Choice of courses, computational methods for SDE or machine learning

[Alexander Westman]

Hi!

Studying for a masters in financial mathematics right now (from a computer engineering bachelors), I want to become a quant, but I don't really know what would be most important for me to learn. I have a choice between two courses this spring, and both are kind of relevant to the subject, but not a complete 100% hit.

The first one is introduction to machine learning, a mandatory course for anyone studying for a masters in machine learning, but optional for financial mathematics.

The second one is computational methods for stochastic differential equations and machine learning, which is a course built around for example determining whether or not Black-Scholes or a Monte Carlo-simulation is more efficient in determining an option pricing.

I'm making the choice in three weeks. The way I figure it, I can probably study introductory machine learning on my own if I have too, but computational methods will be harder to study by myself, so on that basis I'm leaning towards the second one. But any advice would be greatly appreciated.

Links to the two courses:

KTH | DD2421

KTH course information DD2421

www.kth.se

KTH | SF2525

KTH course information SF2525

www.kth.se

 

[Daniel Duffy]

I took a random sample

"formulate, use and analyze the main numerical methods for stochastic differential equations, based on Monte Carlo stochastics and partial differential equations,"

Wow. It could mean anything. It is not specific enough.



"The second one is computational methods for stochastic differential equations and machine learning, which is a course built around for example determining whether or not Black-Scholes or a Monte Carlo-simulation is more efficient in determining an option pricing."
Too much stuff to digest in a short period of time. It is PhD level.

 

[Alexander Westman]

Thank you for your response!

If it's too much stuff to digest in a short amount of time, how long would you suggest to do it? Are we talking closer to half a year, or a full year? Or even longer?

 

[Daniel Duffy]

Thank you for your response!

If it's too much stuff to digest in a short amount of time, how long would you suggest to do it? Are we talking closer to half a year, or a full year? Or even longer?

Hard to say. The requirements are not precise enough. It's a huge area. Try looking

1. Glasserman 2004 Monte Carlo Methods
2. Duffy and Kienitz 2009 Monte Carlo Frameworks in C++.

These kinds of questions have been on QN for years now, some would say ad nauseam.

 

[Daniel Duffy]

Apart from anything else, I suppose programming skills in C++ and Python would do no harm.