switch from quant to machine learning/IT

Switch from quant to machine learning / IT


  • Total voters
    28
Joined
10/23/12
Messages
6
Points
13
The following post isn't meant to offend anyone who is currently working as a quant, I just want some advice regarding my current situation.

About me: MSc from a top school in computer science and a pretty solid statistics and programming background, plus 1.5 years working as a quant in market risk for a large bank. I like my colleagues, the pay is great, and the work is relatively interesting: I get to build and implement new models (although basic in nature, I still find them interesting), I work with traders, etc.

I also have a fairly good grasp of machine learning algorithms, and made it through 3 rounds of interviews at top IT firms. So, from a cynical perspective, if I want to quit finance then I sure can.

The problem: I just don't see how what I do in finance on a daily basis benefits anyone. The actual models which we come up with are no-where close in estimating the real risk (in my opinion). It's almost as if we're playing charades, and are creating models as an elaborate smoke-screen for the regulators. We don't create stress tests. We don't think of creative market scenarios where shit can really hit the fan. We don't work with traders to the level of influencing their trading decision. The list goes on.

I've thought of switching fields within finance, to even something like pricing. I hear similar negative "why are we doing this" messages from other friends who are in pricing; most common seems to be "why risk neutral and not risk averse pricing models?". Those who work with traders claim that they're only doing this for the money because traders have no souls (they are not fully enjoying their work days).

Perhaps I am being too cynical, but in summary, most people whom I know are really just being quants because of the money: they needed a job, banks had more openings than their industries and they made the switch.

In my case, however, I could either go as a programmer for a top IT firm, or as a machine learning scientist. The pay which I got offered is about the same, and what I like about the job is that they are actually building a working system, which does stuff. They are manufacturing something, not creating a mathematical smokescreen for the financial regulators. They deliver a working product.

So, should I try working in a different area? I figure I can always come back to be a quant.

Thanks for all your input!
 
This is a very good question worth discussing. There is nothing wrong with chasing after a high paying job. Most money chasers never make it a career in the long run. If you have the talent and love to do something you feel productive to society, maybe look into this.
http://bits.blogs.nytimes.com/2012/10/23/creating-big-datas-talent-mart/
Great read, thanks!

The "money chasing" that I see is purely promotional: jump from one place to another, taking a higher position each time. In 8-10 years, most of these folks end up as VPs in some area, where the pay is decent, but the usefulness and intellectual interest of the job is questionable. Personally, I'd rather be getting half their paycheck, but doing something I really like.

I also somehow don't see myself applying the financial regulator knowledge to make my own scalable business one day, unless it's consulting, which is not scalable and is tiring.
 
The bottom line in this business is you are compensated to make money for your employer. So let's get that out of the way.
Now, finding an interesting career where your technical knowledge is valued and you are doing something enjoyable everyday is a great question to ask.
I personally would enjoy having my own business and take my own risks and have my own freedom. When you are employee, you are paid to do what your employer wants.
 
Interesting discussion. Prokyon, is it that you are saying you would be in finance IF you were making the trading decisions (or, directly influencing them) based on your quant models? If so, have you tried to get into that area? Most quants I know express the same/similar frustration/disenchantment...and, some others believe that they have some model/algo/strategy that they can mint money with, but for some reason never seem to be doing that.

What you need to ask yourself is: what would you enjoy to do, IF your networth is - not saying "was/were" becuase it could very well be and I don't want to insult you by saying that you are not already worth that much :;)- say, $100M? Then, if you are not already worth $100M, still assume that you are and start your journey in that path. Then, sooner or later you will know if you are happy or not doing that and if not happy, repeat the above. At some point, you would be doing what you enjoy...and, that would definitely make you excel and succeed to your heart's and wallet's content. Just my two cents.
 
