Career choices.. any data scientists/quants here?

[Ammo1337]

I'm currently a second year mathematics & physics student who is now thinking of routes and career options. I definitely want to use mathematics and the skills gained from my degree rather than just get lumped in any old office job.

I've got it narrowed down:

BSc/MSci -> Actuary. I've been to quite a few events run by the Big4 and I'm a treasurer of my university's actuarial society for the next academic year. Nice profession, decent money, interesting, satisfying, safe career, work/life balance etc.

BSc -> MSci/MSc (maybe try get into Oxbridge postgrad?) -> PhD -> Data scientist/quant?

Becoming a quant looks very attractive for me, since they use advanced mathematics, programming, and statistics to solve problems and optimise profit etc. However problem here is that when I've searched profiles of people who are/were quants, a lot of them seem to be oxbridge undergrads. The job is a lot more volatile. I've seen job postings of them wanting people who have taken part in mathematical Olympiads and the like. Not to mention I'll need a PhD and by that time I could be a fully qualified Actuary!

Now, I'm not brilliant, but I am hard working and I am not finding my degree too difficult. I am at a semi-target university for investment banker recruiters and I come from quite a bad background, so I never really had the chance to take part in all these amazing mathematical competitions and the like.

I don't know if I am cut out for it, or even a PhD. I am also quite concerned about job stability. I feel there seems to be a level of snobbery from the banks who want you as well?

Any input is greatly appreciated

 

[Scottland]

Have a look at the Data Science MSc in Dundee, it seems very well regarded and they work closely with large data companies like Teradata.

 

[Ammo1337]

Have a look at the Data Science MSc in Dundee, it seems very well regarded and they work closely with large data companies like Teradata.

I'm thinking about doing a data science MSc, possibly. I've had a look around and also seen a very nice PhD programme at Edinburgh.

I'm wondering if anybody here has any experience in these fields, particularly if they have worked in quantitative finance in a large investment bank!

 

[lukeharvest]

Most quants have a PhD, you could also look into strat positions where only half tend to have PhDs. From my experience of being interviewed for quant/strat and trading positions, the former tend to be more hung up on university prestige (especially if you're going to be hired without a PhD) and to realistically stand a chance with a PhD you'd want to do it at a target. However, only do a PhD if you want to do a PhD, they're not that useful at making you more employable - if you're just wanting to get a target university on your CV you'd be better off doing a masters.

 

[dowie]

Any input is greatly appreciated

disclaimer - I'm not a quant, but here is some input....


(I have worked in finance and financial technology jobs for a few years though and know a few quants)

I know one guy who is a quant with a bachelors, though he did get it from Cambridge and started a decade ago. It used to be possible to get quant roles with a masters and probably still is in some places - a lot of the work doesn't necessarily require a PhD they're able to ask for one simply because they can thanks to an abundance of PhD candidates.(Most quants spend most of their time coding and it is existing models not new research they're generally working on)

firstly you're probably better off looking at places like the quantnet and wilmott forums where there is an abundance of information there from quants (including some quite well known ones)

secondly though you're looking at rather different areas - yes the things you've mentioned in the OP are all mathematical to some extent they're using rather different areas of mathematics - for example a a friend of mine was keen on applied mathematics, did an MSc and got a job as a commodities quant - his skill set is in mathematical modelling, optimization etc.. he's not got a great stats background and knows nothing about machine learning, his quantitative work is rather different to that of say a data scientist at a tech firm

on the other hand quant researchers at buy side firms are the sort of people who are big on statistics and machine learning and do have a similar skillset to data scientists

while this isn't a commonly used terminology take a look at this paper re: 'q quants' (the original 'quants' with applied maths/physics PhDs on the sell side i.e. big banks) vs 'p quants' (statisticians etc.. more commonly found on the buy side - funds etc..)

