The Gateway to Algorithmic and Automated Trading

Pathway to Performance

Published in Automated Trader Magazine Issue 26 Q3 2012

While deploying a trading model as quickly and efficiently as possible is a priority for any trader, a major obstacle to this at present is the constraint on IT resources. Particularly for traders looking to automate their trading but lacking programming skills, this is a major hurdle. But as Gary Stone, Chief Strategy Officer at Bloomberg Tradebook explains, there is another way.

Gary Stone

Gary Stone

Trade automation is a concept with many facets. To some it is all about the automation of an existing manual alpha strategy, to others it is the automation of activity associated with a strategy (such as rebalancing a hedge), while to others it is the implementation of a concept that could never be traded manually for workflow reasons. However, in all cases, the underlying question remains the same - how to get from the drawing board to real time P&L as quickly as possible? The idealised answer is simply to have a team of highly skilled programmers permanently at your beck and call who have the necessary skill set, plus the ability to understand every trading concept in your head.

Unfortunately in the current market that just isn't going to happen. IT resources are now severely limited, so that traders typically have to implement their ideas themselves - which leaves those who aren't already skilled programmers with something of a problem...

The implementation jigsaw

Apart from the immediate problem of encapsulating an idea in code, these traders also face various other challenges, which are best summarised as "How can I assemble all the various components and complete all the other tasks needed to actually get the idea deployed as a real time trading model?" This isn't a trivial task, as this trading jigsaw puzzle can include many, or all, of the following critical pieces:

• Integrating the requisite data
• Connecting to broker/market gateways
• Dealing with multiple order types and data schemas for various markets
• Backtesting and paper trading
• Writing/implementing order execution algorithms
• Porting code from the development to the production environment

The Gamma Hammer

While the automation of alpha models tends to attract more headlines, integrating execution into spreadsheet-based proprietary models can also massively boost productivity and performance when it comes to trade management. For example, the Bloomberg Tradebook Execution Architect ("the Execution Architect") has already proven its value for rebalancing hedges on open convertible arbitrage positions (though it would also add the same value for any other type of trade "housekeeping").

Historically, convert arb has been an intensively manual strategy; multiple calls to brokers, plus frequent delta hedge rebalancing (and/or neutralising position gamma to minimise delta hedging activity). Anything that automates hedging activity therefore frees up the trader to search for new trading opportunities, as well as enabling more granular and accurate hedging. This time-saving can be substantial - it is not uncommon for a convert arb trader to spend more than half of each trading session just on manual hedging.

The Execution Architect enables these traders to create an advanced "Bloomberg Experience" - integrating a proprietary model that calculates a portfolio's gamma rebalancing on the Bloomberg Professional service and integrating execution with Bloomberg Tradebook. This strategy, the "Gamma Hammer", involves simultaneously (re)setting a buy and a sell trade - either of which can be tripped by a model that calculates a delta trigger based upon the gamma of the underlying convertible bond. If a buy or sell order is triggered, the other trade is cancelled and two new buy/sell orders are issued based upon a recalculated trigger level. The crucial benefit is that the model can be fully automated for an entire portfolio of convertibles. It is therefore not subject to the physical constraints of a human trader, so the frequency (and by implication also the accuracy) of the hedge can be increased. For a trading strategy such as convertible arbitrage, where competitive pressure due to multiple new entrants has squeezed potential P&L, this is crucial because the resulting "execution alpha" can ultimately spell the difference between success and failure. Effectively, higher frequency automated hedging can adhere more closely to an idealised hedging curve than human hedging, which by contrast describes a saw tooth profile that oscillates (sometimes widely) around the perfect hedge.

An additional benefit of this integrated approach is that it can be extremely flexible. Multiple parameters and inputs can be used in the design of a proprietary model to determine the placement of trigger levels and buy/sell orders, which can be asymmetric. This can deliver an exceptional level of responsiveness to market conditions such as liquidity, overall market sentiment or even specific news events.

For those who already have the necessary skill set and who can add significant value through proprietary tweaks to any or all of the above, this isn't an issue. For those who do not, there is a pressing need for a one stop automated trading environment that covers all these bases on their behalf. But that alone is still insufficient, because while most traders can deftly manipulate a spreadsheet, they also need to learn how to automate their existing trading ideas and expedite the development of new ones of increasing sophistication.

These challenges were one of the factors behind the evolution of the Tradebook Execution Architect ("the Execution Architect"), together with its tighter integration with other Bloomberg Professional service and Bloomberg Tradebook tools and solutions.

Three routes to go-live

In conjunction with other Bloomberg services, the Execution Architect delivers a complete automated trading environment, but it does so at three distinct levels that cater for traders at all points on the programming skill spectrum.

• Spreadsheet integration:

for those with relatively simple needs, or those new to programming, the Execution Architect provides a complete "Bloomberg Experience" for users looking to integrate their spreadsheet models. Models can be created with the Bloomberg Professional service and executed via the Bloomberg Tradebook Order Builder. The Tradebook Order Builder enables users to set up tables where each row in the table represents an order ticket field to which a user's model can be applied. If the model's conditions are triggered, then the appropriate order, as specified by the user, is sent to Bloomberg Tradebook for execution. Conditions can be built using worksheet formulae, or for those comfortable coding, VBA can be used instead to program these or to create completely separate models that do not use the Order Builder.