Interesting discussion. Prokyon, is it that you are saying you would be in finance IF you were making the trading decisions (or, directly influencing them) based on your quant models? If so, have you tried to get into that area? Most quants I know express the same/similar frustration/disenchantment...and, some others believe that they have some model/algo/strategy that they can mint money with, but for some reason never seem to be doing that.

What you need to ask yourself is: what would you enjoy to do, IF your networth is - not saying "was/were" becuase it could very well be and I don't want to insult you by saying that you are not already worth that much :;)- say, $100M? Then, if you are not already worth $100M, still assume that you are and start your journey in that path. Then, sooner or later you will know if you are happy or not doing that and if not happy, repeat the above. At some point, you would be doing what you enjoy...and, that would definitely make you excel and succeed to your heart's and wallet's content. Just my two cents.
The problem with being around people where I get to influence trading decisions is that those jobs are extremely high-stress and long-hour jobs, from what I can tell. Plus the people there are not of the nicest kind.

I like computer programming and building working systems. Somehow I went astray into finance, because most of the people I knew went into finance and because it was a nice area to get into at the time. So I think I'd rather try switching fields for the time being to see if IT grows on me.
 
Good articles! "something-it" to "yipit" with a computer program lol I like :)

Here's a similar discussion, also by me. I seem to be getting a lot more responses there:
http://www.nuclearphynance.com/Show Post.aspx?PostIDKey=163607#163633
 
Glad to see this discussion. Really happy with a few of your comments, Andy and prokyon. Refreshing to hear people talk about passion for a subject.

I investigated IT as a career option earlier in the year and what I found is while there is a perceived shortage in the sector, most vacancies are due to technologies, particularly mobile development, moving so fast that people can't keep up and that they are fleeing tough roles. This + no training budgets has led to employers being more demaning when hiring, which means 2-5 years experience in IT is an absolute must and "getting the foot in the door" is far tougher than ever (even for quants). One article here explains a bit: http://www.i-programmer.info/news/99-professional/4551-shortage-of-java-developers-.html

Also from a working practices perspective MFE's may want to consider why they want to program (whether as a quant or in IT). Usually it's because they programmed non-stop in college and had a passion for it and the time to debug perfectly, but coding in the workplace is tougher. Having to look at hundreds of lines of code, silly errors crop up and lack of effective communication with project manager/client can slow down production and make a programmer's life a misery. This is something not always seen until you're in the job and varies from one place to another. In the case of the OP I think this isn't an issue as he's clearly succeeding as a quant and surely this skill will transfer into IT.

I would be very interested if people have differing views (as the above may be partially wrong, or at least wrong in certain sectors). For instance is .NET programming a place where a career changer might move in easier, or are there clever ways to get in? What about retraining options for someone like myself that hasn't coded professionaly in a while?
 
There is such a thing as employers asking for 2-5 yrs experience even for junior roles. However, there should be a way for you or a recruiter on your behalf to sell yourself as 'being able to learn fast' and thus get that prerequisite 2-5 yrs entry level experience for the next job. Many people build their experience in this way. Another way is to do contract or freelance work.

The thing about IT is that it changes fast; even moreso than Finance. My friends in IT say that the only 'escape' from frenetically having to learn the next thing is to move into management which obviously everyone can't do.

Coding in a big group versus your own pet project is a change, but having others to work with also means you are not the only 'throat to choke' if something doesn't work. "Hundreds of lines" of code would be something of an understatement too.

This probably not the case in Europe, but technical people in general are doing well in the US from what I can tell:
http://money.usnews.com/money/careers/articles/2011/06/20/job-market-sucks-not-for-techies

There are also some IT Job sites on one of my Google+ pages if it helps:
https://plus.google.com/b/115737827211339621389/115737827211339621389/posts
 
Nice articles! The list of job sites is really helpful, I'll be sure to check it out when I'm ready to interview.

I'm pretty much sold at switching to IT, but the question is how do I get into more ML-type roles. I have pretty basic ML background. I can review C++, data structures, statistics and ML (hidden Markov models, mixture of Gaussians, logistic regression, Monte Carlo methods (rich background in MC methods), Naive Bayes).