http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1717163

Lastly i'd be skeptical about data science MSc degrees until they're more established and you can get some more data about them!(i.e. employment stats vs more established degrees). Generally for these sorts of careers you'll need a mixed skill set some of which inevitably is going to come from self study... sell side quants need the applied maths skills and finance knowledge and C++ programming skills(a data science degree isn't much help there!) - now MFE degrees have been established they can represent a route in, though often more to roles that aren't strictly quant roles... you could end up in quant dev roles or in risk or at a vendor etc.. a PhD who has self studied the finance side and can program is going to be more valuable. Likewise my friend who studied a proper applied maths MSc and then taught himself the programming side and picked up the finance side as he went along (he joined when maths MSc grads were recruited into quant roles) is much better mathematically then the various 'financial engineering' etc.. grads who cram into one year some programming, some finance and some mathematics applied to finance (which isn't anywhere near as comprehensive as a proper mathematics MSc). Basically it would seem that for a lot of people you'd be better off getting the mathematical/theoretical background taught formally and self study the rest.(though in the case of proper quant jobs this could involve a PhD and it probably isn't advisable to do one just for the sake of landing a job).

So ref: the data science degree I'd assume (and this is just a random opinion) that you'd be better placed to do a proper hardcore stats MSc and self study the other stuff

These data science degrees seem to be of two possible varieties - either a watered down Stats MSc with some models replaced with some applied IT/data related stuff or maybe an introductory machine learning module. Or a Computer science MSc with some modules replaced with introductory stats modules.... It just seems like a watered down option, especially if the modules that make it a 'data science' MSc are easy applied ones or just covering some new buzzword technology. Just an opinion but a proper Stats MSc + self studying other subjects via coursera, udacity... - machine learning, mining massive data sets, courses on hadoop, spark, mongoDB [insert buzzword]. Is probably going to get you a much better background than potentially wasting module credits on easy applied/IT stuff you could pick up yourself for the sake of a new trendy 'data science' degree.

Alternative to the stats Edinburgh, Cambridge and UCL are top for machine learning (which is mostly stats anyway) - UCL offers a combined stats and machine learning masters too which looks to give a solid theoretical background.

 

[dowie]

also ref: Mathematical Olympiad stuff - that is the sort of thing a few hedge funds say they look for but it seems a bit dubious... at best it is just a plus for your CV/gets it noticed by those who like that sort of thing, most of their employees won't have won a medal when they were kids etc.. maybe they think having a few that did helps with marketing their fund as being run by a bunch of super geniuses.


re: actuaries, I only know one but he seems pretty happy - they earn less than quants and have to study a lot at the start of their careers, but then again they work fewer hours and are generally less stressed. They're also not confined to working in London but can earn a nice salary all over the UK and it is an established profession embedded within the insurance industry - job security is probably better and career progression is initially automatic and based on passing exams. There are senior roles within insurance companies for actuaries and there are plenty of actuaries in other senior roles including being appointed as CEOs - they're pretty well respected within that industry it seems - whereas senior management in banks often aren't quants and quants are often second class citizens compared to traders(this situation could well evolve though in banks with quant traders/researchers).

 

[Ammo1337]

Most quants have a PhD, you could also look into strat positions where only half tend to have PhDs. From my experience of being interviewed for quant/strat and trading positions, the former tend to be more hung up on university prestige (especially if you're going to be hired without a PhD) and to realistically stand a chance with a PhD you'd want to do it at a target. However, only do a PhD if you want to do a PhD, they're not that useful at making you more employable - if you're just wanting to get a target university on your CV you'd be better off doing a masters.

Yes, for some reason investment banks like to do that. I am hoping to do a Masters or PhD at a target. I think Imperial or Oxbridge, depending on how my results go

I really do like the idea of doing a PhD and I am interested quite a bit in being on the forefront of mathematics in the industry, however at the same time I am worried I won't be cut out for it.

disclaimer - I'm not a quant, but here is some input....


