How is data science applied in investment banking?

Question

What do data scientists do at investment banks? What tools are they using? What kind of analysis are they doing? Why are they doing it? Etc.

Answer 1

In investment banks, this role is called "Quant" (Quantitative analyst). A quantitative analyst is a person who specializes in the application of mathematical and statistical methods to financial and risk management problems. The occupation is similar to those in industrial mathematics in other industries.

https://www.phaidoninternational.com/industryinsights/the-growth-of-quants-in-investment-banks-71503010311421

https://www.investopedia.com/articles/active-trading/111214/quants-what-they-do-and-how-theyve-evolved.asp

https://www.investopedia.com/articles/financialcareers/08/quants-quantitative-analyst.asp

https://en.wikipedia.org/wiki/Quantitative_analyst

Answer 2

First off, realize that in most banks there are what are known as "front office" (FO) and "back office" (BO) operations. Your scope of work is very much driven by which one of these you are at the bank.

The most obvious answer you are going to get here is around FO operations and being a "Quant" trader. But even within the quants, there are still divisions. For example, there are some quants that only do algorithm development on a theoretical level, they do not do any programming. There are also quants that do programming and very little algorithm development. Depending on the bank, you might also find that data engineers are considered quants in the sense that they work in the same department and are constantly looking for way to optimize the dataflow across all teams.

Along with the quants, you're also going to find quants that do nothing but model validation. Trust me, you have to jump through a lot of hoops before a bank will trust you with their money. That process includes a comprehensive model validation process where every part of your proposal will be taken apart and (hopefully) validated for use. Again, depending on the bank, most quants actually start as model validators, going through proposals for models. If you can't provide proper rationale in these papers, they sure as heck aren't going to let you write algorithms for them!

On the BO side, you can still work for an investment bank as a data scientist but not actually be a quant. For example, virtually every bank has some sort of fraud detection function that is staffed by data scientists. That fraud outlook may be looking both externally and internally. There's also data scientists working in compliance to make sure your bank is running legally across any number of regulations.

So, there you have it. There's tons of opportunities for you to be a data scientist within an investment bank.