The Gateway to Algorithmic and Automated Trading

Tech Forum: The OMS - Moving On, Moving Up

Published in Automated Trader Magazine Issue 03 October 2006

One type of trading technology that has been much affected by the growth in algorithmic trading is the OMS. AT talks to five leading OMS vendors to find out how changes in the market are affecting their businesses and how they are dealing with those changes. With: - Eric Bernstein, head of sales for North America at Linedata - Richard Hooke, global product director at Latent Zero - Michael Roberts, director of European operations at ITG Solutions Network - Stephen Engdahl, Director of Product Management at Charles River Development - David Csiki, managing director at INDATA

With the arrival of DMA and algorithmic trading powered by new technologies, the relationship between the sellside and buyside has been changing. In what way does this place pressure on OMS providers?

Bernstein: It's placed pressure on vendors from a technology perspective to allow for the rapid change and implementation of these liquidity providers and destinations. From a DMA perspective, the only pressure on Linedata Services is keeping up to date and ahead of the curve. We've put ourselves in a good position because of our technology and the way we've developed and designed our product. From the algorithmic trading perspective, we have embedded the algorithms in our application, so for the user it's seamless. The way we've architected LongView Trading, clients don't have to do a new install every time we add a new algorithm as all the heavy work was done years ago. This means it's easy for customers to add brokers' algorithms to their systems, and easier for us to add algorithms to ours.

Hooke: It gives us a need to provide a service that supports that model, so that's what we've done and will continue to do. The pressure, if you like, is for us to support whatever business processes our customers want to adopt, and this includes use of algorithms and DMA. Hence we need to provide an integrated service through the OMS so our customers can access these execution paths.

Roberts: Algorithmic trading, along with black box trading and DMA, are all part of a trend towards etrading. This trend has to some extent changed the relationship between the buy and sellside. The buyside trading desk is much more sophisticated these days as a result of these trading methods available to them, so the technologies they need to deploy need to be more sophisticated as well. So what the OMS industry is trying to do is build more sophisticated trading systems; the best example of that is in the demand from the buyside to integrate execution management systems in the OMS. As a result of the buyside being more sophisticated, EMS is now integrated and integral to the underlying OMS. So the difference between the buyside and sellside is less and less.

Engdahl: With the buyside taking on more of the responsibility it had previously delegated to the sellside, it needs more sophistication. So we've added pre-trade analytics and pretrade transaction cost analysis, covering cost versus time risk, which helps the buyside make a decision on how to execute each order and how to set up each algorithm to minimize implementation shortfall. We have built interfaces with a number of DMA and algorithmic providers to ensure the buyside can access these trading venues through a single application, which enables asset managers to take responsibility for trade execution through these channels.
If I have a large order which I'm working via an algorithm, I'll now get executions in far smaller share amounts back from the marketplace so there's a need to process the larger number of execution records coming back. So we've upgraded our FIX engines to manage the growing number of order fill records.

David Csiki, managing director at INDATA

David Csiki, INDATA:" FIX is a big part of what allows us to keep up with the demands of our clients..."

Csiki: It puts pressure on OMS providers to keep up with the changing technologies and the various options that are available.
At INDATA, we have tried to stay client driven, as in the midst of the sellside soliciting the buyside with the various packages that they have available, our focus needs to be on the platforms that are required by our clients. FIX is a big part of what allows us to keep up with the demands of our clients, since we are able to leverage FIX as the standard protocol upon which all client connectivity is based.

Can you illustrate how OMS tools can give buyside traders more independence?

Bernstein: The market has changed; the buyside trader used to deal exclusively with the sellside, but now with tools like the OMS, the process has been streamlined and the buyside trader has more independence and autonomy in the market.

Hooke: Some people look at ways of achieving independence through different stand alone pieces, such as execution management tools. I see it as part of what the OMS needs to provide; enabling a services-based architecture and providing an integrated platform that will enable customers to use the execution paths they want to use. We've integrated algorithmic support into the core of our product. So through LatentZero's product you can capture an order, work out how you want to process that order, and then send it out to a broker. I see a need for the way you support these services to be independent from the actual execution path, of a broker's desk, algorithm or DMA route. It's important that we as an OMS provider are neutral from these execution paths, and that we strive to make sure our architecture is open to new execution paths; a lot of execution management systems are effectively broker-specific meaning if you wanted to route an order to another broker you would need another product. We want to provide our customers with an overriding framework to support the full range of execution paths on a broker neutral basis through the OMS. The benefits for them include a better use of desktop real estate, which is quite precious on a lot of buyside desks where lots of applications are running different things. You avoid the need to run a lot of execution applications with our system, which is a key goal. DMA in particular is a reasonably new development for the buyside. It's been around for some time, but in terms of adoption and interest by the general buyside community rather than specialist hedge funds, it's new. So having DMA built into the OMS means you can look at a particular order in the context of what is available on the market.