From what I've seen so far though, most ML positions already require knowledge of collaborative filtering, for example. Some roles already require Hadoop/MapReduce as well. Any ideas? I need to find more basic ML roles to start.
 
From what I have seen mostly ML roles ask for Phd or relevant experience. Without such background you only get closer to ML in Data Mining where you mostly use ready made software which also get boring after sometime.
 
I don't agree with the above. Go to www.icrunchdata.com and search for ML jobs and you will find some jobs that do not require PhD (and some that do) and they are advertised as real 'software' jobs (i.e. not using canned third party tools).

I agree that a PhD has more opportunities in this field though.
 
Hi Prokyon,

Why don't you try to join the best of both worlds ? there are many IT companies that develop serious software for risk management, pricing, etc.
Maybe you could explore this possibility too.

Good luck
 
Liam is right. Many people think that getting into the quant world will make you millions. unless you are an outlier (smart and outstanding, be able to play the politics and extraordinary good at people skills). I know someone will try to shoot me with a bullet if I say that most traders are actually from the Top MBA schools, with a small portion of quants are in HPT which required more technical skills. However, please bear in mind that no ones in this world is able to estimate the "True" risk unless you live in your wonderland. It's just like one quant was telling me he could create a horse racing model. Come on.....horses are not maths or stats!

If you think that quants are making a lot of money that motivate you to switch career (how much is considered a lot?). Furthermore, the global financial markets are regulated by laws and insider trading is prohibited. It's getting harder for you to make money if you do not have a "trading" network, and this network takes a lot of time to develop. In addition, you need to convince people to play your game with rules specified by you. Can you?

You can ask the average salary from quants in this forum. Frankly speaking, people who make millions will not hang around in this forum. MFE is a product and apparently many universities think that they can sell like a hot cake. Most people are living in their illusions by thinking that they can make money once they obtain the MFE..and know C++ programming. It is far more than this. You should use your intelligent instinct to make a sensible decision instead of asking people in any forum to decide your future. No ones can help you. In addition, some blokes think they are smart enough to give you advice but in fact it's type II error in your hypothesis testing.

If you already have a career in your current field. Why don't you try to excel in that field you already spent years to develop. If you can't succeed in your field and you try to move into other fields seeking a fortune, it's just like random walk mate! My two bits.

P/S: Like Sal Khan left his hedge fund career to start Khan Academy. He doesn't make a lot of money (I think) in Khan Academy but he is happy.
 
@Shanatanu: strongly disagree. The only area which might require a PhD, and other credentials, is consulting of any sort. They need people on the frontline who are easily marketable, and titles matter there.

For more technical roles, it’s what you know and what you can do. Your degree/educational background/looks don’t matter one bit. I speak from experience.

MonteCristo: sure, but I don’t know of any software shops like that which do machine learning. They basically re-implement already developed algorithms for pricing / risk management. Code quality is also questionable.

If you know of any software shops which use ML for finance, then I’d love to hear about them! :)

Jocase: accent on “politics”. I can’t stress this enough. Banks are highly-beurocratic and politics matter a lot there. Hedge funds and pension funds are different, but are still heavy on people skills and your ability to manipulate people to move up the ladder.

What’s HPT?

Yeah, you create your network, and therefore create legal insider trading. That’s how winning is done in this game ;)

People, Jocase’s summary is pretty accurate, read what he’s saying a few times over, let it sink in and marinade.
 
Prokyon, Jocase, you are coming across as a little high and mighty.

People are offering to exchange thoughts and advice with you for free (and the whole thread was prompted by Prokyon asking for advice so I don't know what he/she is talking about). This is generally how internet forums work and there's nothing particularly bad or good about reading forums about an MFE or Quant Finance versus any other topic.

No one is under any illusions like 'this internet forum is the source of truth'.
 
Back
Top Bottom