Embracing the Numbers: Analytics to Improve Trader Performance

The game of baseball has been played for over 100 years so its rules are more than a century old. Analytics in the sport have historically revolved around batting averages, numbers of home runs hit, and a pitcher’s win/loss record. Then the Oakland Athletics broke new ground when the team took a deeper dive into the numbers to produce a winning organization on a shoestring budget. Their strategy crystalized in 2003 in Michael Lewis’ book—and eponymous film—Moneyball: The Art of Winning an Unfair Game.

Fast forward to 2019 when almost every Major League Baseball team embraces analytics—also known as sabermetrics—to some extent. The data generated during games can be analyzed to help a player better see their strengths and weaknesses, and help the club optimize its roster.

Trusting analytics wasn’t an easy sell to baseball traditionalists because it is a science and ignores gut feelings and instincts. If the analytics suggests a course of action that contradicts a manager’s instincts, the question becomes: listen to your gut, or trust the numbers? Similarly, trading has traditionally relied on gut feelings and instincts, but everyone has specific strengths and weaknesses.

Financial institutions use a myriad of technologies to trade and manage a portfolio. Inside those actions are metrics that can be analyzed and used to improve how an individual trader or portfolio manager accomplish their job.

This is a new idea for many financial institutions, but a few vendors and asset managers have begun to take advantage of technologies available, such as behavioral analytics, user-experience processing, and machine

learning. These technologies create a performance baseline to follow or exceed, and offer guidance as to where to focus decision-making.

A Decision Journal

The need to compete with algorithms when picking stocks was what drew Cambridge Global Asset Management, which manages $20 billion in assets, to using data and behavioral analytics to figure out if their portfolio managers are making the best decisions.

Brandon Snow, chief investment officer at the Toronto-based asset manager, says the environment in which humans now have to compete has significantly changed, so it’s important for them to have the skills that will keep them competitive in the long term.

“I think our ability to stock pick is still a competitive advantage. So we wanted a third-party view of what we do, to see where value is being added and what decisions are driving those,” Snow says. “There is a shrinking market, so it’s more competitive out there, but if you start early in the process of recording your decisions and improving on them, you can be years ahead of your competition.”

Snow says if his portfolio managers can understand how they make their decisions, they develop a better grasp of their abilities, lean into their strengths, and have insight into their instincts. This way, they don’t have to be worried all the time about competing against algorithms.

Cambridge tapped data analytics firm Essentia Analytics to teach it to optimize the decision-making processes. The London-based vendor gathers data from individual traders—starting with their trade and holdings

data—and analyzes it for behavioral patterns. These patterns can show where and when a trader or portfolio manager is more comfortable making a decision. For example, a manager may have more confidence in entering a trade when the market is on the downswing, and Essentia advises the trader on what they’re better at leaning into, and what they should avoid doing.

Clare Flynn Levy, founder and CEO of Essentia, compares traders to athletes who want to constantly improve. And one of the ways athletes have found to become better is look at the data they themselves generate and analyze it to find patterns.

Levy notes that buy-side firms have taken on criticism in recent years, as passive investments have been consistently outperforming active fund managers. It’s hard to justify a fee if you’re getting beaten by an index fund.

“It’s a well-trod story, but at the end of the day we believe that some fund managers can continuously or consistently outperform the index. ... But are you at the level of skill or competition as a human that you need to compete?” she says.

Levy says Essentia coaches clients to lean on these habits or to get them more comfortable with different market movements. She adds that they employ “nudges,” which are questions tailored to the individual manager during critical decision-making moments.

Essentia mainly focuses on equity fund managers, but does work with companies with other strategies such as small caps, emerging markets, and environmental, social and governance (ESG) investing. They currently work with portfolio managers at 30 firms around the world.

Cambridge’s Snow points out that it isn’t enough to have a record of your decisions—he personally keeps a decision journal—because most people don’t have the insight into the impact their decision could have had. He says he wanted to understand the connection between his own investing decisions and its portfolio impact and see if he could have done something different.

“Inherent biases convince you to keep making bad decisions because you don’t know they’re bad decisions. But if it’s tied to your portfolio, you’ll see its impact better,” Snow says. “A data scientist can take my decisions and interpret what their impact was.”

This inability to self-reflect was exactly the challenge Cambridge experienced. Snow says there was—and still is—reluctance from other portfolio managers to submit their decisions to a third party. He was the lone guinea pig for a year before he convinced his entire team to work with Essentia. Even then, the rest of the firm remains wary. Snow notes that it is hard for many traders, who often say they rely a lot on their gut and market knowledge to make investment decisions, so he understands why a lot of people may not want to offer up their decisions to data analysts.

“Humans, in general, have documented behavioral biases about the decisions they make about money and what we see in our clients’ behaviors is very reminiscent of that,” Levy says. “People have a tendency to hold on to losing positions for too long; we see that all the time in the data. But it’s not until you actually see yourself in the mirror doing that that you say, ‘OK, how do I fix that?’ There’s something about seeing it in your own data that is actually key to change.”

