn how to write algorithms to their CEP of choice, or how to integrate a standalone CEP into their overall trade workflow.
Portware provides a comprehensive solution that addresses all the shortcomings of a typical CEP deployment. First, Portware Enterprise, our flagship trading solution, is itself an event-based platform. Used independently, Enterprise meets or exceeds the throughput requirements and analytical demands of most users. Using Portware, traders can create proprietary rules-based strategies and algorithms with ease, leveraging the system's rich, open application programming interface (API) and flexible trade architecture.
In addition, firms that want to deploy a dedicated, highly secure trading engine can opt for Portware Strategy Server. Unlike raw CEP engines, Strategy Server is a comprehensive solution. It comes pre-equipped with everything that other CEP solutions lack: integrated market data handlers, full state and position management, pre-certified connections to hundreds of market destinations, true multi-asset support, and so on. And because Portware is based on industry standard programming language, any custom development can be done with a minimum of human resources.
Waters: In terms of disaster recovery, what are the issues and challenges with algo trading? Is it available in recovery sites if the main trading offices are closed?
Khatchikian: We have always thought about disaster recovery in the context of different scenarios: localized hardware failures that can be addressed with on-site solutions; and truly catastrophic events, such as a building collapse, or extreme weather, that would destroy or disable a firm's entire trading infrastructure. For the latter, so-called "cold" or "warm" backup solutions that reside in secure, alternative trading locations are required. Portware has always worked closely with clients who maintain remote recovery sites, ensuring that our systems are ready to go at a moment's notice.
For localized hardware failures, firms traditionally had to rely on variants of warm backup solutionssystems that could be up and running in an hour or a few minutes. In today's fast-moving market, however, even a few seconds of downtime can lead to significant financial loss. While hardware failures are extremely rare, the potential impact of a trading system failure can be enormous. As a result, firms are looking for more advanced backup solutions.
To address this need, Portware recently introduced Portware High Availability (HA). Unlike standard disaster recovery and backup systems, Portware HA eliminates the risk of downtime by delivering an instantaneous failover solution. Portware HA continuously monitors the client's core trading environment, duplicating messages and maintaining real-time copies of positions, outstanding orders and order status, in a separate server environment. As soon as Portware HA detects an interruption to normal trading operations, it immediately switches to the backup server environment. This eliminates the need to restart trading systems, check positions and re-establish connections. Most importantly, Portware HA is extremely efficient, requiring a minimum of system resources.
Waters: Given the prevailing market environment, will the reliance on algos and advanced trading systems increase or remain the same?
Khatchikian: Market participants on both sides of the Street are certainly facing tough times. More than ever, firms are focused on increasing workflow efficiencies and reducing trading costs. At the same time, firms are facing an increasingly fragmented global marketplace. These factors, combined with tougher best execution mandates in the US and Europe, are leading more and more firms to rely on advanced execution algorithms that can help them navigate a very challenging marketplace and seek out liquidity in its numerous forms.
Of course, clients want and need access to a variety of different broker destinations, not just one or two. However, using multiple independent broker front ends introduces the dreaded "swivel chair" effect and a host of other workflow inefficiencies. In response, firms are moving away from rigid broker front ends to systems like Portware that offer a single point of contact to any global electronic trading destination. This kind of flexibility and control over order flow ensures that traders can route orders to the most appropriate trading destination, allowing them reduce trading costs, minimize market impact, and achieve maximum trading efficiencies.
Waters: What will algos look like in the next three to five years?
Khatchikian: We certainly expect to see strong growth in algorithmic trading across additional asset classes, as brokers continue to develop and market new strategies for FX and listed derivatives. Algorithms that seek out non-displayed liquidity will become more advanced, as will those that address more complex order types and illiquid securities.
Ultimately, firms will have to be ready for whatever the future brings. A constantly changing marketplace will bring with it new opportunitiesand new risksthat will give rise to a new class of automated trading strategies. Those looking to leverage these new algorithms must ensure that the tools they adopt today can meet the demands of tomorrow's marketplace, whatever they may be.
Ary Khatchikian is president and CTO of Portware, a financial services solutions provider. For more information, please visit www.portware.com.