The Proper Characteristics of Data Scientists and How to Use Them

Why domain knowledge matters when it comes to firms' data scientists.

ia-science-scientist-cut

There is nothing quite like a good debate between two panelists at a conference.

Our conference producers do a great job of assembling some of the best and brightest in financial technology to speak on a variety of subjects, and whenever we can get a few of them to go back and forth on a topic they disagree on, it usually leads to some great insight.

That was the case last week on an algorithms and analytics panel at the North American Trading Architecture Summit in New York. A few of the panelists discussed the merits of whether or not it was critical for data scientists to have domain knowledge.

I've already written about the panel, but to give you a brief summary of how it all went down: Gary Bhattacharjee of Morgan Stanley Wealth Management said data scientists should have domain knowledge; Ron Papka, formerly of Citi, disagreed, saying sometimes he feels it's better if a data scientist doesn't have any domain knowledge.

Know the Business

I'll admit, after hearing the two go back and forth, I tend to side with Bhattacharjee. His anecdote about data scientists coming up with a hypothesis that the firm should promote everyone to managing director was a perfect example of why everyone, even those in technology, should have at least a slight understanding of the business they're solving problems for.

I've written about this before, but I've had the opportunity to interview several CTOs and CIOs about their career paths. Not all of them had master's degrees in business administration, but they did all spend a period of their careers learning the financial markets, whether it was spending time in a specific vertical or going back to school.

I even posed the question to a CTO at a technology vendor I met with Tuesday: How important is domain knowledge to a data scientist? He sided with Bhattacharjee, citing a similar anecdote about data scientists who didn't understand the markets coming up with a solution for a problem that was in no way practical.

That's not to say I don't see validity in what Papka said. I do believe bringing in a person with a fresh perspective that isn't biased toward certain processes is helpful. More often than not, it's a fresh set of eyes that hasn't been tainted by traditional methods that can be the most innovative.

However, I think you always need at least someone on staff with that domain knowledge who has the ability to make sure the solution is practical.

Getting Work Done

One part of the panel I did find interesting that I wasn't able to include in my story was Papka's criticism of the lack of actual successful use cases of data scientists on Wall Street.

"People are investing in it and they haven't really shown benefits from it," Papka said. "Data science is a buzzword. A lot of people don't know what it actually means, but they really believe that this is the future. It's very hard to find an algorithm that actually helps the business, a predictive model that actually leads to something actionable. So you could find a correlation, but it doesn't mean that you could use it for your business purposes."

In Papka's opinion, there are only two formulas for the successful use of data scientists. The first is what he called a unicorn: a one-person wrecking ball who reports to the top of the business and has access to all the data. The other is a team effort that consists of a group of scientists that have gone through a formal science education.

Papka admitted the two approaches are on opposite ends of the spectrum, but said it's the only two ways he's found data scientists to be successful.

The unicorn approach seems particularly difficult, in my opinion. To find one person with that type of range and management skills would be a big ask, which Papka admitted. Also, I'm not sure how successful a single person could be at a massive asset manager or bank.

Still, I do agree with Papka that some type of hybrid approach might not necessarily be successful. Too often, firms try and hedge their bets by not fully committing to a technology. More often than not that only leads to failed projects and lost capital.

 

This week on the Waters Wavelength podcast ─ Episode 14: NATAS, Sell-Side Technology Awards

If you haven't already, subscribe to the podcast on iTunes here. Also, check out our SoundCloud account here.

 

 

Food for Thought

  • I'm heading to Miami Beach for Sifma Ops 2016. I'll be arriving Monday and leaving Thursday. If you'll be down there and would like to meet up, drop me a line here.
  • "Game of Thrones" has gotten all the hype this week, but for the technology folks, I truly suggest "Silicon Valley." I know I'm not breaking any news here, but it's an absolutely fantastic show and worth checking out for those of you who haven't gotten around to it yet.

Only users who have a paid subscription or are part of a corporate subscription are able to print or copy content.

To access these options, along with all other subscription benefits, please contact info@waterstechnology.com or view our subscription options here: http://subscriptions.waterstechnology.com/subscribe

You are currently unable to copy this content. Please contact info@waterstechnology.com to find out more.

You need to sign in to use this feature. If you don’t have a WatersTechnology account, please register for a trial.

Sign in
You are currently on corporate access.

To use this feature you will need an individual account. If you have one already please sign in.

Sign in.

Alternatively you can request an individual account here