Roberts: From ITG's perspective, it extends beyond the OMS. We increasingly talk less and less about the OMS, and more about best execution trading systems. OMS is the engine, with other component pieces on top that help that process.
We're integrating functionality-rich pre trade analytics with the OMS, as well as integrating post trade cost transaction analysis capabilities on the trader's desk, with the OMS. This provides the bigger picture. In the old days, the OMS and blotter is all the trader saw. To empower that person to do more, we're providing him with analytical tools to help and assist him in that process.

We are a broker neutral system. We have brought into our desktop over 150 different algorithms, provided by over 20 different brokers, with one tool on the desktop. This gives the trader all the information on pre trade analysis he needs for a trade, and one central tool to manage executions with a variety of different brokers, which stands up against a non broker neutral platform, where a buyside trader needs to manage and configure multiple tools on the desktop.

Csiki: The OMS really acts as a central gateway that allows traders to connect to the platforms that are out there. The OMS allows traders to connect to multiple platforms, making integration easy. Traders can go to lots of platforms and access pools of liquidity and different algorithm trading platforms, and whatever tools and packages add extra value to their efforts. This gives them a lot more flexibility as they can connect to all of these options from just one system.

How are OMS providers responding to the technology infrastructure demands required to deal with the increasing messaging traffic and data volumes associated with algorithmic trading?

Eric Bernstein, head of sales for North America at Linedata

Eric Bernstein, Linedata:"...we went from a client-server architecture to an N-tiered system"

Bernstein: We planned for a couple of things. In the late 90's and early part of 2000 there was massive consolidation of asset management companies and the number of accounts, executions and users were increasing.
In response to that we went from a client-server architecture to an Ntiered system. We've had that in place for about four years now. In the US, we had it in place post-decimalisation when volumes were increasing, so it allowed us to be very prepared for the volumes of orders, executions, and messages that were coming with DMA and algorithmic trading.

Hooke: Some OMS vendors will struggle because they don't have the architecture in place to handle volumes and FIX messages; but we've had that capability in place since 2001, because we have a messagebased architecture. So the potential bottle-neck caused by upwards of a million fills a day coming in are distributed out intelligently to servers and over to the services that need the information, rather than being channelled through one choke point.

Latent Zero has always prided itself on its real time infrastructure and ability to handle large volumes of data. Some of our largest customers, which are by definition some of the market's largest players, have done their own benchmarking on our products' ability to handle volumes and FIX messages, and we've demonstrated to them our ability to handle it.

Roberts: Algorithms and DMA are a challenge through the increased throughput and volume platform environment algorithms have created. But if you need to process a transaction in one tenth of a second, the cost of the hardware needed to do that today versus 10 or 15 years ago is significantly lower now.

The tools and architectures available today support that level of procuring, so it's all about adapting the latest and greatest today. If you're an OMS vendor that first created a system in the late 1980s or early 1990s, the technology available then that your system is based on is probably not up to today's throughput demand. Our platform was designed in 2000, so the technology is very current today.

It's about adoption of better technology, so you have to design your system with the technologies available today like .net and Java. For the industry as a whole, this presents a challenge. If you made your OMS in 1996 and its live today, reengineering that is not going to be easy.

Engdahl: A lot of OMS vendors have been in this space a relatively long time, so they have antiquated technology. We've invested a lot of research and development and are rolling out our multi-tier, Web services-based platform. This gives far more scale and far more flexibility, and fits better into the customer environment of today. If an OMS vendor is not consistently investing heavily in R&D since some time ago - it's not something you can do overnight - they may not survive.

It's all about scalability, throughput and being real time, event driven; if you're working with DMA, you can't wait. Every event needs to be fresh, now, on your screen. We hear a lot of speculation about what some of our competitors are doing. However, the key differentiating factor is that anyone can get through a quick demo or presentation, but with the size of our customer base we must provide something that doesn't only look good on the surface. When we release a product it has to be enterprise quality.

