Should You Take Data Science Classes for Quant Finance?
Vložit
- čas přidán 9. 07. 2024
- Should you take a data science or machine learning course if you intend on working in quantitative finance? Yes!
But there is something to be aware of. I would recommend taking a probability and statistics course before you take any machine learning or data science. The foundation of machine learning is traditional statistics (ML is just a sub area of statistics). By taking some basic statistics and probability courses you'll better understand what problems you are trying to solve and the weaknesses of the different methods.
CMU Data Scientists in Quant Finance Panel
apply.mscf.cmu.edu/register/f...
Free ML Resources:
www.fancyquantnation.com/free
Website:
www.FancyQuantNation.com
Quant t-shirts, mugs, and hoodies:
www.teespring.com/stores/fanc...
Connect with me:
/ dimitri-bianco
/ dimitribianco
I would love it if you could make an updated version of your 2019 video on Quant Reading List. There are so many interesting books that I keep going back and get stuck in a tutorial hell...
What topics specifically are you stuck on?
Very nice Dimitri! I signed up for that machine learning/ data science event thing @ CMU thank you my friend
Thank you so much Dimitri, I will arrange both of these books.
yes, but before statistics you should be good at math, at least at the level of calculus
Thanks for this. There's no way to get around stats!
What is your opinion on "The Elements of Statistical Learning"?
Same authors as Intro to SL, but more maths heavy and theoretical.
I have not read that one but a buddy of mine has high praise for the book. The more rigor the better for quant finance.
Hi Dimitri, first of all thanks for all your content!! I'm currently an undergraduate economics student from Peru, in a program that is very numerical based (very mathy micro and macroeconomics, linear algebra, convex and dynamic optimization with notions of topology, calc 1 and 2, ODEs, statistics, econometrics, and ML focused on statistical learning). All my math courses began with a proof-based focus. With that being said, i feel that i still lack from math creedentials for an MFE master, and im taking in consideration doing a master in applied maths (with a focus in stochastic processes) here in Peru before applying abroad. My question goes about internships: i've been in a bank doing advanced pricing analytics (about credit products) and models (econometrics/ML in Python), but i lacked seeing capital markets. Then i changed to a pension fund internship into market risk area, but it is not as statistically-driven as i expected (it is mostly traditional finance view). My end-goal for now is to be a Market risk Quant model developer, or a role where the core is to model about investments. Is it better for me to find an internship in traditional finance market risk, or i'd rather go for a data science/machine learning role? Thank you in advance, you are one of the reasons i'm so passionate about giving everything to become a quant someday!!!
I would go with data science and machine learning. Being technical is more important experience wise compared to being in investing but on the traditional finance side.
For PD modelling, I was thinking what's the difference between an approach if we take log_odds of the default indicator (i.e. whether an account defaulted next 12 months or not) for each month and develop a linear regression mode for these log_oddsl with MEV's as explanatory variables using OLS vs developing a logistic model on default indicator with MEV's as explanatory variables. Will both the models be theoretically similar?
For OLS vs logistic, logistic is bounded while OLS is not. However it depends on your data and usage. I've seen OLS regressions outperform a bounded problem. The question that needed to be answered was, "will these inputs create unrealistic answers if they shift?" The business was uncomfortable with this issue and logistic regression was used.
For clarity, do you mean take a course in calc-based probability (usually just called probability theory) and calc-based statistics (which is usually called mathematical statistics)?
The more math the better but just a simple dumbed down probability and statistic courses will put you in a lot better position than most people just taking data science.
@@DimitriBianco well said, thanks. It concerns me that some people “getting into” data science don’t even know the basics of stats, and are not at the level of basic programming as an actual cs major (unless that was their major of course).