Are Computational Projects Useful For Undergrads Who Want to Do S&T?

It would make you an excellent candidate for an analytics or quant development position at most BBs. Ultimately, those guys are going to be the folks running a lot of the algorithmic trading strategies in 10-20 years, so I say go for it if you've got a computational background.

Thanks illiniprogrammer, but don't most quant groups require at least an MFE, if not a PhD? I was wondering to what extent are these types of projects valued in traditional undergrad sales and trading roles?

If you are looking at any derivatives group you won't get backlash for having a maths background. My desk (and my friends in equity derivs) is almost all masters and phds in math, applied math, physics, engineering, etc.

Nope. To be a quant, you probably need a graduate degree or several years of industry experience. A quant developer is just someone with a CS undergrad and a strong math background- you can get hired in out of undergrad if you're a strong programmer and you've got a good understanding of linear algebra, stats, and calculus. Obviously, algorithms, data structures and OO are foundational as well.

my biggest fear of putting a lotta quant shit on the resume is, during OCR, what if your resume is read by some random chick who is in cash equity sales, and she does not even know wtf C++ is or what the difference between a put and a call is, how would her reaction to such a resume be, vs a resume of some other random chick who was in the same sorority as her? i'm a CS major myself, and I always fear having my resume picked some completely non-quant person on a vanilla desk, and get dinged because I might be "labelled" as someone fit for backoffice or middle office....is it true that only people with quant background themselves value a resume of someone with a quant background? i also have some "quant" related work experience (i interned for some algo trading HF), and if i put some of the stuff i did at work on it, i might instantly get labelled as some type of quant nerd or IT person and that will result in ding

some people have managed to convince me that recruiting for IBD, trading, or whatever at BBs is pretty much like rushing for the greek houses we have in college, and you are more likely to get a job or interview if you appear "likeable" or "cool" on your resume as opposed to actually being smart and having some hard skills....sorry for the rant, i just realized that OCR is right around the corner and i've been getting paranoid lately

Ehhh, 60% of the folks on the trading floor these days have TI-89s sitting in front of them.

The bottom line is that as a CS major, you don't want to go straight to the trading floor. You want to spend a few years as a quant, an analytics developer, quant developer, or in algorithmic trading. You need to learn how financial programming is done, how to build an efficient enteriprise system, how to work with other technologists, and how the models really work. (90% of your time with any model will be spent looking around for someone who can explain the model and its assumptions but discovering that everybody is more clueless than you and being forced to reverse engineer what the hell is going on in the code yourself.)

Once you've done that- and assuming you're not already in algorithmic trading, go back and get an MFE. By the time you've gotten your MFE, some market that you're probably familiar with will be going electronic and some prop shop will be looking for a developer who knows his stuff.

The CBOT is setting up a market for CDS Index Tranches? Hey- you spent three years developing the model for that and running your own pricing system, and now you've got an MFE from Stanford to boot. Why not hire somebody who's already built a system who's only difference from a black-box trading algorithm was that it was unplugged from the market to run the system? Heck, there's only about 50 developers in the world who are as qualified as you to put together the system, and 30 of them have families and don't want to take any risks.

Illini, would BBs try to pigeonhole this kind of background into analytics/quant development? Or would they just consider the candidate more suitable for A/QD, and ask why they prefer S&T instead (so story is important)? As Revsly mentioned, S&T derivatives groups like people with math backgrounds.

I'm speaking from the perspective of someone who is going to be a junior in the fall at an ivy target, and who is going to do OCR and drop my resume into a pile of 300 or 400 (or maybe more, i hope not) other resumes, which will get looked at for not more than 10 or 20 seconds by analysts in sleep deprived states...if only 60% of people on desks use Ti-89s (im assuming for elementary calculations?), then there is a good amount of people who don't even know what C++ is, and my resume might seem very strange and alien to them, as opposed to football players, sorority chicks etc who take BS classes and also have the same GPA as me...i fear that in those 10 or 20 seconds my resume is looked at, relative to all other resumes, i may as well stand out in a good way and get an interview, but i am just as likely if not more likely to get dinged because I come off as an IT/quant guy, as opposed to some random lib arts major who the resume reader is biased towards....after reading threads like the one below, i kinda wished i did not major in CS and math and majored in econ & art history or poli sci like the vast majority of my school :(

