Behind The Wheel
The algo wheel is a major step towards establishing an optimal workflow, making trade execution faster, systematic and accountable.
Mention the concept of “algo wheel” to a buy-side trader and visions of a passenger sitting in a driverless car might ensue. Far from that, its implementation at State Street Global Advisors proves that the buy-side trader remains “the one behind the wheel”.
If you grew up in the 80’s, this unsubtle Depeche Mode reference has probably not escaped you. “Depeche Mode” is pretty much how you could describe the hour preceding the market open where hundreds of orders must find their way to the auction. Removing trader bias and freeing up time to work difficult orders are the primary advantages of the process. A few clicks and voilà, 90% of the pad is at work in the most optimal conditions as objectively possible.
Even the most proficient algo users have a tendency to favour certain electronic destinations and select algorithms they feel most comfortable with, using settings they estimate optimal to work a given benchmark. Give a trade with an “arrival price” benchmark to 10 buy-side dealers and you’re likely to see 10 different outcomes. Yet, just like picking the wrong cash desk to route your flow deprives you of price-improvement opportunities, selecting the wrong algorithm won’t deliver the best execution.
As the use of electronic trading concentrates more risk on the shoulders of a buy-side trader, they will naturally favour destinations where sales trading support is most reactive and provides a “cushion” of safety as a second pair of eyes. Most orders though, certainly from large passive funds, require little to no oversight from brokers and therefore the wheel is a solution to optimize flow distribution on a scalable and efficient manner.
In practice, what is the wheel and what does it do?
Say you have 5,000 orders waiting in your incoming blotter in the morning, differentiated by benchmark and average daily volume. Select anything under a certain liquidity threshold and, rather than picking a broker manually, elect for each benchmark the wheel as destination in the execution management system and it will automatically dispatch the orders to brokers, based on a ranking.
The ranking is established according to previous performance. How many brokers compete in each benchmark, what percentage of the business should go to each is up to the desk to decide. This ranking process is not static. As brokers compete to improve their performance, after a certain period of time, again set by each desk, a review will be done and a new ranking will be implemented.
Another advantage of the process is to spur competition and innovation on the broker side. Each client benchmark and each market get a specific attention. One strategy setting will not fit all markets but also not all market conditions. Price prediction models, stock mapping, historical data versus real-time triggers, and more are an ever-evolving source for quantitative research.
Sales-traders are also encouraged through this process to pay closer attention to outlying orders, being part of a package expected by the client who can easily cut a broker’s wheel target if expectations aren’t met. Free from research-based commission models under Markets in Financial Instruments Directive II, attracting pure trading commissions is the new endgame and dropping off a wheel “bucket” (a benchmark-specific ranking) means a substantial loss of income.
Trends among brokers in each benchmark category eventually emerge overtime while improvements in settings or trading logics can finally be quantified. The process arms buy-side traders with an unbiased ranking of which algorithm is best candidate to trade that “tail” of less liquid orders, which typically would not be routed through the wheel. The model gradually shifts the electronic broker role to an execution consultancy one and both parties can learn as the wheel is not intended to be a static black box but rather an ever-evolving outperformance think-tank.
Keeping data uncorrupted is paramount. Events such as large benchmark rebalances, while benefiting from the wheel ranking to direct the massive flow, are best left out of the periodical re-evaluation process as they can represent a substantial share of total turnover with such wide price swings that they would trump the knowledge gained from day-to-day data analysis.
Similarly, removing “outliers”, those unusual trades that weight heavily in the balance, helps yield more consistent and meaningful performance figures. Building up a statistically meaningful database is the cornerstone of the wheel deployment and can prove a challenge for smaller operations.
Can your traders explain why a trade was executed through a certain broker in a certain way to the firm’s stakeholders or upon a regulator’s enquiry?
The wheel brings traceable accountability into the process. Can we envision a constantly re-optimized wheel capable of handling all orders, from the most to the less liquid ones? There is little reason why we could not and already some desks have implemented a straight-to-market workflow that bypasses any trader input provided that a set of conditions are met. Even the interaction with liquidity providers’ flow, such as Liquidnet, could be hardcoded and automated.
On the heels of the worldwide implementation of algorithmic trading, the wheel is a landmark step towards establishing an optimal workflow, not only in terms of speed, but also in terms of improving accountability to stakeholders by systemizing best execution. It spurs innovation and strengthens the framework within which buy-side desks operate.
So, take control of your flow and embrace this milestone evolution in global trading to make a difference for your clients.