Course advice

[Menime]

I am currently pursuing a quant program.
I have an option to select electives and I can take one of the following paths:
1. Data mining course followed by Machine learning which will give me strong computational statistics foundation.
2. Advanced PDE course followed by a non linear PDE analysis thus taking a mathematical route.
Although I know where my interests lie, I would like to know which of these two are more relevant in the quant industry currently.
More specifically, what kind of job roles look for each of the above set of options.

Thank you.

 

[Daniel Duffy]

It's like asking if judo is better than karate

2. I would learn how to solve PDEs numerically. Pricing solvers in PDE are the fastest.

1. is very trendy/hype. Just sayin'

 

[qdogg]

The advanced math courses in quant programs were traditionally intended to prepare people for jobs in derivatives modeling, the vast majority of those positions disappeared after the financial crisis, and the tiny handful of them left now nowadays almost invariably require people to have PhDs, not just Masters degrees... If priority #1 is your career, just focus on learning as much programming now as possible and leave the PDE shit to the guys getting PhDs.

 

[Daniel Duffy]

PDE shit

LOL. Maths is being dumbed down.

Some MSc students in UK do do PDE

Datasim Financial Forums • View topic - MSc theses in computational finance

 

[bigbadwolf]

Some MSc students in UK do do PDE

He's not saying they don't; he's saying it will be difficult to get a job after a PDE-focused master's because of the marked preference for Ph.D.s for the relatively few jobs that are out there. Although I agree that PDEs are forever while the machine learning might just be trendy and faddish.

 

[Daniel Duffy]

He's not saying they don't; he's saying it will be difficult to get a job after a PDE-focused master's because of the marked preference for Ph.D.s for the relatively few jobs that are out there. Although I agree that PDEs are forever while the machine learning might just be trendy and faddish.

I agree.
It's the expletive that got my back up :D

There was a time when maths students did PDE/FDM at undergraduate level.

 

[bigbadwolf]

There was a time when maths students did PDE/FDM at undergraduate level.

And still do at many US universities. I actually prefer the computation-based undergrad PDE courses to the theory-focused grad courses (Sobolev spaces, etc.)

 

[Daniel Duffy]

And still do at many US universities. I actually prefer the computation-based undergrad PDE courses to the theory-focused grad courses (Sobolev spaces, etc.)

In the 70's numerical analysts (Strang, Aubin, Lions and the French school) formulated PDE in Sobolev spaces and _solved_ them using the Finite Element Method (FEM).

Very esoteric. Proving convergence in H_1 space is part of the deal. Ad infinitum..

It is possible to do FEM w/o Sobolev but then it becomes engineering (nothing wrong with that).

 

[pingu]

He's not saying they don't; he's saying it will be difficult to get a job after a PDE-focused master's because of the marked preference for Ph.D.s for the relatively few jobs that are out there. Although I agree that PDEs are forever while the machine learning might just be trendy and faddish.

Even though I agree, PDEs are not going to help you a lot now to get a job in Finance. It's the whole P vs Q world. P won. You rarely solve a PDE anymore in mainstream finance.

 

[bigbadwolf]

Even though I agree, PDEs are not going to help you a lot now to get a job in Finance. It's the whole P vs Q world. P won. You rarely solve a PDE anymore in mainstream finance.

You are right.