//www.wallstreetoasis.com/forums/ugly-truth-about-resume-reviews

Also, Illini, is your recommendation for A/QD/algo->MFE->quant hedge/prop specifically for CS majors (you said "the bottom line is that as a CS major...")? I believe the OP is just doing financial math and not CS. Would someone with a finance+math background (and some data structures/algorithms/OO knowledge, but no assembly/operating system/higher level CS knowledge) still be better off going into development first instead of S&T?

Maybe the initial recruiter, but we have several guys in S&T with CS backgrounds.

That said, it's a waste of a CS degree to go directly into paper S&T.

In 20 years, most of what you see when you go down to the trading floor today will be replaced by banks of servers in a datacenter directly across the street from the CBOE's or the NYSE's network gateway. Institutional investors will get bids and asks directly from servers at different I-banks, and the market makers in every product from convertibles to mortgage bonds to plain old vanilla stock options will all be computers.

Without industry experience as a programmer, you'll never get into algorithmic trading. In fact, my firm's algorithmic trading group looks for people with more than five years of experience as developers, first. I believe GETCO, Jump, and the other program trading firms are pretty similar.

Getting into algorithmic trading without a strong industry programming background several years after graduation is like graduating medical school, going into the cup sales business and never practicing a single day as a doctor afterwards, and then applying as a doctor at Mt. Sinai 20 years later. What few skills you may remember will be outdated, whatever pedigree you might have had upon graduation is irrelevant, and the industry has moved beyond giving aspirin and bed rest as treatment for a heart attack.

Now, if you had at least spent five or six years practicing medicine and knew how the business worked, you could get back in with a fair amount of getting yourself current. But without practical industry programming/development experience (and this doesn't have to be from a bank- actually, some firms prefer Google, Microsoft, and IBM) more than three or four years out, there's no way you're going to get back in as an algorithmic trader.

By graduating with a CS degree and going directly into S&T with no programming involved, you are giving up on your chances of getting into algorithmic trading- where all of the money is going to, anyways.

I'm speaking from the perspective of someone who is going to be a junior in the fall at an ivy target, and who is going to do OCR and drop my resume into a pile of 300 or 400 (or maybe more, i hope not) other resumes, which will get looked at for not more than 10 or 20 seconds by analysts in sleep deprived states...if only 60% of people on desks use Ti-89s (im assuming for elementary calculations?), then there is a good amount of people who don't even know what C++ is, and my resume might seem very strange and alien to them, as opposed to football players, sorority chicks etc who take BS classes and also have the same GPA as me...i fear that in those 10 or 20 seconds my resume is looked at, relative to all other resumes, i may as well stand out in a good way and get an interview, but i am just as likely if not more likely to get dinged because I come off as an IT/quant guy, as opposed to some random lib arts major who the resume reader is biased towards....after reading threads like the one below, i kinda wished i did not major in CS and math and majored in econ & art history or poli sci like the vast majority of my school :(

Yes, you'll knock out a few of the firms run by idiots who don't understand people with any kind of quantitative skill. But you'll wind up at a firm that can use your skills a lot better- a place where you can advance faster.

I'm only a sophomore, so long ways away for me haha. I'm at a finance target which is why I'm biased towards S&T instead of development (and trading also seems more selective out of undergrad, so I'd be getting more out of my 3.9 GPA instead of development where coding ability is probably the only important requirement). I did a lot of CS in high school (data structures/OO mainly, some algorithm analysis and algorithm design for CS contests but that's about it), and do programming projects in my spare time, but I don't plan to do any other CS courses in uni (not interested in computer/operating systems/assembly stuff, and my school isn't great for CS anyways). Thus my uni courses will just be business core+finance+prob/PDEs/stochastics/other necessary math (should I bother with analysis?) while I program on the side to show interest/ability. Looking into quant HF internships for the school year too, though dunno if they would take me. Goal by graduation is either BB S&T, quant HF, quant prop shop, or actuarial (another career I've been considering, more stable than the 3 finance ones), and maybe BB MO risk or something. Might consider BB A/QD now, but I guess I don't want to be pigeonholed as a quant too early in my career (though I probably already am).

