Adopting Intelligent Automation
Most any buy-side executive would like for their trading desk to be fully automated, with the most cutting-edge artificial intelligence baked in to that automation. The trick is getting there, as moving from just plain automation to intelligent automation is no small task.
For one, transaction cost analysis — the generally accepted industry scorecard for trading efficiency — must be rethought.
“You need to understand what it is you’re trying to optimize,” said Henri Waelbroeck, director of research at trading-technology provider Portware, a FactSet company. “You cannot use TCA benchmarks as they exist today as an optimization objective for an automated workflow. Doing this would likely hurt rather than help portfolio performance, because TCA benchmarks are typically endogenous: they’re formed in part through your own trading decisions, such as how many shares you trade today versus tomorrow or where you set a limit, and the impact you have on the market.”
“Current TCA methodologies don’t adequately take into account opportunity costs or selection bias that results when traders use discretion on the number of shares they fill each day, for example,” Waelbroeck continued. “These may seem like technical issues but they’re not technical at all — you simply cannot have a quantitative optimization framework if you don’t know beforehand what it is you’re trying to optimize.”
Implementing intelligent automation entails two steps: understanding its capabilities, and identifying where to apply it. That’s according to Ajay Bhutoria, vice president of banking and financial services at consultancy Cognizant.
“It is the introduction of logic which allows these programs to make decisions autonomously, or on their own, when they encounter exceptions or other variances in the processes they execute,” Bhutoria explained. “Intelligent automation systems are also characterized by their ability to analyze vast amounts of dynamic and unstructured input, as well as to execute processes that are highly dynamic and non-rules-based.”
From there, “for many organizations, the processes that benefit most from intelligent automation are those where these smart technologies and humans interact together – each working on the part they do best,” Bhutoria continued. “In the business world, imagine intelligent automation running in tandem with work conducted by research and development teams, sales organizations, manufacturing and logistics operations, or customer service departments. The impact – on everything from financial trading systems, to real-time pricing engines, to patient care, to completely individualized insurance programs — is enormous and is just beginning to be recognized by early adopters.”
Large, white-shoe institutional investors are conservative and risk-averse by nature, and as such they typically aren’t first movers when it comes to technology. But as more firms implement intelligent automation and optimize processes in doing so, the methodology presumably will move toward being a sort of standard of care, necessary to keep up with competitors.
An investment firm can test the IA waters before jumping in. “Rather than looking at wholesale system changes, process re-engineering, or complex studies, companies are realizing that intelligent automation can be tried, tested and scaled in very short cycles,” Bhutoria said.
In a broad sense, “the promise of intelligent automation is real, and it’s here now,” Bhutoria concluded. “Organizations that get started on the automation journey will soon experience the benefits of process acceleration, greater efficiency, quality gains and optimized work teams and begin improving outcomes like never before.”
Buy- and sell-side traders need to process more data than ever before.
Nasdaq CTO assesses how cloud, data, and AI will impact market operators.
It was once about nudging traders to the mouse; now it’s moving them away from it.
Adaptability becomes the hallmark of the latest generation of routers.
GMAG says vacation season is a good time for design, implementation and testing of machine-based trading.