Csiki: The biggest thing algorithmic trading is doing that OMS vendors have to deal with is increased volumes in terms of messaging traffic. Essentially, the algorithmic programs take large block orders and will only execute if the conditions of the algorithm are met, which means an increased volume of trade execution activity. From a system processing standpoint this puts an additional burden on the OMS; the OMS has to perform as seamlessly and as fast as it was in the lower volume environment. We've kept up with system architectures; we use Microsoft as our technology partner and we use 64 bit processing, which is twice as fast as the 32 bit systems that are typical out there. This means our system can handle large volumes efficiently.

How are OMS providers likely to differentiate themselves to cater for the needs of algorithmic trading?

Bernstein: Right now it's hard to differentiate. Some may try to write their own algorithms, which we hear is going on in the OMS space. Linedata Services is not delivering our own algorithms; we're a software and service company. We'd rather partner with these companies that provide the intellectual property in this arena and have a strong working relationship than develop them ourselves.

Pretty much all the larger vendors can roll out algorithms to clients pretty quickly. So I'm not sure how we'll differentiate. For us, being somewhat agnostic and connecting to as many brokers as possible is our differentiation point. Other than that, I'm not sure how others in the OMS industry will differentiate themselves.
Richard Hooke, global product director at Latent Zero

Richard Hooke, Latent Zero: "The OMS has to provide access and availability to the service that the broker has offered..."

Hooke: The key is around our ability to quickly deploy support for new services. The OMS has to provide access and availability to the service that the broker has offered; we're not in charge of that service, but we must provide access to it. The way we see ourselves differentiating is in providing fast access, deployed quickly, with no need to upgrade core applications every time we need to support a new algorithm from a new broker; it's a configuration activity rather than a fundamental change for customers.

Roberts: Differentiation from my point of view comes from the breadth and depth of algorithms you can present. Customers want to know how many algorithms you connect to; if I say two or three, they're not going to be interested. They want breadth and scope; we currently connect to 15, including all of the big guys like Merrill Lynch.Where ITG is really leading is its inclusion of pre and post trade analytic capability, so you know what the best algorithm to select is and how it performed.

Engdahl: Our broker neutrality is a big part of our differentiation. We're connected to multiple different broker sources and we support that full worklflow including both the screens on the desktop and the FIX network behind it. Our addition of independent and broker provided pre trade analytics, plus new shortcuts to enable one click access to even the most complex algorithms, further streamlines the workflow between our product and the algorithm.

Csiki: Really, to offer options and to the extent that options are available, to offer those that are widely used. This will and is happening in two ways: one, offering multiple options and connecting to all algorithmic platforms that customers are demanding; and two, offering very tight integration with the different algorithm packages that are out there so everything operates seamlessly from the trader's standpoint.

How is the arrival of direct market access and algorithmic trading changing OMS transaction fee structures?

Bernstein: From our perspective, we don't think it is. We're not a brokerdealer so we haven't changed our model. We charge the brokers a license fee and do not participate in commission sharing.

Hooke: I think it's interesting. I know that different vendors are adopting different pricing models. Currently, there is a move towards a transactionbased fee, but I don't think this is sustainable in the long term: I see it as a relatively short-lived phenomenon. I do believe that we'll see a move towards a monthly service fee, rather than a transaction based fee.

Michael Roberts, director of European operations at ITG Solutions Network

Michael Roberts, ITG Solutions: "We've stopped talking about just OMS and instead we talk about the best execution platform."

erts: We charge on the cost of the OMS platform. I don't price my OMS any different to how I did prior to algorithmic trading becoming a fashionable thing; now it's there, it provides greater opportunities.
The McGregor Financial Network is a private IP network that connects our brokers with the buyside. The sellside pay ITG Solutions Network for connecting to the buyside, which is the normal model typically based on a connection basis, rather than a transactional basis.

Engdahl: We are an independent technology company. Our customers buy the license for our software, so from our standpoint an annual fee is the model we're standing by. There are some competing products becoming more aligned to a particular broker, but you then lose the benefit of the neutrality that you get from a truly broker neutral platform.

Csiki: A lot of it is being driven by some OMS providers moving to a transaction-based model. What some are doing is driving a fee structure off connecting to a particular package into the broker itself at the expense of others.
While some OMS providers may do that, we've always stayed broker neutral, not aligned with one algorithm or package over any other. Essentially, we just connect to the different packages that our clients use, and keep connected to those platforms as they change.