Workflow Insights

It isn’t just behavioral analytics that asset management firms are turning to. Some banks are experimenting with using their workflow to figure out what an ideal trader process looks like. Understanding how people use applications, including which applications they open first, how long they stay, how many mouse clicks it takes them to get information before setting up a trade, offers insights into their processes. This allows companies to create a benchmark of workflow behaviors others can follow.

Desktop integration firm Glue42 offers what it calls user-experience process mining. Process mining lets managers look at each of their traders’ workflows and advises them about what they could be doing better.

James Wooster, COO of Glue42, says understanding the detailed process in which traders begin their trades also lets companies measure what

stronger traders are doing compared to those who are lagging.

“Where it gets even more exciting is where you happen to know if someone is doing something above and beyond and you want to understand what behaviors they’re exhibiting that are not present in others,” Wooster says. “So we can therefore do a baseline of you and a baseline of either another individual or a group of individuals, and then compare the flow between applications. From that we can start to see what it is that you uniquely do that others cannot or don’t do.”

Companies install Glue42’s platform and can look at how their traders use each application running on the background of its interoperability system. The platform can tell when someone’s opened up an application like FactSet, see if it remains in front of other windows or if it is thrown to the back, how long they spend using it and even employs mouse tracking to see how many times they’ve clicked on a button in an application. Once this information is available, the company can then begin to compare individual application interface behavior and possibly set up some best practices for others to follow.

Wooster notes looking at how traders use applications already helps with determining if people are following best execution and for him, best execution ties into the idea of setting up best practices to become a better and more efficient trader.

Glue42’s biggest and first client for UX Process Mining is JP Morgan Wealth Management, which Wooster says has deployed the platform to

15,000 desks. JP Morgan Wealth Management declined to talk about how it is using Glue42’s product.

Wooster says getting insight into how people use applications in tandem also allows operational leaders within a team to make sure everyone is performing up to standards. And if they aren’t, they can quantify the hows and whys of this non-performance.

“The problem is that human beings are sometimes hell-bent on breaking processes, are doing things on their own, and are just not taking kindly advice,” Wooster says. “So even with the best applications in the world where you’ve created this beautiful desktop and patchwork of applications that the traders need, the question is, what are they actually doing? Are they following the correct procedures in terms of doing all of their data research before speaking to a customer? Are they recording the call notes at the time that they are speaking to their clients and not afterwards? That’s where this user behavior analysis comes in to play and UX process mining also comes in.”

Repurposing

Asset managers are also looking at how they can repurpose compliance monitoring platforms using analytics to help improve performance. One such way is using machine learning to look for conversation patterns in communications channels.

Most financial firms monitor the communications of portfolio managers and brokers to prevent fraud and mitigate risk. This same technology is now being used to look at successful sales patterns.

Lee Garf, general manager of communications and financial market compliance at Nice Actimize, says the same technology that searches for abnormal patterns in risk mitigation can also become a means to find or generate alpha opportunities.

“There’s a growing trend for companies to say, ‘I’ve got this interesting data; is there something we can do with that data to make it not just a cost center and a risk reduction, but can we generate alpha or new opportunities based on this data?” Garf says. “Some of the things that we’re looking for are correlating the type of communication channel to the result—does voice or WhatsApp or chat generate better results for the company. The other thing we can look at is what the mix between social conversation and business conversation by better performers.”

He adds the platform can figure out if a trader or a broker is more successful in buying or selling if they introduce more banter into their conversations.

Take, for example, a broker selling to an important client. His company would like to know if he is more likely to generate additional revenue or commissions if the product is pitched through a call versus an email versus a chat. They can also figure out if, for this particular client, it’s better to build a more social rapport or not.

“A lot of the technologies that we use for compliance, or for a risk use case, were things around communications using natural language processing and machine learning to identify patterns. The patterns, in that case, are looking for anomalous behavior. Using those same techniques we have built models that look for this other scenario. We’re looking at anomalies or patterns that can help them improve,” Garf says.

Nice Actimize uses a mixture of machine learning and natural language processing to parse through communications data clients provide them and find those patterns. NLP identifies the tone and sentiment of calls or texts and figures out a percentage of the communication that was devoted to work and non-work conversations. Garf notes that early adopter clients want to see if there’s a trend showing how important the social piece of a transaction call is in settling a deal.

Most firms using these technologies are still the early adopters. As passive investment increases, active managers have had to find a way to compete and justify the fees they charge. While it is challenging to go toe to toe with a machine, humans can at least make incremental improvements to their performance thanks to new technologies.

And either way, it doesn’t hurt for firms to look at how their employees are using technology and make sure it is actually aiding them in the quest for alpha rather than hindering them.

“If you are playing with your company’s money or your client’s money, you better make sure you got exactly the right data source and applications and have no downtime or dead spots in the journey to raise a trade or to acquire some assets,” Glue42’s Wooster says. “When the process has to have a human being involved, you have to ask, ‘How do I get the insight in order to optimize things?’ and technology is the answer.”

Originally published on WatersTechnology, Oct. 2019 https://www.waterstechnology.com/trading-tools/4584781/embracing-the-numbers-analytics-to-improve-trader-performance