• C++ Programming for Financial Engineering
    Highly recommended by thousands of MFE students. Covers essential C++ topics with applications to financial engineering. Learn more Join!
    Python for Finance with Intro to Data Science
    Gain practical understanding of Python to read, understand, and write professional Python code for your first day on the job. Learn more Join!
    An Intuition-Based Options Primer for FE
    Ideal for entry level positions interviews and graduate studies, specializing in options trading arbitrage and options valuation models. Learn more Join!

Is Stochastic Processes still a relevant course to take for a prospective quant?

Joined
1/5/20
Messages
18
Points
13
Hello,

What are everyones thoughts on this question. I have read that stochastic processes are less relevant in the industry now vs pre-08 as derivatives trading has scaled back considerably. From this list of modules, what would be the most relevant in preparation for a career in quantitative research?

45 - 75 from the first set and then 45 - 15 from the second group. I have been guided towards Applied Bayesian Methods, Numerical Optimisation, Graphical Models, Probabilistic and Unsupervised Learning, Advanced DL and RL as well as the Intro to DL. Any suggestions?

1579789537688.png
 
What do you already know? How much probability, mathematical statistics have you done?
 
What do you already know? How much probability, mathematical statistics have you done?

Currently taking a final year undergrad level course in mathematical stats covering Classical Inference, Bayesians stats and inference, simulation and multiple regression
 
It kind of depends on what you mean by quantitative finance. Sell side relies heavily on StoCal, Buy side more on forecasting / probability. It seems a lot of solid folks start on sell side and go to buyside, but I don't hear many (any) examples of going the other direction. Here's a good overview of the differences. 'P' Versus 'Q': Differences and Commonalities between the Two Areas of Quantitative Finance by Attilio Meucci :: SSRN
I was hoping to go straight to the buy-side in a junior research role so was thinking the StoCal would not be as relevant as say the Bayesian stats and DL and RL
 
Anything you have seen which is used a lot in particular? Anything with promising applications in the future?
i don’t think he knows what he’s taking about. Most of the jobs that require ML on buy side is overhyped bs nonsense. ML didn’t get invented automatically. It was there many many years ago and many mfes didn’t have ML in curriculum and still got into quant roles in buy side.
 
i don’t think he knows what he’s taking about. Most of the jobs that require ML on buy side is overhyped bs nonsense. ML didn’t get invented automatically. It was there many many years ago and many mfes didn’t have ML in curriculum and still got into quant roles in buy side.

I agree that the investment case for ML is (with some exceptions) largely unproven and it's not unfair to call the hype BS at this point in time; at least tentatively.

However, the question of investment utility is entirely different from trends in required skills.

These change over time; 5 years ago, ML / Data Science were rarely if ever mentioned in buyside job descriptions, now they are much more prevalent. It's true many MFE's did get jobs without these skills. It's not clear they would still be as competitive for those (medium - long term horizon) kind of jobs today. Several top programs have tossed out a bunch of sto-cal courses to do more Data Science / ML. If traditional MFE curriculum were enough, why would they need to add those capabilities to be competitive for buyside recruiting?
 
almost all buy side quant jobs r now asking for ml skills. and the requirement is increasingly not "i can use blah function from blah library" but "i understand the basic theory cold and can program my own ml library"
 
almost all buy side quant jobs r now asking for ml skills. and the requirement is increasingly not "i can use blah function from blah library" but "i understand the basic theory cold and can program my own ml library"

is linear methods for regression and linear methods for classification is enough for more junior roles that mostly recruit bachelors and masters? maybe support vector as well. deep learning seems like application of undergrad linear algebra so don't think it'll be too hard...
 
deep learning seems like application of undergrad linear algebra so don't think it'll be too hard...

Everything sounds easier when you put it that way. My continuous time models course in grad school was an application of undergrad calculus and probability theory. It was one of the hardest courses I've ever taken.

To answer the op, I never would have thought I needed stochastic calc after I graduated but I did end up using it. My advice would be to keep it a part of your studies, but by no means does it have to be the core of it. Always good to leave some doors open for a variety of opportunities.
 
Last edited:
is there anyone from the buy side here with a mfe who uses ML daily. Who needed to code a neural net or whatever from scratch in his interview, and is successfully adding value with deep ML knowledge ?
 
Everything sounds easier when you put it that way. My continuous time models course in grad school was an application of undergrad calculus and probability theory. It was one of the hardest courses I've ever taken.

To answer the op, I never would have thought I needed stochastic calc after I graduated but I did end up using it. My advice would be to keep it a part of your studies, but by no means does it have to be the core of it. Always good to leave some doors open for a variety of opportunities.
Well said.
In my case I am using stuff (e.g. PDE) that I learned, starting in 1972. And I now teach it to MSc/MFE students.
Stochastics are important.

No one can say if ML -as we know it Jim - will be around in 5 years time. AI has had a very checkered history..
 
Last edited:
almost all buy side quant jobs r now asking for ml skills. and the requirement is increasingly not "i can use blah function from blah library" but "i understand the basic theory cold and can program my own ml library"

It is extremely idiotic and downright unproductive to "program my own ml library". "Understanding the fundamentals" is a generic advice that is applicable to all fields and endeavors in life. Why on earth would you duplicate all the hard work that has gone in to make scikit learn fast, modular and uniform across so many regression and classification algorithms?

I am pretty sure you have never worked in any capacity as a ML researcher/data scientist and just making banal and amateurish comments to people actually asking for help. Yeah go ahead and make your own deep learning library while we all try to make some real progress with free frameworks designed by Google and Facebook.
 
i mean sometimes there’s no off the shelf solution from sklearn and/or need to do ml in the company’s in house programming language. like if u need some special optimization function or least squares scheme or neural network architecture that is fit for a particular situation. i doubt pdt or tgs or renaissance actually uses those
 
i mean sometimes there’s no off the shelf solution from sklearn and/or need to do ml in the company’s in house programming language. i doubt pdt or tgs or renaissance actually uses those, at least for production

Most of the hard work is in data collection/labelling, network architecture and finally tuning the parameters while training. All of this can be done in Pytorch or Keras with less than 200 lines of code. "Programming ml" isn't where most of the time goes, if you dont have a large labelled dataset then most ML algos will give you crap results.

I am only speaking from my experience - I mostly use Pytorch as it is extremely streamlined and fast enough for most NLP projects I am working on (personal efforts). It is impressive how easy it is to construct your own network using a little OOP and just stacking the elements (convolution layer, batch normalization, pooling layer etc). You can design your own new architecture just using Pytorch. The breakthroughs in the last five years - residual networks, inceptionnet - were all through new architecture and not "new in house programming language".
 
Back
Top