(I have worked in finance and financial technology jobs for a few years though and know a few quants)

I know one guy who is a quant with a bachelors, though he did get it from Cambridge and started a decade ago. It used to be possible to get quant roles with a masters and probably still is in some places - a lot of the work doesn't necessarily require a PhD they're able to ask for one simply because they can thanks to an abundance of PhD candidates.(Most quants spend most of their time coding and it is existing models not new research they're generally working on)

firstly you're probably better off looking at places like the quantnet and wilmott forums where there is an abundance of information there from quants (including some quite well known ones)

secondly though you're looking at rather different areas - yes the things you've mentioned in the OP are all mathematical to some extent they're using rather different areas of mathematics - for example a a friend of mine was keen on applied mathematics, did an MSc and got a job as a commodities quant - his skill set is in mathematical modelling, optimization etc.. he's not got a great stats background and knows nothing about machine learning, his quantitative work is rather different to that of say a data scientist at a tech firm

on the other hand quant researchers at buy side firms are the sort of people who are big on statistics and machine learning and do have a similar skillset to data scientists

while this isn't a commonly used terminology take a look at this paper re: 'q quants' (the original 'quants' with applied maths/physics PhDs on the sell side i.e. big banks) vs 'p quants' (statisticians etc.. more commonly found on the buy side - funds etc..)

http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1717163

Lastly i'd be skeptical about data science MSc degrees until they're more established and you can get some more data about them!(i.e. employment stats vs more established degrees). Generally for these sorts of careers you'll need a mixed skill set some of which inevitably is going to come from self study... sell side quants need the applied maths skills and finance knowledge and C++ programming skills(a data science degree isn't much help there!) - now MFE degrees have been established they can represent a route in, though often more to roles that aren't strictly quant roles... you could end up in quant dev roles or in risk or at a vendor etc.. a PhD who has self studied the finance side and can program is going to be more valuable. Likewise my friend who studied a proper applied maths MSc and then taught himself the programming side and picked up the finance side as he went along (he joined when maths MSc grads were recruited into quant roles) is much better mathematically then the various 'financial engineering' etc.. grads who cram into one year some programming, some finance and some mathematics applied to finance (which isn't anywhere near as comprehensive as a proper mathematics MSc). Basically it would seem that for a lot of people you'd be better off getting the mathematical/theoretical background taught formally and self study the rest.(though in the case of proper quant jobs this could involve a PhD and it probably isn't advisable to do one just for the sake of landing a job).

So ref: the data science degree I'd assume (and this is just a random opinion) that you'd be better placed to do a proper hardcore stats MSc and self study the other stuff

These data science degrees seem to be of two possible varieties - either a watered down Stats MSc with some models replaced with some applied IT/data related stuff or maybe an introductory machine learning module. Or a Computer science MSc with some modules replaced with introductory stats modules.... It just seems like a watered down option, especially if the modules that make it a 'data science' MSc are easy applied ones or just covering some new buzzword technology. Just an opinion but a proper Stats MSc + self studying other subjects via coursera, udacity... - machine learning, mining massive data sets, courses on hadoop, spark, mongoDB [insert buzzword]. Is probably going to get you a much better background than potentially wasting module credits on easy applied/IT stuff you could pick up yourself for the sake of a new trendy 'data science' degree.

Alternative to the stats Edinburgh, Cambridge and UCL are top for machine learning (which is mostly stats anyway) - UCL offers a combined stats and machine learning masters too which looks to give a solid theoretical background.

also ref: Mathematical Olympiad stuff - that is the sort of thing a few hedge funds say they look for but it seems a bit dubious... at best it is just a plus for your CV/gets it noticed by those who like that sort of thing, most of their employees won't have won a medal when they were kids etc.. maybe they think having a few that did helps with marketing their fund as being run by a bunch of super geniuses.


re: actuaries, I only know one but he seems pretty happy - they earn less than quants and have to study a lot at the start of their careers, but then again they work fewer hours and are generally less stressed. They're also not confined to working in London but can earn a nice salary all over the UK and it is an established profession embedded within the insurance industry - job security is probably better and career progression is initially automatic and based on passing exams. There are senior roles within insurance companies for actuaries and there are plenty of actuaries in other senior roles including being appointed as CEOs - they're pretty well respected within that industry it seems - whereas senior management in banks often aren't quants and quants are often second class citizens compared to traders(this situation could well evolve though in banks with quant traders/researchers).

