Coursework in Masters of Statistics

little_fermat · Post by **little_fermat** » Wed Aug 08, 2018 8:26 pm

Hello,

First time posting, I have been reading a lot of the phorum over the past several months and the wealth of information here is incredible. I am starting a Master's in Statistics and I was hoping to get some advice from the members here on course selection.

My program offers several different "tracks" within the stats dept. Quantitative finance and machine learning are the two that I feel are most relevant to my career goals. The quantitative finance track is heavy in things like stochastic processes, and time series modeling. There are also electives in everything from numerical methods to portfolio theory. In ML there is the standard set of data mining/statistical learning classes and also things like deep learning and natural language processing.

Ideally I'd like to get a job where I could be part of a team that researches and implements automated trading strategies. Time frame/asset class aren't really that important to me I'd just like to work in a group that applies a scientific approach to trading, i.e. hypothesize strategy, research and backtest, implement and track results.

My question is whether anyone has an opinion on which track would better serve that career goal. My bias is towards ML but this only comes from analyzing keywords in job postings. I've looked at ~15 companies which seem to do what I'm interested in (mostly hedge funds and prop shops), and the job postings seem to emphasize data analysis & programming skills, and in some instances ML specifically. I see less regrading stochastic processes or portfolio theory but these postings are pretty vague.

I'd guess that both are viable options and there are a variety of firms applying all sorts of strategies. I've spoken to a few people who work adjacent to the industry and that seems to be the general consensus. One person I talked to gave the advice that I should choose the track where I was most interested in the material. Gun to my head I would say ML but honestly I think the technical aspects of both are quite interesting and I could see myself being happy with either choice.

I should add that I have some basic academic background in CS (C, python), have taken courses in databases and operating systems and my undergraduate was in pure math.

Any input would be greatly appreciated,

Thanks

TL;DR: Quantitative finance or ML for systematic trading strategies?

day1pnl · Post by **day1pnl** » Wed Aug 22, 2018 4:59 pm

Gun to my head, I'd say theoretical or computational statistics (machine learning) is superior choice here.

The problem with quantitative finance is that the academic level is too low at the same time as it lacks professional specialty. From an academic point of view, your skills in measure theory will be too watered down to be competitive in PhD positions in stochastic analysis. On the other hand, you will not learn a trade or a profession so you won't get specialized educational training like lawyers, medical doctor,s or carpenter,s do. In my opinion it's not a great compromise. Though I haven't gone through the statistics I think supply/demand isn't great anymore either.

That said, education is not everything - you also have to consider what your interests are, and I think smart, hard working people can get good jobs regardless of differences in educational background.

little_fermat · Post by **little_fermat** » Wed Aug 22, 2018 10:17 pm

@day1pnl thank you for your response. That's a good point you make about the academic level of the quantitative finance track and it's something I hadn't considered. Perhaps those kinds of courses may be well suited for someone who is looking for a job in risk or something similar. I appreciate your feedback.

Rashomon · Post by **Rashomon** » Tue Sep 04, 2018 3:59 pm

little_fermat, you've probably seen that people post questions like this all the time. What you are expecting to work won't.

Academia is the wrong place to look. I would suggest speaking to a recruiter. There exist skills which are incredibly in demand, including by trading firms, and these guys are hungry to place those who have them. Academia has turned itself inside out to deceive people like you.

day1pnl · Post by **day1pnl** » Tue Sep 04, 2018 9:24 pm

@Rashomon, bit harsh I'd say. I dont know how much they screw with the students in general. Maybe I am naive, but think selling quantitative finance certificates is more herd behaviour than outright deceit.. "All other schools are selling this, so we need to do it too " kind of thinking.

@little_fermat:

> "just like to work in a group that applies a scientific approach to trading, i.e. hypothesize strategy, research and backtest, implement and track results."

Wouldn't worry. 99.9% do this and are making money doing that. Else they wouldn't be in the market. You do have to make sense. Eventually you have to explain what is going on in your books. Its never just "I sold because I wanted to see what happens". (well, sometimes it is. but thats not the point). Its about being rational about trade ideas, hedge strategy, expected p/l, realized p/l, capital costs, risks and position size, new regulation, flows/technicals driving the market, unwind costs etc etc..

deeds · Post by **deeds** » Tue Sep 04, 2018 9:44 pm

Data pointlet: Interviewed 3 fine, fresh MFE (master in financial engineering) grads, sorted and selected from fine west coast schools, for a position valuing financial derivatives.

Not one could talk about any aspect of finance meaningfully, nor were interested in finance. All said their courses had left them untroubled in the ways of quantitative finance. Each wanted to help machines to learn using regressive procedures. ; )

Each wanted to be a 'data scientist'.

i was really surprised. The world has a very bad hype infection.

All that by way of saying curriculators [sic] seem responsive to what sells.

little_fermat · Post by **little_fermat** » Wed Sep 05, 2018 2:29 am

@Rashomon, thank you for your input. That's a great idea and I'll definitely try to speak with a recruiter.

> "There exist skills which are incredibly in demand, including by trading firms, and these guys are hungry to place those who have them. Academia has turned itself inside out to deceive people like you."

I'm not expecting to take classes which are perfectly tailored towards trading. I know that none exist. The crux of the question was whether I'd be shooting myself in the foot by not taking the opportunity to learn stochastic processes/portfolio theory.

As @deeds points out I think a lot of students these days want to work in a more data driven role. I would probably include myself in this set. Just trying to better understand what skills are utilized in various roles.

doomanx · Post by **doomanx** » Wed Sep 05, 2018 10:20 am

If you're planning a (successful) career in trading you will need to be able learn new things throughout your entire career. The most important thing one can take away from uni is learning how to learn and how to solve challenging problems, but the specific knowledge you will need is always growing and always changing depending on what project you're working on. To that end, take whichever option seems more interesting/lets you do the most sport/beers/whatever you like while you still have the time.

deeds · Post by **deeds** » Wed Sep 05, 2018 1:52 pm

@doomanx, +1

Rashomon · Post by **Rashomon** » Fri Oct 19, 2018 2:46 pm

day1pnl: “Hey, whatever gets me laid” is not acceptable coming from a teenager, let alone from a grown-up who should (and often does) know better.

Universities receive public and private largesse precisely because they promise an alternative to the world’s tendency to race to the bottom. David Starr Jordan and William Rainey Harper would not have recruited stock-market pontificators or self-styled “data” experts—and JD Rockefeller/Jonas Clark/Leland Stanford/Landon Clay/C. F. Mueller wouldn’t have supported such a faculty if they had.