This professor wants to help Wall Streeters understand their emotions
Brian Uzzi, Kellogg School of Management professor, wants to change that. He has developed a system that could alert traders or their respective companies of their emotional states- through analyzing their emails and instant messages in real time.
"There's a whole body of work in psychology that says based on the language that we use to describe our situation, we're revealing our underlying emotional state," Uzzi told Business Insider. "I could describe today's stock market as very changeable, tumultuous, volatile ... all these words are basically describing the same underlying concept but reveal the emotions I have in relation to that trade."
This product, dubbed SYNC, is largely built on research he's done earlier this year with Bin Liu and Ramesh Govindan. Together, they sifted through 1.2 million instant messages among day traders over a two-year period, and found that traders who expressed too little or too much emotion made relatively poor trades.
The sweet spot
Emotions color the way we view and interpret fact. They also impact our ability to take risks once we make a decision on a rational ground.
Uzzi and his team found an interesting relationship between stock traders' emotions and profitability of their trades. Using a dictionary to assess the level of emotional activation in the day traders' messages, they concluded that the best trades came when people hit the medium level of emotional activation - traders expressing neither very little emotion, nor an overly large amount of feelings.
"That kept them clear headed enough but at the same time allowed them to react appropriately to risks," he said.
'A killer man + machine partnership'
Since the start of his research, Uzzi has given two hedge funds part of his findings in exchange for use of their data to find patterns.
Uzzi said his product is ready for wider adoption and is unobtrusive in its use. This is because financial firms already have their own archiving system of email exchanges and instant messages, and Uzzi's algorithm looking for emotional states in those messages can simply take advantage of that existing archive.
"We can pinpoint what contacts a trader or analyst might have in their network that tend to provoke more or less of their emotions," he said. This helps them assess their relationships with their colleagues, and arrive at the right emotional activation.
Those real-time analytics can then sent directly to the traders or head office of a trading floor, allowing the higher-ups to assess if they should intervene in the market or not. They could also match emotional level and the timing of each trade to further explore the correlation between emotions and effectiveness, and possibly even integrate their findings into more automated forms of trading.
"What our system does it to make unstructured text data that's coming from human communication, quantify it, then adapt your algorithm that's used for machine trading to incorporate the new quant data that you didn't have before," he said. "So you get this powerful man and machine partnership in trading."
While Uzzi's product involves the collection of sensitive data, he doesn't see privacy as a problem. The professor said he's not going to resell it to somebody else, nor create a firm on his own.
"We wouldn't have the competencies anyway," he said.