Which areas of applied and pure mathematics (at a postgraduate level of background) would be ideal for a quantitative researcher? For link to part 2 - non math subjects: https://us.teamblind.com/s/TLL8mbTH
Sell-side quant-turned SWE here. I think the importance of graduate-level mathematics (relative to software engineering skills) is on the decline. The maths for flow derivatives has largely been worked out decades ago. Exotics are typically priced using Monte Carlo methods. Algo trading is a statistics game, but very engineering-heavy.
Software engineering is usually picked up in industry though. Most computer science professors can't teach (soft) SWE skills because they don't have experience dealing with large-scale systems in production, excluding some folks who engage in distributed systems or open-source work. I knew practically nothing about software until a full-time job in tech; my limited, previous experience at quant firms didn't amount to much in the way of applying design principles and solid engineering practices.
A quant told me the same thing. All the exotic stuff where the Physics PhD’s were very useful has been worked out, and currently it is all about finding signals in vast amounts of data i.e the same math used in Machine Learning is what is useful in quant finance today.
Sage advice from my grad school stochastic calculus TA (postdoc): “don’t bother with a Ph.D., just get an MFE and go into industry”... Haha
Different utility functions and regret minimization lol
There's another thread from a while ago that says the opposite, that MFE is useless and it's PhD only for quantitatively meaningful work 🤷♂️
The real answer IMO is none of the above, but optimization, statistics, and stochastic calculus are good to know, on top of linear algebra. The idea that you need to be good at math is pretty exaggerated, as anyone with proper math training will feel like their math skills are total overkill. Before some of my investments made my NW skyrocket and made me stop worrying about money, I was pretty resentful of the fact that statistics and computer science degrees seemed to get my friends much further in their careers. It's very sad to study esoteric abstract nonsense and then have to struggle with explaining some really simple linear algebra the rare occasion you have to use it.
But esoteric abstract nonsense (math) is cool though ☹️
It’s full of weird symbols and shit lmao
Makes no difference, all falls in the oh he’s a math PhD bucket
Even at top firms like Edgestream, Rentec, TGS?
Lol wtf? Abstract algebra? You just posted a list of courses offered in a math department
An inteviewer I met from D.E. Shaw did his dissertation on additive combinatorics, so clearly it has a lot of applications /s We can make a sort of (bad) argument along the lines: algebra -> algebraic topology -> TDA (betti numbers, homologies, etc.) Or, actually I threw in a few oddballs to estimate the number of troll voters
Nobody really does tda, it was trend a few years ago with Gunnar carlsons work and that startup he founded
Oh boy, I admire you guys. This is a whole new level of nerd.
Of course my specialities (differential and algebraic geometry) are down at the bottom
Wasn’t founder of rentech an algebraic geometer
Differential geometer, but Rentec is an oddball; lots of algebraists and mathematical physicists too
😱
Hahaha category theory for the win
That can actually be useful for normal software engineering. One of my senior engineers used it to justify engineering designs.
Would be interesting to hear what that was about.
This should cover most of the modern popular branches. I deliberately excluded econometrics and ML because they're not traditionally considered of interest by faculty in mathematics departments. Obviously one shouldn't choose a doctoral research topic based on applications in industry. I do have some weak preferences and priors at this point, but if the academic path doesn't work out at least there might be a partial consolation prize. There are a few ways to interpret "ideal background": 1. Economic signaling of aptitude 2. Academic-industry connections 3. Relevance to job by intuitive understanding of markets data 4. Usage of language in papers or implementation Q. Why not use interview questions as a proxy? A. I've gathered a small sample, but questions understandable by a large proportion of applicants are overrepresented, and interview content is not always a good indicator of the actual job. Q. What about using Indeed and LinkedIn data? A. We can compute posteriors, although it's hard to distinguish whether ex-academics in one field of study currently work as a quant researchers in higher frequency due to targeting by headhunters, ability to land a top gig, or other reasons. Too many confounding variables.
Came here to say I like your username
Thanks boss