Yeah, you used to be able to get away with a Bachelors but now you need pretty much need a PhD, although I have heard of getting in with a Masters through some serious connections and experience.

I have been lurking those places, but I thought I'd give a shot here too, maybe get a bit less of a biased opinion

I have been looking at a variety of masters but I am still more drawn to mathematics/theoretical physics ones more. In particular I've been looking at Part III at cambridge where I have a lot of flexibility. I could take modules on PDEs and optimisation, stochastic processes and calculus, and then general relativity & quantum theory .

I've also been looking at kaggle to get started on using R & python for data mining etc and hopefully start up on learning C++ this summer, as well as hopefully nailing some kind of internship. Busy summer for me but it has to be done!

Thank you for clearing the olympiad stuff up. I'm not a genius but I am quite interested and motivated, and a hard worker.

I've read up a lot on actuaries and it does seem a lot less stressful (except the first few years where you are studying as well) and it is quite an attractive career. I'm just wondering if it will be "enough" for me and that I won't have any regrets about not pursuing a PhD.

Oh dear. That's what I was afraid of. I have heard about a lot of snobbery and swallowing your pride when you work in a bank and that is one thing that I really wouldn't want to deal with or do. The actuaries I've met at the local offices have been very nice and welcoming, and certainly seemed interested in us!

Thank you so much for all of your input! I did not expect such detail

Also, just in case it means anything, these are three particular PhD programmes I've been looking at about possibly doing in a few years:

https://www.maths.ox.ac.uk/study-he...athematical-modelling-epsrc-cdt/infomm-course

https://www.maths.ox.ac.uk/study-he...rential-equations-epsrc-cdt/pde-cdt-structure

http://datascience.inf.ed.ac.uk/about-course/

 

[dowie]

again I'd suspect it would be a bad idea to pursue a PhD purely to try and land a job, you're looking at different areas too applied mathematics vs 'data science' - if you've got a couple of years left maybe figure out what area you enjoy the most, that is probably the best option to get you through an MSc or PhD, not just picking an area because it is hot right now. That applies to jobs too, do what you'd enjoy most as you're more likely to excel at it.

I think if you did do part III you'd likely still be able to go for quant jobs without a PhD(you could always test this with a fake e-mail account, burner phone and CV under some random name... see what sort of response you get in terms of interviews with a profile similar to how yours might be in a year or two)

if stats/data science does turn out to be your thing and you want to apply it to finance then this group at oxford looks interesting, it is sponsored by a massive hedge fund:

http://www.oxford-man.ox.ac.uk/

then again if you're keener to solve PDEs then the first two links you've posted might be more appropriate

anyway, main point of my post was just to emphasize that doing what you enjoy is fundamental else you're highly unlikely to work at it in the same way or to the same level

 

[Muz333]

it would be a bad idea to pursue a PhD purely to try and land a job

I'm in a different field doing a PhD which I'm in the last few months of. I would agree with this statement based on my own field. A PhD is normally a route into academia as opposed to getting a head start in industry.

It may even be the case that you end up overqualified for any entry positions in industry but lack the real life experience for the higher up positions.

 

[Quartz]

I'm biased because I should have become an actuary, but I'll point out that data science is incredibly specialised and a new field which may not last whereas there have been actuaries for centuries. And is there not a large element of data science in being an actuary?

 

[Ammo1337]

again I'd suspect it would be a bad idea to pursue a PhD purely to try and land a job, you're looking at different areas too applied mathematics vs 'data science' - if you've got a couple of years left maybe figure out what area you enjoy the most, that is probably the best option to get you through an MSc or PhD, not just picking an area because it is hot right now. That applies to jobs too, do what you'd enjoy most as you're more likely to excel at it.