• Scripting language: trading strategies can also be written within the Bloomberg desktop using an intuitive proprietary scripting language. Furthermore, strategies written in this way can also be backtested (see "Backtesting" below).

• .NET: soon, for users who are already competent programmers, a CS.NET framework will be available within the STDY Bloomberg function. This will offer users greater flexibility and functionality. Complex models, such as the ones that options traders encode in spreadsheets can be written in CS.NET and backtested in the core Bloomberg backtesting engine.

These three programming levels provide a smooth migration path for those either new to automated trading or who are seeking to design and implement a system before devoting expensive IT resources to building a black box. However, for those already wishing to take the final step to black box trading, it is now possible to run fully automated models via the Bloomberg Managed B-PIPE service on a remote server.

Backtesting

As part of a client's subscription to the Bloomberg Professional service, Bloomberg has developed a number of backtesting tools of varying degrees of sophistication and rigour. Traders can design and backtest their strategies on the Bloomberg Professional service and then implement a successfully backtested strategy with Bloomberg Tradebook. At the most basic end of the spectrum, the backtesting functionality these tools deliver enable traders to test the potential profitability of a new strategy - without having to revert to a back-office team of quantitative analysts or application developers. The objective is to provide traders and front-office staff with a straightforward tool for creating real-world simulations that they can apply to their trading algorithms. The backtesting functionality includes a highly robust strategy template, as well as providing a broad selection of P&L and risk factor analytics. The process can also support models that include money/position management strategies such as pyramiding. Once backtesting is complete, the user can easily migrate the model to Bloomberg Tradebook's Execution Architect to bring the model into a production execution environment.

All in one

While electronic trading enables strategies that could not be implemented in a manual environment, many traders also find themselves confronting a different roadblock. Yes, they can automate their trading, but because of competitive pressures they now need to automate it simultaneously across multiple markets and security types. As inter-market strategies become more complex, this often begets further complexity as traders find themselves struggling with multiple data schemas. For example, code written for the equities leg of a strategy may well not be immediately portable to the futures leg of the same overall strategy, but will require tweaking to conform to a different data schema.

In an environment where there is intense pressure to deploy trading models as quickly as possible, this is unacceptable. For this reason, the Execution Architect has consolidated the various security types to three data schemas (Equities, Derivatives, and Pairs) so as to streamline the implementation process for traders by enabling "write once, run many". In a further streamlining step, the number of events within the common schema is also being reduced and the service upon which it runs made multithreaded to remove the risk of bottlenecks occurring in high activity conditions. This rationalisation has a benefit at an ergonomic as well as a programming level. Traders managing multiple automated strategies that traverse multiple security types and markets do not need multiple screens to monitor individual trade legs. Furthermore, the vast majority of Bloomberg Tradebook's execution algorithms and tools (such as pairs trading) that are available to manual traders using Tradebook are also accessible via the Execution Architect to automated traders.

Maximising the opportunity

The Execution Architect steers traders away from the dirt road trade execution they often encounter on other platforms. Nevertheless, making the most of the new superhighway is inevitably easier given the right support, which is provided by Tradebook execution specialists. In addition to generic benefits, this can also save clients from drawing erroneous conclusions about new automated trading models.

A classic example of this is where the real time performance of a trading model is significantly below that of its backtest results. In this situation, it is easy to conclude that the model was over fitted during development and is therefore unusable for live trading. However, this situation can also arise when the underlying assumptions of the model are perfectly valid but it falls at the final hurdle because an inappropriate execution strategy is being used. There have in fact been a number of such situations recently where Tradebook execution specialists have worked with clients to refine the execution strategy so that live model performance has been boosted from sub to super optimal in comparison with the backtest.

Conclusion

Whether a trader is moving from a manual environment, looking to add an execution component to their hedging, or streamlining/enhancing the development of a complete portfolio of proprietary models, the challenge remains the same - how to prevent the development process impeding the business? Creating an integrated "Bloomberg Experience" across multiple skill sets and levels of complexity accomplishes this, so the user can focus on generating alpha - not building the infrastructure.

Bloomberg Tradebook LLC member of FINRA (www.finra.org)/ SIPC/NFA. Bloomberg Tradebook do Brasil is the representative of Bloomberg Tradebook LLC in Brazil registered with the BACEN. Bloomberg Tradebook Services LLC, Bloomberg Tradebook Australia PTY LTD ABN 36 091 542 077 ACN 091 542 077, Bloomberg Tradebook Do Brasil LTDA., Bloomberg Tradebook Canada Company Member of CIPF, Bloomberg Tradebook Limited, Bloomberg Tradebook Hong Kong Limited the first ATS authorized by the SFC-AFU 977, Bloomberg Tradebook Japan Limited member of JSDA/JIPF, Bloomberg Trading Services Japan LTD, Bloomberg Tradebook Singapore Pte Ltd Company No. 200104338R, Bloomberg Trading Services (Singapore) Pte Ltd Company No. 200101232G. Bloomberg Tradebook Bermuda LTD, licensed to conduct Investment Business by the Bermuda Monetary Authority.

BLOOMBERG, BLOOMBERG PROFESSIONAL, BLOOMBERG TRADEBOOK are trademarks and service marks of Bloomberg Finance L.P.("BFLP"), a Delaware limited partnership, or its subsidiaries. Bloomberg Tradebook is provided by a BLP subsidiary, Bloomberg Tradebook LLC, and its affiliates and is available on the BPS.