Hi all,I have a couple of queries for your good selves. My background is in physics (BSc Hons.) and electrical/nuclear engineering, and I'm just about to complete a doctorate (Eng.D*) focused heavily on applied machine learning techniques in the energy industry. Throughout my studies, I've worked extensively in ML and have good experience in developing techniques with R, Java and MATLAB. With this in mind, I'm looking at taking my skills and applying them to a potential career in quantitative finance.In looking at some of the advertised positions, I've noticed a few HH agencies are looking for 'machine learning quants', 'big data quants' or similar, with fairly general job descriptions and very little in terms of solid requirements, even for quantifiable skills like programming. This is in contrast to the rest of the reading I've done on these forums and beyond, where there seems to be a quite well-defined path for a good quant candidate as the competition is fierce.(1) Do these 'machine learning' or 'big data'-focused quant positions even exist, where I'd perhaps compare more favourably against the competition with my experience? I get the feeling that without formal qualifications or experience in the topics covered in, say, MFE programmes, I'd struggle to compete with even the average candidates for the vast majority of quant roles at this minute. Are these positions just HH agencies mining for CVs with buzzwords? Some of the descriptors even stipulate that experience in finance isn't mandatory, just an interest to solve problems, which seems ludicrous.(2) My school - University of Strathclyde - is respected in terms of engineering, but not so much generally in the UK. Would I not even be considered for roles when placed against Oxbridge, LSE and Imperial candidates? If so, would it be worth my while pursuing an MFE or similar in order to improve my qualifications?* For clarity, the Eng.D (short for engineering doctorate) is a doctoral-level scheme equivalent to the PhD. Essentially, it's PhD in engineering with a strong industrial component, working closely with a sponsoring company who benefit directly from the research.Cheers for your time.

Get stuck in a find out.Expect a high percentage of recruiters to be no good. But sometimes they are gold. You have to use your intuition on that. Somtimes 'big data' means an IT role involving masses of SQL and problems with systems having completely fucked integration. Most roles involve a little bit of this 'data plumbing' but if the amount is too high: Avoid.There are plenty of interesting data science issues in finance right now however: particularly relating to market 'tick' data. So my advice would be to gravitate to anything that is to do with trading rather than 'big data' per-se. Dont underestimate the importance of getting out there in person and assesing the lay of the land. Dont completely rely on the recruiters for this: Their interests are not completely congruous with yours. You are correct that many roles recruiters post look lush, but are just bait. They then try to switch you in to whatever mandates they have lying around. The scenario can rapidly degrade from 'algo trading at major hedge fund' to 'matlab contractor' or 'operations system analyst'. BTW not necessarily knocking those former two roles, but be clear what you are getting.

QuoteOriginally posted by: neuroguyGet stuck in a find out.Expect a high percentage of recruiters to be no good. But sometimes they are gold. You have to use your intuition on that. Somtimes 'big data' means an IT role involving masses of SQL and problems with systems having completely fucked integration. Most roles involve a little bit of this 'data plumbing' but if the amount is too high: Avoid....Dont completely rely on the recruiters for this: Their interests are not completely congruous with yours. ...OP I would highly stress the above points by neuroguy. be very careful abt the big data roles - many, maybe even most, are just DBA roles adapted to the big data space. they pay well but the work is just as dull.and definitely be weary of recruiters, they'll tell you everything you want to hear to get their commission. not all but most, anyway.

Thanks for the responses folks. I'm at a fairly early stage of applying for roles, so haven't actively engaged with recruiters yet. I had the idea I needed to be careful but your advice has solidified that in my mind more, thanks. I suppose the best way to approach this is in asking directly for more details about the expectation and nature of the roles (or indeed if they even exist).

It should become quite clear quite quickly.Its all pretty benign, but the standard advice applies:- Dont give them details about other places you are looking at- But don't let recruiter B represent you again somewhere you have already been introduced by recruiter A (or indeed yourself)- You are under no obligation to give them any information you don't want to- Dont give your CV out to recruiters willy-nilly, i.e. only supply it to ones that seem to have a genuine role.Of course the exceptions to this are if you have a good recruiter who you want to keep onside by giving them useful information (but only in such a case as it will not negatively effect you).This is not the hardest part of getting a job, so don't sweat it too much!

Headhunters stockpile bidders in generic roles. You want a fairly precise role. There is a good chance you could headhunt yourself.1) Log into hedge-fund performance websites to get lists of well-capitalized traders,2) Don't be afraid to go around sites like this, and say you are a machine-learning phd looking for a job, willing to relocate, whatever.I'm not going to hire you, but I am curious: For what amount of money would you work hard? For what amount of money would you punch six hours doing company stuff, then go surfing? What cities can you relocate to?

Hi and welcome to the forum. Care to share some words on what you were doing in your thesis?? Stale

Hi neuroguy,Can you mention some of these interesting data science issues relating to market 'tick' data? I would really appreciate it, because it is hard filter out some useful information from the internet (most of the material is either superficial or IT oriented).Thanks in advance!

There are 2 areas I have come across:1) Trading: using tick data (i.e the actual quote-by-quote data stream coming of the exchanges) to inform systematic trading strategies. On the shortest timescale ('high-frequency' trading) this more or less amounts to determining the short term expected market direction and then betting accordingly. On the longer timescale (eg. systematic commodities trading advisors) this amounts to using raw tick data to build timeseries models to identify trends and reversals. 2) Algo-Trading: using tick data to build models that enable you to trade with minimum 'market impact'. Banks and other vendors sell execution 'algos' that predict the cost of moving into or out of a position. A simple example might be that trading 100,000 shares of some name in the morning session might be rather expensive (in that you might move the market and hence obtain a less favourable average price). If you know a bit more about liquidity you might find that splitting the order 50/50 on the open and at the close gives a better average price. Its possible the make this much more complex, but the basic idea is the same. If you can model market liquidity you can reduce trading costs. Many firms are also bringing this kind of work in-house and seeking to integrate it into the wider investment processes. Both of these endeavours involve modelling liquidity, which is an interesting area IMO.

QuoteOriginally posted by: farmerI'm not going to hire you, but I am curious: For what amount of money would you work hard? For what amount of money would you punch six hours doing company stuff, then go surfing? What cities can you relocate to?I'd be looking to remain in the UK, so London essentially. Money-wise, I've seen ranges of values for commonly offered salaries so putting a number on it would be difficult for me at this stage. I can't see myself ever taking a job with too low a salary in my mind to warrant not working hard, though; the issue for me in that instance is probably more about the role itself.QuoteOriginally posted by: StaleHi and welcome to the forum.Care to share some words on what you were doing in your thesis?? StaleMy thesis looks at the application of a variety of machine learning techniques to various diagnostic and prognostic problems in machinery condition monitoring, specifically rotating machinery in the energy generation context. There's a variety of reliability and operational tasks that were identified at the outset of the project for decision support development - and effectively my work has been looking at these tasks and creating a toolkit to tackle them in scenario. This can range from simple behavioural classification through clustering, automatic changepoint extraction from large volumes of data or investigating the potential for a data-bourne degradation metric using SVMs.A lot of the work is focused on taking outputs from ML techniques (coded a variety, but quite a bit of kernel methods) and presenting any findings in a useful manner to the domain. This has also involved a little work in statistical inference metrics, generally looking at issues surrounding data suitability and any assumptions used when selecting historical behaviours for study.