I think if you did do part III you'd likely still be able to go for quant jobs without a PhD(you could always test this with a fake e-mail account, burner phone and CV under some random name... see what sort of response you get in terms of interviews with a profile similar to how yours might be in a year or two)

if stats/data science does turn out to be your thing and you want to apply it to finance then this group at oxford looks interesting, it is sponsored by a massive hedge fund:

http://www.oxford-man.ox.ac.uk/

then again if you're keener to solve PDEs then the first two links you've posted might be more appropriate

anyway, main point of my post was just to emphasize that doing what you enjoy is fundamental else you're highly unlikely to work at it in the same way or to the same level

I don't plan on doing a PhD just for a certain job, but because I want to get into research and possibly end up using said research in industries. R&D is also something that I'm quite interested in post PhD. I just wouldn't want to be churning out papers and research, I'd like to be in the field too.

I'm in a different field doing a PhD which I'm in the last few months of. I would agree with this statement based on my own field. A PhD is normally a route into academia as opposed to getting a head start in industry.

It may even be the case that you end up overqualified for any entry positions in industry but lack the real life experience for the higher up positions.

How was your experience? What are your plans? I hope you make it!

Yes I've heard about this, one of the PhDs programmes I have posted up there feature 10 week projects at companies, previous students have worked at oil and metallurgy companies in norway and gone to the netherlands etc. I'll also definitely be looking for internships and I'm doing so right now, all the way from now to the end (whether that is PhD or masters)

I'm biased because I should have become an actuary, but I'll point out that data science is incredibly specialised and a new field which may not last whereas there have been actuaries for centuries. And is there not a large element of data science in being an actuary?

What did you end up becoming and what do you mean by you should have become one?

There is, but one thing that concerns me is that I haven't been able to get a glimpse into the kind of models and maths they use.

 

[Muz333]

How was your experience? What are your plans? I hope you make it!

Again, I have no idea how it would be in your field but I do my research in Synthetic Biology constructing artificial gene networks to engineer organisms to complete a set function.

In this I collaborate with physicists/mathematicians as I do quite a bit of in silico modelling to predict the function of network topologies. The physicists come into again as we've made a microfluidic platform that required knowledge of fluid dynamics especially as micro scales where friction overtakes the system.

I'm due to finish at the end of September and currently have one paper published and two ready for submission in the next couple of weeks. Additionally I have some early but promising data that could lead to writing up a fourth after I do complete my thesis.

My experience has been...eye opening I guess. While I've definitely enjoyed a lot of my research it quite commonly involves working up to 100 hours a week for a few months on end at peak times. Put it this way I could count on my hand the amount of days I've not been in the lab this year. As for holidays? No chance, I got told no time off for the last two years of the research project.

Now this is manageable when everything goes well and you have exciting results to report. However, research rarely goes like that, or at least those times don't last forever. There will be times where you hit a wall as you attempt what feels like a sisyphean task while everything refuses to work. Now trust me, working seven 12 hour plus days in a row for a big experiment only to realise it failed and having to repeat it, maybe even four or five times grinds away at your very soul.

All in all I'm glad I've done this PhD, that said I'm running straight to industry as soon as I finish. Continuing down the academia route means working way too many hours, temporarily contracts with no security and having to move constantly to chase down those few jobs.

 

[Ammo1337]

Again, I have no idea how it would be in your field but I do my research in Synthetic Biology constructing artificial gene networks to engineer organisms to complete a set function.

In this I collaborate with physicists/mathematicians as I do quite a bit of in silico modelling to predict the function of network topologies. The physicists come into again as we've made a microfluidic platform that required knowledge of fluid dynamics especially as micro scales where friction overtakes the system.

I'm due to finish at the end of September and currently have one paper published and two ready for submission in the next couple of weeks. Additionally I have some early but promising data that could lead to writing up a fourth after I do complete my thesis.