So does this sound about right: anyone quant should just go into development since S&T is becoming increasingly automated even if they want to work at a (probably quant) HF later, whereas someone more qualitative that wants to work at a macro or long/short equity HF would probably be in IBD instead of S&T anyways (so S&T isn't very useful for anyone)? Would HF exit opps actually be better for a developer than a trader? And I don't want to be the developer in a HF getting screwed with an essentially-fixed 6-figure salary when the PM with a trading background is making 7-8 figures.

You say you need algo trading experience to actually know how the models work and to be up to date with industry standards. But wouldn't the experience of actual trading in S&T give you a more realistic view of where profit is made, and then you switch to development in a quant HF afterwards to use a semi-automated semi-human method? Sure algo trading is good for arbitrage/market making, but those opportunities will shrink and profit will need to be made on qualitative investment judgment.

But what do I know, I'm only a soph.

So does this sound about right: anyone quant should just go into development since S&T is becoming increasingly automated even if they want to work at a (probably quant) HF later, whereas someone more qualitative that wants to work at a macro or long/short equity HF would probably be in IBD instead of S&T anyways (so S&T isn't very useful for anyone)? Would HF exit opps actually be better for a developer than a trader? And I don't want to be the developer in a HF getting screwed with an essentially-fixed 6-figure salary when the PM with a trading background is making 7-8 figures.

There's not a whole lot of developers in the country who can handle finance, math, programming, and speak English at the same time. Worst comes to worst, you get pushed around by the traders, but you keep your math/finance/programming creds and leave the door open for algorithmic trading.

You say you need algo trading experience to actually know how the models work and to be up to date with industry standards. But wouldn't the experience of actual trading in S&T give you a more realistic view of where profit is made, and then you switch to development in a quant HF afterwards to use a semi-automated semi-human method? Sure algo trading is good for arbitrage/market making, but those opportunities will shrink and profit will need to be made on qualitative investment judgment.

In any case, most algorithmic trading groups refuse to hire folks in from the finance side. Heck, a few of them refuse to hire in even financial programmers and only stick to Google, Microsoft, and Amazon. Without the programming background, it will be very difficult to get hired as a trader who, by definition, writes code to do his trades for him.

In algo trading, your ability to efficiently and accurately get your model to the market is what dictates how much money you make. You can be the best trader in the world, but if your cache runs in O(n) time rather than O(c), you'll show up to market too late. If you don't know how to rapidly and accurately check your code for bugs, you've got to unplug the system and/or set less aggressive bids/asks longer while the other guys are making the big bucks as the market changes.

Like the military, the market is moving from F16s to drones. The engineers and technicians who once reported to the pilots are soon finding themselves as their effective replacements.

IlliniProgrammer:

Yes, but if you tell an engineer how to program a drone to think like a pilot and guess psychologically what an enemy plane would do and how to combat it, they would probably have no idea without actually knowing what it's like to fly a plane in a fight. Sure they could tell it to do something "logical" that is "conservative", but that means it's probably predictable too. Like telling a poker bot to only bet/call with +EV hands: I'll just bet every single hand, and if they ever call or raise then I stop betting or fold because I know they have something.

Made me LOL at work.

Also, think you hit the nail on the head with a lot of the comments in this thread; my experience so far has been consistent with what you've described.

Eh, I wrote that. So is my thinking consistent with your experience, or Illini's thinking?

I understand both sides: trader's would argue what I just said, Illini would argue what he's been saying about automation. My problem is that I can't pick a side because I know both are partially right. Ugh.

