Five Pillars of Modern Electronic Trading
By Kathryn Zhao, Global Head of Electronic Trading, Cantor Fitzgerald
The electronic trading business is highly competitive. In my view, to build a competitive low-touch offering, one needs to, at least, tackle the following five key areas.
Over the last 20 years, the electronic trading technology landscape has experienced dramatic changes. Innovations in hardware, networking, and software have had an immense impact on the current state of the art. To truly stand out from the rest of the competition, an electronic trading product must perform from a latency and throughput point of view and be resilient to failure without compromising capabilities and quality of execution.
There is a prevalence of cutting-edge open-source libraries for almost everything that needs to be built. A tremendous amount of attention should be paid to reliability and failover. A comprehensive suite of automated testing and a sophisticated simulation environment is essential to validate and guarantee the quality of a product.
Buy-side firms clearly demand better performance and predictability of execution results, without sacrificing the ability to source liquidity, transparency and control of their executions. Connecting to a wide range of liquidity sources, selecting venues to route algorithmic child orders based on advanced analytics, regularly reviewing venue selection, and providing transparency of order routing logic are all integral parts of best execution and drivers of quality execution performance.
Advanced quantitative models are the cornerstone of a competitive electronic trading offering. It is the “brain” that drives the algorithmic behavior. Taking into account both historical and real-time market data, a limit order placement model determines when, where and how much quantity to place at any point in time throughout the order lifecycle, a venue-ranking model makes informed stock-specific routing decisions, and a volume profile model dynamically tilts towards either a front-loaded or back-loaded distribution of volume.
One should choose the most suitable models for every situation. We treat our order placement problem as a Markov Decision Process solved by dynamic programming technique. This provides a mathematical framework that can naturally integrate various sub-models to address different aspects of security trading, in particular, market microstructure models, such as spread dynamics, fill probability and adverse selection, all modeled with thorough statistical analysis. Short-term signals can also be easily incorporated into the framework.
At every state, an optimal decision among crossing the spread, improving the best quote, joining the best quote, placing deeper in the order book, and staying away from the market will be selected in order to minimize trading cost.
The use of market microstructure models and dynamic programming technique makes a number of well-known optimal decision concepts become the natural outcome of an order placement model in a quantitative manner:
1. Opportunistically cross the spread when the spread is tight or the market is moving in the same direction as
the algo trading activity.
2. Remove the order from the market if there is significant adverse selection or the market is moving in the opposite direction as the algo trading activity.
3. Step into the spread to improve fill probability while still capturing the spread.
FIX connectivity/Vendor certification/Client onboarding is one of the most crucial elements to get right in the electronic trading business. Seamless certification reduces the number of issues down the road, and proper set up of FIX connectivity and tag translations creates a positive client experience, with a reduced number of rejects and erroneous algorithmic behavior. In addition, expedited response to FIX inquiries enables electronic sales and coverage to respond to clients’ demands in a timely manner.
No two buy-sides’ needs are exactly the same. Modern, competitive electronic trading offerings are required to deliver customized solutions tailored to each client’s specific trading preferences.
With growing sophistication of quantitative models and ever-increasing electronic solution varieties, buy-side firms face a unique challenge: how to choose a product that is straightforward to use and yet easy to customize.
The first question one should ask when evaluating a provider’s electronic offering is whether it has the capability to quickly customize its product to meet all the requirements and what level of customization can be achieved. There are roughly three types of customizations:
1. Customize within a particular algorithm and adjust parameters and behaviors based on price, time, or other market conditions;
2. Customize across algorithms, i.e., switching between different algorithms based on certain signals or market conditions; and
3. Bespoke algorithmic offering based on specifications from buy-side firms.
Customization needs to be done expediently to enable the buy-side to take advantage of market dynamics as they occur.
On top of that, it is necessary to build a flexible experimentation framework that allows conducting A/B (randomized test running two versions side-by-side) test and an advanced analytics framework that enables generation of performance statistics easily. Only doing so will ensure having a complete circle of capabilities for a modern customization framework.
Electronic trading is a “low-touch” business with “high-touch” service. The most competitive service model would be a combined approach of low-touch speed to market and high-touch level of consultancy. As the number of providers, algorithms, and venues grow, it is increasingly important for sell-side firms to provide a high level of customer service to their electronic trading clients. In order to deliver “high-touch” service, one needs to have proper sales trading tools, capable FIX support personnel, and dedicated algorithmic support staff to answer buy-sides’ inquiries on the fly.
What is “high-touch” service?
• Stay on top of trade blotter and watch orders closely: orders that are not executing, orders that are falling behind the schedule, performance metrics (slippage against Arrival Price, Interval VWAP, etc.), orders that are causing impact, orders that are getting poor fill rates, stocks that are on the move, etc.
• Provide prompt responses to Instant Bloomberg messages and client inquiries.
• Most importantly, catch things that clients care about (even if it is not good news) and bring issues to their attention.
• Provide actionable suggestions.
An algorithmic product is only as good as the people behind it. “High-touch” service means the algorithmic coverage is clients’ eyes and ears, and provides them with value-added information in a frequency acceptable to them.
Enabling/papering the client is step one, getting the client to trade is step two, retaining and increasing the clients’ business is step three. Simply getting the client to send an order on auto pilot mode is not something that will retain clients, and more importantly, increase the clients’ flow.
There is a need to standardise ESG data.
Solve Advisors says demand for data is growing as fixed income markets become more electronic.
The UK regulator said ETF primary markets are highly concentrated, particularly for fixed income.
System helps traders manage data agreements electronically.
Using machine learning in securities lending business enables greater pricing efficiency.