Python vs MatLab

[kdmfe]

Dear experts,

I need guidance on which programming language to learn for further my skills.

I have developed the knowledge to understand Stochastic Calculus-Continuous time models, by Steve Shreve'. I am currently doing Numerical methods in Computational Finance by Daniel Duffy.

With the above knowledge, learning which programming language is suggested -Python or Matlab.

I have some basic understanding of C language.

Thank you in advance for your guidance.

 

[CrossGamma]

MATLAB has been or is being replaced by Python as prototyping and analysis language almost everywhere.

 

[kdmfe]

Hello CrossGamma

thank you for your comment.

Is there any particular module/book in python which is specific to stochastic/quantitative finance.

Also, are you able to suggest a few certifications.

Thank you

 

[Andy Zhang]

For Python books, you can check out Wes McKinney and Yves Hilpisch.
There's also the Python course on QuantNet, depending on your budget.

 

[Daniel Duffy]

Go for Python. Many do Python these days.

Online Courses :: Datasim

www.datasim.nl

Online Courses :: Datasim

www.datasim.nl

Online Courses :: Datasim

www.datasim.nl

As student you have a discount and courses after the first one are even more attractive, price-wise. Please contact if you any queries.

 

[kdmfe]

Hello Andy, thank your for your suggestion.

I looked at the online course, there are four of them. Where do you think is the best to start. budget is not a constrain. I need to the curriculum to be synergetic and advance my knowledge and job prospects.

Online Courses

quantnet.com

Thank you

 

[Andy Zhang]

Hello Andy, thank your for your suggestion.

I looked at the online course, there are four of them. Where do you think is the best to start. budget is not a constrain. I need to the curriculum to be synergetic and advance my knowledge and job prospects.

Online Courses

quantnet.com

Thank you

There's just one for Python.

 

[CrossGamma]

Hello CrossGamma

thank you for your comment.

Is there any particular module/book in python which is specific to stochastic/quantitative finance.

Also, are you able to suggest a few certifications.

Thank you

I personally don’t care much for certifications. For quant graduates, I typically give a take home exam and to test your understanding of some language features in the interview. The former is a challenging algorithmic problem and also assesses the quality of your implementation (tests, abstractions, extensibility, …). The latter is to test how deeply you thought about how thing work behind the scenes. I am not a fan of Leetcode-like assessments but know they are used a lot in the industry as easier to administer in scale. Best preparation for the interview I use is to implement small projects yourself and be very conscious about design choices, writing good tests, …

 

[Daniel Duffy]

kdmfe, Since you are doing my numerics course you can send me your ODE questions.

 

[Daniel Duffy]

I personally don’t care much for certifications. For quant graduates, I typically give a take home exam and to test your understanding of some language features in the interview. The former is a challenging algorithmic problem and also assesses the quality of your implementation (tests, abstractions, extensibility, …). The latter is to test how deeply you thought about how thing work behind the scenes. I am not a fan of Leetcode-like assessments but know they are used a lot in the industry as easier to administer in scale. Best preparation for the interview I use is to implement small projects yourself and be very conscious about design choices, writing good tests, …

Yeah, but how does one become "conscious about design choices", sounds a bit of a dumbing down, no? You learn by making mistakes.
It takes years to be good at software design. For production, I suppose time to market might have a high priority.

BTW this discussion from me is not about certification (altho' I give them) but developing skills.

 

[CrossGamma]

Yeah, but how does one become "conscious about design choices", sounds a bit of a dumbing down, no? You learn by making mistakes.
It takes years to be good at software design. For production, I suppose time to market might have a high priority.

BTW this discussion from me is not about certification (altho' I give them) but developing skills.

Totally agree with you and maybe didn’t express myself well. It’s something you can only learn by getting your hands dirty and making mistakes.

 

[kdmfe]

kdmfe, Since you are doing my numerics course you can send me your ODE questions.

Daniel, I have not yet registered for your course. I purchased your book and I am writing my notes for the topics in the book.

Hope I am clear.

If your course helps me enhance my knowledge and helps me understand your text book well, I will certainly register very soon for both ODE/PDE and Python course.

I quit job about six year ago, and started building my understanding of maths right from High shcool to PG level. I am an accountant by background.

Thank you

 

[Daniel Duffy]

Totally agree with you and maybe didn’t express myself well. It’s something you can only learn by getting your hands dirty and making mistakes.

No problem. It's good that the new MFE generation don't think that programming is a dawdle.

 

[alovya]

I typically give a take home exam and to test your understanding of some language features in the interview. The former is a challenging algorithmic problem and also assesses the quality of your implementation (tests, abstractions, extensibility, …).

Not to be cheeky, but can you give an example of a question similar to these take home exams?

For preparation you suggest doing small projects while being very conscious of design choices (tests, abstractions, extensibility, etc). I have experience in this and feel relatively comfortable re: abstractions and extensibility, but not tests. I'd like to hear what you expect re: tests.

For reference, the tests I've written for physics simulation codes during my PhD have never gone much beyond "automatically run several test cases and compute the error between my numerical solutions and reference solutions, and if the errors are small enough this means the code is working". Not to say that they can't be better than this, I'm just giving my background.

EDIT: Example of a test suite I wrote.

EDIT 2: This is possibly relevant.