• C++ Programming for Financial Engineering
    Highly recommended by thousands of MFE students. Covers essential C++ topics with applications to financial engineering. Learn more Join!
    Python for Finance with Intro to Data Science
    Gain practical understanding of Python to read, understand, and write professional Python code for your first day on the job. Learn more Join!
    An Intuition-Based Options Primer for FE
    Ideal for entry level positions interviews and graduate studies, specializing in options trading arbitrage and options valuation models. Learn more Join!

Which maths area should I go into?

Joined
5/14/08
Messages
1
Points
11
Hi I'm new here and need a little advice about what sort of maths is needed to work in the investment banking/quant sort of area.

Basically I'm nearing the end of my maths degree and need help choosing an area of maths to focus on for my honours thesis.

My first question is:

Should i focus on something from stochastic processes or statistics?

If I should focus on stochastic processes which area should I do out of Extreme-value theory, time-series analysis, stochastic calculus or Copula (sorry not entirely sure what the last one is)?

Or alternatively, if I should focus on something from statistics, what area should it be from?

Basically I'd like to know which is the most used sort of area in quant

Any advice would be great

Thanks

Squishy
 
I think that if you have taken this courses, it would be okFinancial mathematicsAdvanced analysis Stochastic processLinear RegressionMethods for statisticsProbabilityC++ programmaing
 
Copulas were very popular during the good years of Credit derivatives, so although this has been a valued skill, I think it's price is in decline.

I'd go for this order:
Stochastic processes
Time Series
Stats
Copulas

But be aware that to prosper as a quant you need at least some of all these.
 
mmm... Copula has an interesting definition that might not be proper for this forum :D:D:D

lol this reminds me talking to a colleague quant (whose native language is English, unlike mine) and telling him something like "....so we use this copula to *scratching my head trying to find a right word*...to copulate these functions". :tiphat:
 
Paul Wilmott last night advanced an argument on maths that is similar to my view on Java.

For some years there has been a contraction in the breadth of maths used in the mainstream, and it's possible this has career implications...

Firstly, if some type of maths becomes more useful, the 'older' ones you know will lose value, possibly quite quickly. This is the standard model in IT where a skill may lose much of it's market value in a year. Hasn't happened in quant, but I will not say it can't. This is likely to be slower, but there is a crowd effect that hurts your market price when an area stops being a source of employment and people move to the 'nearest' in search of employment. That applies to both physical locations and types of skill.
It's not pretty.
Been there.
Did not like it at all.

People are are good but not great at maths are the most vulnerable here. Remember that employing a person is like buying a portfolio of futures contracts. The buyer cares about what he sees as his needs after hiring you, not the stuff he's got already. People will be solving SDEs until long after I'd dead and you have retired, I will state that as a fact.

I can't tell you how much they will be paid,

If you are someone who finds mean reversion models attractive for forecasting prices, recall that the hourly rate for this sort of maths has increased more than any other skill in any other profession that I can think of. You want to bet on the price of that going up or down in the next 10 years ?
You are betting more money than you are worth today.

Also, standardisation is part of automation, and since a quant costs a lot more than most bits of software, we are already seeing increasing use of it to make quants more productive.

A rational response is diversification.
To get a job you have to learn the same shit as everyone else, that's the way the system works.

But my advice is once employed, to invest a bit of time in something else.
Even Paul Wilmott who some see as a better mathematician than me doesn't get more precise than "non linear dynamics look interesting". I have a bit of an interest in finite state automata, others like Ayache and Taleb believe that our basic idea of what is 'probability' is wrong, and they are honest enough to say that they don't know what if anything is better.
But ignorance is definitely worse.
 
So I'm guessing most abstract math is out the door?

This discussion is great and highly relevant to me. I'm in a very similar position as the original poster squishy. I want to be a quant, am heading into the spring of my junior year, so have a choice between abstract math and more practical stat classes.

I'm always stuck in choosing the more abstract math classes than the more practical statistics ones and the (bad) thing is that I'm always leaning towards the former. I always have this glorified view (maybe wrong) of mathematicians in Wall Street who walk in a room and devise models mostly from an abstract point of view. They come in and use what they know of measure theory, stochastic calculus, elliptic PDEs to form some theorems, prove it using analysis, and call it a day.

Am I wrong to think this way because honestly, if there's an ideal job I want, it would be this. So doing some research I see in JP Morgan's website the role of Quantitative Analysis is to use all the aforementioned math to build models. Or in Citigroups website where a requirement for a Quant is 'Mathematical Analysis'. None of these job seem to rely on the more statistical, applied math which DominiConnor have mentioned to be more in demand.

I'm not saying that statistics or 'careful' calculations are not important. My main argument is to question the usefulness of this analysis, topology, PDEs math which I hope does in fact have it's presence in Wall Street, at least among a selected few. Then it would make more sense to send a semester in one's final year, or even four years in one's PhD career, to use measure theory to prove the connection Kolmogorov-esque between expectation value and it's related PDE of some SDE equation. (perhaps the quickest example I could think of form my Stochastic Calc class.)
 
I always have this glorified view (maybe wrong) of mathematicians in Wall Street who walk in a room and devise models mostly from an abstract point of view. They come in and use what they know of measure theory, stochastic calculus, elliptic PDEs to form some theorems, prove it using analysis, and call it a day.

Am I wrong to think this way because honestly, if there's an ideal job I want, it would be this.
Once upon a time, these positions are call blue-sky quants/researchers. There are groups within banks that still do this but much of the work is now done by financial services firms that run by ex-quant/mathematicians/academics. (See Ito 33, I met people running this group when they demo'ed their product to my desk). If you are a small hedge fund, firm, desk, you are most likely will use products from an another specialty quant firm. Some big banks create their own groups.
Morgan Stanley has a group called Morgan Stanley Strats and Modeling (MSSM) which is run by Roy DeMeo. This group is part of IDEAS.

I don't know who they hire into this group but probably not out of undergrad.
 
Back
Top