Another thing you could do, if it's deep-stacked: just overbet the pot everytime. The bot will never be justified to call because the hand strength/pot odds EV will be poor (betting 4x the pot means they need 80% chance of winning to call, which is only justified when they have the nuts).

That's the thing, if we could get an algorithm to think like that (overbetting/betting more often when the player is too conservative/plays textbook, etc.) then being a developer would be useful. But it's the traders that think up this strategy. Thus they earn more money, and just hire developers to program it for them.

Hopefully, either Illini's arguments or the arguments for the trading side can ultimately win based on logic and not preference, otherwise I still won't be able to decide what to do career-wise. But if one could logically pick one over another without using any preferences to decide, the other job probably wouldn't exist. So it comes down to preference. So then I'll pick the one with more money. Trading yes?

Oh I know, I just used that as an example. I made a probabilistic model myself too, but most simple AI (like poker on Blackberries) just does something as I previously described (with maybe say a 5% chance of bluffing, not evolved to the players on the table). I'm sure one of my higher level math/probability courses will cover Monte Carlos, but I'm only starting my sophomore year so I haven't gotten there yet.

Yeah, and I'm still unsure which to pursue. I don't plan to major in CS though; will I still be considered for an algo trading developer job with double major finance+math+programming ability? But really, how intelligent can algos get? Is it just arbitrage/market making so profit opportunities will shrink as the market becomes more efficient, or do most algos use some sort of AI to generate alpha? Do these systems still require human monitoring? If I were to start my own quant fund, I'd want my algos to be able to make money on its own for a few weeks (or forever...) while I take a vacation haha.

Between these two jobs (analytics/quant development and complex product trading) what's the salary distribution like? I'm guessing the former has a higher median and less downside, but also significantly less upside and fewer high outliers. Which has better exit opps for HF (either pure quant or semi-quant HF) (actually running the fund and making 7-8 figures, not being a developer making 6 figures)? What about job stability?

Yeah, and I'm still unsure which to pursue. I don't plan to major in CS though; will I still be considered for an algo trading developer job with double major finance+math+programming ability? But really, how intelligent can algos get? Is it just arbitrage/market making so profit opportunities will shrink as the market becomes more efficient, or do most algos use some sort of AI to generate alpha? Do these systems still require human monitoring? If I were to start my own quant fund, I'd want my algos to be able to make money on its own for a few weeks (or forever...) while I take a vacation haha.

As with any trader, algorithms always have to have some level of monitoring.

Between these two jobs (analytics/quant development and complex product trading) what's the salary distribution like? I'm guessing the former has a higher median and less downside, but also significantly less upside and fewer high outliers. Which has better exit opps for HF (either pure quant or semi-quant HF) (actually running the fund and making 7-8 figures, not being a developer making 6 figures)? What about job stability?

You get to your maximum pontential earnings by spending your first five years focusing on doing your job and learning everything you can about the markets. In the meantime, figure out how you can add the most valuable competitive advantage to the firm. That might not be in trading- or as a quant developer. You might turn out to be a great risk manager. (A very underrated job, BTW).

Quant developers and risk managers have GREAT job stability. Better than most bankers and traders. I worked in credit analytics but survived a credit crisis and a merger. A number of the people who came in as analysts the same year as me as TAs and salespeople are still looking for work, so I made out pretty well.

What about a position focused more on mathematical modeling and designing new products than quant development and trading (design vs implementation vs execution, math vs CS vs finance)? Would this be structuring? I'm guessing the position typically requires a Master's/PhD though. I suppose as I continue to take more math/finance/CS courses I'll naturally lean towards one. Hopefully.

I'm not sure if I would enjoy risk management, I would want to actively contribute to something that adds revenue instead of managing it. If anything I would just become an actuary instead, since they do a combination of risk management and product pricing/design, and can exit into actuarial consulting (if I want to get out of NYC). The main concern though is pay - actuary upside is somewhat limited to 200-300k. As well, actuarial methodology and predictions are quite uniform and predictable compared to finance, so it can get repetitive (and less inefficiency+less volatility = less profit). They are the most stable with the least hours out of all these options though.