All market participants have felt their margins tighten in the post-crisis regulatory squeeze. As the competition for liquidity from traditional providers has grown, agency algorithms have emerged as a remedy for the increasing inefficiencies and diminished liquidity in the fixed income markets. Not only do these sophisticated algos smartly work orders, but they also open the way for valuable transaction cost analysis.
|Jonty Field, Head of EMEA, Quantitative Brokers|
Quantitative Brokers (QB) is a leading provider of algorithms to the futures and fixed income markets, which will soon be available on the TT platform. We sat down with Jonty Field, Head of EMEA, to better understand QB’s unique approach to agency algorithms and what they have to offer users on TT.
TT: Let’s start at the beginning. How did QB get started?
Jonty: QB was founded in 2008 during the financial crisis. Robert Almgren and Christian Hauff, both at a large sell side institution at the time, realized that fixed income traders were lacking the quantitative execution and cost measurement tools that were so valuable in equity markets. It would have been difficult to build these tools within a large bank, because the challenge of providing everything to everyone coupled with the requirement that it be distributed across different desks and regions, resulted in numerous, complex obstacles.
TT: You mentioned how Robert and Christian saw the industry need to focus on fixed income. Can you tell me more about the needs QB set out to address?
Jonty: QB was really looking to design algorithms that would address the particular features and behaviors of fixed income markets that were not represented in traditional equity algorithms.
There’s a key difference between cash equity trading and fixed income. In cash equity trading, you have one individual stock to execute on, but in fixed income, you have a term structure that is interrelated or structurally related. Therefore, there is a need to focus on these structural characteristics, the impacts of economic announcements, and the actual mechanisms of how exchanges fill incoming orders.
Because these behaviors are not represented in traditional equity markets, firms executing in fixed income with equity-centric algorithms were missing important characteristics and a real understanding of the specifics for these markets. Robert and Christian thought it was important to move away from the cut-and-copy, plug-and-play model to something that would really concentrate on them.
TT: As leaders in agency algorithmic execution, can you tell our readers about how QB will fit into their day-to-day trading and long-term strategies?
Jonty: Our position in the market is unusual in that we are independent. We act as an extension of the trader’s own execution. We will work with the trader’s existing execution relationships to operate inside their execution framework and workflow—both in terms of how they stage and execute orders.
With regard to longer-term strategy, we work with clients to really understand what their objectives are. Performance is only achieved through a clear and accurate understanding of the benchmark. We sit with our clients to listen to and understand their objectives and work together to identify the benchmark that maximises their performance. With this benchmark in hand, QB can suggest which algo is optimal for their specific requirements.
We also work closely with clients on their new strategies, these may be particularly cost sensitive or perhaps complex to execute. By leveraging QB’s core expertise and infrastructure clients can test ideas that would otherwise be too costly and time consuming to explore.
TT: To elaborate on your first point, Jonty, what is unique about independent agency algorithms?
Jonty: The process of developing algorithms is expensive and requires many components such as infrastructure, data and domain knowledge. The benefit of having an independent algorithm provider is that a client is not required to use a given counterparty to access those algorithms, and can also save on the cost of proprietary development.
In addition, our core skillset and primary focus is developing algorithms, which allows us to do our jobs well and develop our business to the ultimate benefit of our clients.
TT: So, how does QB maintain and develop their algorithms? I can’t imagine that a “one-algo-fits all” approach meets the complex needs of today’s traders.
Jonty: Great question. QB doesn’t design algorithms as much as we design strategies for addressing certain objectives and all of the details about how to manage the distinctive characteristics of the market—including market structure, timing, trade sizing, the existence of short-term directional signals and how best to manage scheduled economic announcements. These details are really a part of the algorithm; you choose your objective and the algorithm follows. And when you really dig down into what your goal is from a trading perspective, there aren’t that many different objectives.
In terms of maintaining and developing algorithms, we have built a real-time simulation environment where we actually run and create the order book of all the exchanges we run on. We can regression test an idea by introducing the logic into the algorithm to see how that behavior would perform historically, and equally, in real time, we take all of our production orders and run them against our simulator to verify the quality of that simulation engine is at the right standard. This gives us the ability to test and evaluate any changes that we make against a good cross section of markets, which is particularly important to ensure the algorithms are going to work well in different conditions and markets.
One thing that particularly impressed me when I joined QB is that, at the start of every day, QB automatically sends a random subsection of orders to the real-time simulator across different markets and strategies to see how those strategies are performing. Monitoring and evaluating performance in real-time is essential for performance improvements and quality control.
TT: Can you tell me about some of the paramount algorithms that QB has designed to meet trader’s various objectives?
Jonty: I’ll touch on three algorithms—the BOLT algorithm, the STROBE algorithm and our multi-legged trading algorithm, LEGGER. The BOLT algorithm has been designed and optimized to focus on the arrival price objective, which is generally appropriate for clients wishing to preserve directional alpha.
On the other hand, if the objective is to get an average price over a window, a client can use our STROBE algorithm, which can be TWAP or VWAP-based.
With regards to LEGGER, a client may have multiple legs of an order where they have a synthetic target price they want to achieve. The LEGGER algorithm has two benchmarks, firstly the target price for traders who want to achieve a specific level and secondly the arrival price where traders wish to minimize the cost of execution relative to midpoint of the structure at the start of the order.
TT: Finally, what insight can you provide into the future of algorithmic execution?
Jonty: I think the future of algorithmic execution is twofold. First is for firms to be more comfortable using an algorithm for larger trades and larger parts of the execution, which is predicated on the ability to understand and visualize how the algorithm is behaving. In turn, I think we’ll see a lot more transparency with regard to transaction cost analysis representing where the value is coming from.
Secondly, there is greater competition and benchmarking. We have seen algorithms developed significantly in cash equities, and we’ve seen the entrance of independent providers, such as QB, in the fixed income space, but I think we will see more progress in measuring and evaluating performance in the future.
Putting these two trends together, the future is in ensuring that there is no disintermediation between the trader and the market. Algorithms certainly add value and improve execution, but it’s important to ensure substantial linkage between what’s going on in the markets, the color you get from that, and the underlying trading entities that are actually taking advantage of their alpha.
The key for us in providing and developing agency algorithms is to ensure that that transfer of insight is still held. The tradition is that if you call up a broker or trader in the market, they would give you the color in the market. How do you ensure that color is still available through an algorithmic execution venue? That’s where QB is really taking the lead, both in terms of consulting with clients and visualization of our trading to explain what we’re doing and how we’re doing it.