Artificial Intelligence | should one pursuit a degree after a MFE?

[lucasnutz]

If you want to participate in the HFT world, would it be necessary to have an AI degree in conjunction of an MFE to become a HFT? I ask this because even though I am enrolling this fall in the MSCF program, where machine learning and other statistical classes are taught, I don't know if it is really necessary or if it is more like "learning by doing" kind of thing, should you land a job in that field of course.

If the answer is yes, then what topics in AI do you consider key to develop and how would you develop them?

thanks

Lucas

 

[lucasnutz]

No one, really??

 

[Liam]

A lot of the people on this forum are from a quant background so can't definitively answer your question. Based on experience of switching between different fields I'd say try to make the AI work look as little like Plan B as possible both in how you write your CV and what courses you do (doing financial work in your masters is very distracting to employers and this isn't the 1950s where employers couldn't care less if you took a job as a 'gap job').

It's about having skills and how you present them - the question might get a lot is "why did you do MFE then midway decide to do AI?". Whatever answer you give it has to be your own and ensure employers feel you won't get distracted by another matematical area in another 6 months time. HR will ask and I wouldn't really worry about, they're usually twats and get overruled, but your prospective line manager will probably ask this.

First thing is do a thesis in AI, but I'd also do as much research and work outside of coursework. At the very least do programming of your own (quant employers, including the one that hired me, were very impressed when I said I'd programmed basic games in C++ - this was also as I'd learnt a lot of libraries even a PhD wouldn't necessarily know). Your thesis doesn't have the conditions of work experience - internships cost time, but also you can grab a public dataset and work on it, publish a blog and the code goes on GitHub. What employers bitch about a lot is when grads come in and can't hack the real world problems because they don't like the data wrangling (possibly 90% of any job involving data analysis or algorithms. I mean I love it and while the actual sexy work is 5% for me it's 95% sexy... Nice...) or using stupidly out of date systems. You won't replicate all of these conditions, but you can go beyond the confines of academia and can go some way to bridging the gap between academia and work.

 

[Liam]

Also I would say come across a bit better, I was about to log off when I just noticed how many acronyms you used - it sounds like you think it's all about qualifications, but I don't think you see it that way. The only bit that makes sense in your post is "I don't know if it is really necessary or if it is more like "learning by doing" kind of thing" which is buried in a lot of crap about HFT this and MFE that...

In terms of topics maybe look at Mahout's machine learning directory for areas of interest. I'm subscribed to a lot of Big Data feeds on Wordpress, Twitter and LinkedIn - prehaps pick up some issue there. e.g. I was recently reading how Google used search terms and textual analysis to map flu outbreaks, but failed as text searches turned out to not correlate with flu outbreaks. Maybe grab a public dataset and solve an issue like this (as it would involve practical work that would be useful to employers).