While most OMS providers say that they have taken this stance, if you look at the ownership structure of some of them, several are owned by major brokers. Others, including INDATA, are truly independent.

The OMS market is a mature market. In what ways will algorithmic trading provide new market opportunities for OMS providers?

Bernstein: The OMS market is mature and we do have dominant players in the market place, but we are seeing more of an EMS/OMS focus of late.

New opportunities, from a financial perspective, come from the OMS adding functionality to differentiate itself. As the OMS is adding functionality, it is becoming more of a hybrid between an OMS and EMS product. This is allowing smaller firms to come into the OMS market to get the benefits; OMS used to be portfolio management, trading and pre-trade compliance but now it's expanded and giving smaller firms more independence and autonomy.

I don't think necessarily that the OMS is mature. In our own product planning, we're constantly looking to see what's next to do; the range of asset classes required by our customers is always changing so we're always adding more. Some areas like equities and fixed income are more established, but integrating and providing execution capability into the OMS is what we see as 'the next stage of evolution of OMS'. The opportunity for providers like us is to differentiate ourselves by our ability to do that.

Roberts: There's no question that the traditional OMS market is mature. We've stopped talking about just OMS and instead we talk about the best execution platform. The opportunity for the vendor is to work up and down stream as pre and post analytics will increase take up of algorithmic trading. It's really through that that you get increased differentiation and opportunity.

The OMS space has grown and expanded. In a trading environment, as we bring in more algorithms and DMA, we believe firmly that given the time requirement of trading, having execution and order management rationalised into one system eliminates a critical breakpoint in the trade life cycle, and increases a trader's efficiency.

I would say that it's the evolution of the product itself that's making the opportunity, for a more integrated product. The OMS is a very deep package and algorithmic trading is only one part of it. Other key functions include portfolio management functionality and compliance.

What new functionality have you developed and are offering on your own platform in response to algorithmic trading?

Bernstein: We launched our FIX network, Longview LyNX. LongView LyNX is more than just a FIX network as we view it as a trading portal providing FIX connectivity access to alternative liquidity pools.

It has put us in the position where we can bring new products and asset classes to market very quickly. In doing this we've made it easier for clients to select what they want.

Hooke: We provide fully integrated support for execution to an algorithm destination or broker algorithm through our OMS. In a similar way that you can send an order to an agency desk, you can send it to an algorithm and automatically pick up the fills you get back, so you can check the progress your order. Brokers will always come up with new ideas for algorithms, which means there will always be more execution choices available for an order. As the buyside uses algorithms and DMA more, they in turn will develop more ideas on how they would like deals to be executed. As a result, the buyside will be able to influence the development of the algorithms it wants and the ones they see as being of greatest benefit to them.

Related to this is the increased interest in broker analytics, which show how much an order can be expected to cost if placed in a particular part of the market, and then after the event how much it actually cost, why, and whether it might have been better to have used a different algorithm or broker. This is another area we have invested in. These things are quite important to the buyside; as they take on more choice on which routes they can use, they need more help on execution so they can make better judgements next time.

We work very closely with e-trading desks; we fully understand their algorithmic and DMA trading and what their offering is about, and then we offer that in a very sophisticated way in our blotters. The key for us is to have a really close relationship with algorithmic providers. We have to understand what algorithms they're offering and reflect that in our system. Brokers and dealers are constantly offering enhanced and new algorithms, so we have to try to stay on top of what they're offering concurrent with their services.

The evolution of our company has seen us add additional functionality to our OMS, such as post trade confirmation and settlement as well as analytics.

As providers of algorithms are moving into next generation solutions, there are more options a trader needs to trigger. So a new solution from us allows the platform to store the parameters needed for one particular algorithm provider so it can be traded quickly; you can turn 20 clicks on an algorithms' parameters down to one. Additionally, you can set up automatic routing rules to automatically transmit low risk trades to an algorithm so a trader can concentrate on the orders which do represent risk and where they can add the most value.

We have to keep up with changes so our clients can take advantage of them. Offering connectivity with packages as they come along and keeping up with the evolution of functionality in these systems is key.

What integration/functionality issues do you see the adoption of multi asset algorithmic trading raising as regards the interface between the OMS and back office systems? How do you think those issues might best be resolved? And by whom?