My experience has been...eye opening I guess. While I've definitely enjoyed a lot of my research it quite commonly involves working up to 100 hours a week for a few months on end at peak times. Put it this way I could count on my hand the amount of days I've not been in the lab this year. As for holidays? No chance, I got told no time off for the last two years of the research project.

Now this is manageable when everything goes well and you have exciting results to report. However, research rarely goes like that, or at least those times don't last forever. There will be times where you hit a wall as you attempt what feels like a sisyphean task while everything refuses to work. Now trust me, working seven 12 hour plus days in a row for a big experiment only to realise it failed and having to repeat it, maybe even four or five times grinds away at your very soul.

All in all I'm glad I've done this PhD, that said I'm running straight to industry as soon as I finish. Continuing down the academia route means working way too many hours, temporarily contracts with no security and having to move constantly to chase down those few jobs.

Sounds very interesting, especially your collaboration work. I guess that is the volatile nature of research, and overcoming setbacks and failed experiments which is what gets to most people. I won't be in any labs.. much

The lack of holidays is a bit meh, but I would hope that they could give you a bit of time off if an excellent external opportunity arose like an internship or so.

Good luck to you, what industry are you breaking into? :0

 

[Quartz]

What did you end up becoming

I went into IT.

and what do you mean by you should have become one?

I wasn't aware of the profession until I was in my 40s, and it would have been a perfect fit for my numerical skills and my personality.

 

[dowie]

people still become actuaries later in life AFAIK, no reason why you couldn't try and become one in your 40s

 

[Muz333]

The lack of holidays is a bit meh, but I would hope that they could give you a bit of time off if an excellent external opportunity arose like an internship or so.

Good luck to you, what industry are you breaking into? :0

It's really a seven day a week affair, but we microbiologists often have to work on the schedule of our yeast, worms or whatever cells we work with.

I would aim to get into a Biotech industry. They tend to work with biotherapeutics, drug testing and R&D. Quite a lot up in Aberdeen here but I'd really want to get out of this city if at all possible!

 

[Quartz]

people still become actuaries later in life AFAIK, no reason why you couldn't try and become one in your 40s

That time has passed. I'm nearly 50 now. I'd need to relearn mathematics, then go to university for 3 years, then convince a firm to take me on. Who's going to take on a 55-year old apprentice? I'd be nearly 60 before I qualified.

 

[dowie]

you're probably pushing it, but while you'd need to brush up on your maths skills if they're rusty you wouldn't have to go back to university

you've got 15-20 years of working life left if you're not yet 50, there are people completing OU degrees who are older than you, no reason why you couldn't have a go at the first couple of actuarial exams, you don't even need to register with the institute/faculty to take the first one which seems to mostly be fairly basic financial mathematics - the sort of stuff you'd cover in say the first week or two of a grad program at a commercial bank or similar and the second exam seems to be mostly rote learning

you can take a look here at CT1 and CT2

https://www.actuaries.org.uk/studyi...llowshipassociateship/core-technical-subjects

some of the later 'core technical' ones seem to be more like undergrad mathematics and statistics and I'm sure the exams beyond that are probably more at a post grad level, but the first two seem pretty approachable with just a bit of self study... no harm in doing one or two and approaching some firms... depending on your background and whether it represents a significant pay cut maybe you could get a job doing some grunt work for a small consultancy etc...

 

[dowie]

quick google reveals this:

http://www.actuarialoutpost.com/actuarial_discussion_forum/archive/index.php/t-150051.html

second reply is from a guy who started aged 50

 

[dowie]

I went into IT.

in particular this appears to be a plus

you won't find much resistance to an older job candidate as long as you have an exam track record. If your current work is something you can spin as an asset (insurance related, or with a strong IT bent, or some such) it could go quite smoothly AFTER you prove you can pass exams.

from the link above

it seems to me that, given the high failure rate of young 20 somethings with these exams, if you've got some useful experience in IT and can prove you can pass some exams by taking some on your own then actually some employers would take you seriously