Mapping Out the Future of SORs

DWT: How would you describe the current generation of smart order routing (SOR)?
Choël, SunGard: In Europe, the current generation of smart order routers mainly deals with displayed liquidity and these have not yet completely integrated with other algorithms that are driving trading. We think this will change dramatically in the near future given the developments in the U.S. and the progress of the underlying technology. As firms develop multi-listed trading algorithms and integrate dark liquidity pools into their smart-routing strategies while seeking the lowest latency, these issues will contribute to the birth of a new breed of smart order routers.

Bailey, smartTrade Technologies: Smart order routing is firmly entrenched in the market today. Current discussions within financial institutions focus on how to make these routers handle more complex transactions, including multi-product routing that can perform sophisticated arbitrage between products as well as provide a deeper level of customization where the router can manage the parent and child orders. The routers also should address advanced connectivity issues, such capturing/measuring market data throttles and managing venue disruptions. They need to be designed to support cross-asset trading and integrated into the organization’s trading backbone.
Multi-product routing is being driven from both a geographical and trading perspective. Inter-listed securities that can trade on multiple exchanges located in different countries have a foreign-exchange (FX) component to them; smart order routers need to be able to analyze and provide best execution not only on the security, but on the FX trade as well. Add to this scenario that the ability to leave the security on multiple exchanges at the same time—i.e. queue jumping—will require a smart order router that is “state-aware.” Arbitrage opportunities can arise in all the product groups. For example, in the auto-hedging of cash versus futures in fixed income, the router should be able execute the cash, futures and spread legs based on the criteria of the user. In the case of trading exchange-traded funds (ETFs), the router needs to price each underlying security to make the proper routing decision.

DWT: How can firms leverage their smart order routing technology into other functions, such as trade reporting or clearing?
Bailey: In order to provide for proper trade reporting and clearing information, the smart order router should be “ground zero” for all details relating to the transaction. If the router has been designed properly, it has all the granular information that is required for the reporting and can relay that information real-time. If we extend the possibilities of a well-designed router, it is possible to automate the control of the post-trade features with regards to how this can affect the trading. For example, it is possible to limit the access in trading a certain security if the clearing has failed, or if there is a mismatch signal that has been sent from the post-trade layer.

DWT: How should smart order routers be designed to scale to meet market fragmentation?
Bailey: Each new venue has its own microstructure that consists of fees, rebates, execution conditions, order types, trading protocols, latencies, and refresh rates, among other nuances specific to the individual venue. For smartTrade, we have cleanly isolated those pieces of information that are redundant among the venues so that the connection is not burdened with undue weight. For example, there are a range of accepted order types in the market; smartTrade can process these order types directly on the level of the router or emulate them as well, which is important. We have given our customers the flexibility to define their own additional order types. Even more important to our connections is the fact that we are a state-aware system, which means that we know at all times where an order is, what state it is in, if a connection is up, or if data is being throttled or refreshed. In being state-aware, our smart order router can make decisions in real time like redirecting flow when a venue goes down, reacting to throttled data immediately, and calculating the fees/rebates before executing even across multiple venues at the same time, and of course, managing passive orders dynamically.
We have found through our experience that our clients prefer that we do the development and maintenance of the connectors and give them high flexibility in defining and customizing their smart order routing rules and order types. This is where smart order routing’s real value lies for them.

Choël: Further market fragmentation and the continual rise of high-frequency trading will increase pressure on smart routing technologies to be faster and more adaptable. We expect that smart routers and algorithms will benefit from the latest progress in low-latency direct market access (DMA) infrastructure. Complex event processing (CEP) technology will help the next generation of routers become more efficient and quicker at adapting algorithms to changing liquidity dynamics, and handling higher degrees of fragmentation.
We also see much closer integration being achieved between “how” and “when” trading algorithms, which will help considerably in reducing inter-process latency.

DWT: Which new metrics are we seeing introduced into the smart order router calculations?
Choël: One current area of emphasis for many firms is integrating market data histories and adopting dynamic intraday routing logic, so one key objective for metrics is to assess the resulting gained execution improvements. This is classical transaction cost analysis (TCA), which has always been the key metric for smart routers, but the range of possibilities and the finesse of the techniques involved are increasing steadily.
The use of a smart order router means that broker-dealers have to manage a constant conflict of interests: How can they improve their  clients’ execution prices without increasing their cost of trading excessively?
It is very difficult to optimize overall trading costs and still achieve the best price for the client due to all the venues’ business models—central counterparties (CCPs) and clearers—which you have to take into account in Europe. Tracking the relationship between execution quality and trading costs is clearly a necessary guide to striking the right balance.
A full analysis of trading costs requires front-to-back measurements, including the use of dark liquidity pools and in-house liquidity. Those brokers who are able to optimize their flow management and trading costs front-to-back will be best able to deliver best execution and make it cost-effective for the investor.

Bailey: We see the addition of more sophisticated venue monitoring to smart order routers that will also need to become much more analytical in order to support multi-product trading.

DWT: What is the best way for smart routers to deal with dark and “gray” markets?
Bailey: It is always by respecting the rules of the market. An institution may have had its own unique experience with each venue and this information can be used in the smart order router as part of its decision-making process. Heat maps are key to finding liquidity in dark or gray pools. Producing the appropriate heat map depends on the trader’s strategy, but in my eyes, cannot be an off-the-shelf package. At smartTrade, we focus on our LiquidityOrchestrator smart order router to make these decisions based on multiple factors or monitored data, including complex heat maps if needed.

Choël: The optimal type and degree of use of dark liquidity is very dependent on the type of order flow involved. When executing in large scale, exploitation of dark and grey liquidity can give the advantage of limiting, or even removing, market impact. But the cost of trading on these platforms is usually considerably higher than the equivalent activity on the displayed markets. The first key issue for a smart order router is to balance these factors and then strike the right balance between what is available and known and what is potential and unknown—this is the larger challenge. The search for dark liquidity should not compromise the use of the available lit books.

DWT: How do smart order routers meet the demands for low-latency execution?
Choël: It is similar to the way that they will address market fragmentation, which I mentioned earlier.

Bailey: My sense is that if a smart order router adds less than 1 millisecond of latency, it is perfectly acceptable in most of the market cases, whatever the asset class.
However, the real trap is in the coding of the rule or execution strategy. Since complex strategies may add some additional loops of code in their execution, this could lead to additional latency because of the additional required CPU cycles needed to accomplish the strategy, especially when reproduced through the entire trading workflow. To obtain ultra-low latency, firms have to take into account collocation, transport, network and so forth.