Bernstein: I don't think there's much difference in what we see between the back office and OMS now. I don't really see multi asset class trading via algorithms causing any more problems than there already are today.

I don't see that issue at all here. When we send an order to an algorithm, we get back lots of small fills from that algorithm which we treat the same as a large block. We treat it exactly the same as we would treat anything else. The only issue is one of volume, which we are comfortable that our architecture is more than capable of handling.

For me the back office and OMS are interfaces, and they're moving targets with a never ending group of changes you have to remain concurrent with; it's a challenge. But you need to stay on top of it in the ever decreasing time windows we have to do that.

In Europe we still have - versus the US, and even with the sellside versus the buyside - some way to go with regard to FIX. But the adoption of these standards, especially in the financial integration space, makes things a lot less painful.

When you look at back office systems, not all systems support all asset types particularly in rapidly evolving areas such as derivatives. We have a host of certified interfaces and are working closely with our third party
Stephen Engdahl, Director of Product Management at Charles River Development

Stephen Engdahl, Charles River Development: " As providers of algorithms are moving into next generation solutions, there are more options a trader needs to trigger."

back office partners. We are very clear with our customers about what each back office system interface does and does not support in terms of asset class coverage, geography, and other factors.

Other vendors may say they have particular interfaces but they may not say which geographies and asset classes each interface supports, which can lead to surprises for the customer at the time of implementation. Users can see what our current and planned capabilities are with our partners, as they may want to use some of these new instruments, so knowing what our plans are and what our third party business partners are doing with us helps customers move into these new areas.

Fundamentally, the work flow is different for asset classes, especially fixed income, so providers need to have a multi asset platform to begin with. A lot of platforms may favour equities or a hedge fund type of approach, as it's a challenge to have a multi asset trading approach. We have comprehensive fixed income functionality but algorithmic trading as it stands for us is actually more of an equities function. To resolve these issues, it's more of an industry-type solution. Clients will actually resolve a lot of issues based on their changing needs. The integration with back office systems is a challenge; to begin with back offices are very different from each other and that reflects on how you have to connect up to them, but the goal is to have seamless integration between the OMS and back office. We offer our clients a back-end portfolio accounting solution as well.

Do you see the task of facilitating the integration of sellside algorithms as being primarily the responsibility of OMS vendors, the sellside, or organisations such as FPL?

Bernstein: For the most part, there is collaboration between vendors and the sellside. The sellside has a responsibility to show us how they want their algorithms presented and from our side, we have to work out how to facilitate that to our customers.
With the sellside making changes to their product offering it requires a good working relationship to ensure all parties are satisfied, the client, vendor and broker.

We're involved with the FPL and other industry bodies to ensure we have got a set of standards appropriate for this industry. But at the end of the day, the commercial imperative is on us and the sellside to provide buyside traders with what they want.
Furthermore, since we provide the services framework for our customers to get access to these algorithms, we need to be responsible for the technology integration; to involve other parties in this stage would be to slow down the process immeasurably and would reduce our control over the products we offer to our customers.

What's more, brokers create, run and promote the algorithms themselves, so we need to develop a mechanism for them to differentiate themselves. The ability to provide access to new algorithms quickly is key for us; it enables brokers to be innovative, and encourages them to work with us and incorporate their algorithms into our OMS. Standards are a way of potentially ensuring broker neutrality, but we're confident we have that anyway using FIX. There have been discussions around enabling a richer service delivery from brokers and there are some FPL working groups talking about that, but the delivery of that is between us and the brokers.

The industry as a whole shares responsibility; we're all stakeholders, from the FPL, Swift, the buy and sellside, and vendors. It's a question of the industry working together to deploy standards, as these interfaces change all the time and it's easier to keep up with them using standards.

The FPL is doing an outstanding job with regard to the sellside and now much more, the buyside. But we all need to understand the integration issues and market initiatives.

Currently it's between the sellside providers of algorithms and the OMS vendors. Across different sellside providers there's a lack of consistency. Ultimately the task of making sure the system works is down to the OMS vendors.

That role will become easier when standards are in place for providers of algorithms. The currently favoured standard that is being discussed at the FPL is the delivery of algorithms into the OMS using XML.

Over a year ago, we developed our own XML-based language so that we could more quickly and easily make new algorithms available to our customers, without requiring our customers to take new code or a new release for each algorithm. With a similar method being considered by the FPL, we are well positioned to take advantage of such a standard when it comes available.