• 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!

hi, everyone, I need some advice : statistical arbitrage

Joined
3/16/08
Messages
2
Points
11
Hi, everyone. I am currently a PhD student in statistics and I really want to be a quant. I heard that there two types of quant, and one is more related with statistics. I wonder what is that? What kind of knowledge and techniques are required. Does it need C++ also? Anyone can help me with this? Thanks!
 
Hi, everyone. I am currently a PhD student in statistics and I really want to be a quant. I heard that there two types of quant, and one is more related with statistics. I wonder what is that? What kind of knowledge and techniques are required. Does it need C++ also? Anyone can help me with this? Thanks!

Desk Quants for IBs, and quants for hedge funds require strong statistic skills. :)

It's would be a big plus if you can code /understand C++. But I think quants somewhat have to code more or less.

:) Good luck.
 
Hi, everyone. I am currently a PhD student in statistics and I really want to be a quant. I heard that there two types of quant, and one is more related with statistics. I wonder what is that? What kind of knowledge and techniques are required. Does it need C++ also? Anyone can help me with this? Thanks!

There are people who say that you must know C++ and there are those who say you don't need C++. My opinion is that you need C++, however, the amount of C++ you need depends on your position. Also, I think that all quants must know statistics as well, but many people will disagree. As a future statistician, I plan to make some money using statistics :)
 
If you want to be a quant analyst, I would say you need to be more proficient in the use of analysis tools such as Excel, Matlab etc... Quant analysts need to prototype models and tools such as Excel and in particular Matlab are superb for this. Then on top of all that knowing C++ can only be good for your CV. However C++ is mainly used for writing large scale software for speed used more by software developers and quant developers.
You need to work out which route you want to follow. If you want to use your statistical skills then I would suggest a Quant Analyst role. There are Quant Analysts in the company I work for, for example, who have learnt C++ but almost of them use C++ in a secondary role. There primary focus is using and taking advantage of tools I mentioned above. If you enjoy software development and visualise yourself as developing and implementing models then take the Quant Developer path.
Regards,
Imran
 
Hi, everyone. I am currently a PhD student in statistics and I really want to be a quant. I heard that there two types of quant, and one is more related with statistics. I (*snip*)

There is one particular type of quant jobs that is much more statistical than all the rest and the math for it is completely different than your typical quant job. If you work in statistical arbitrage (a.k.a. "stat arb") or in other high-frequency trading areas then you'll be using a lot of statistics and econometrics. As someone who has interviewed for a few of these jobs before, my experience is that the interviewers are *very* interested in both your statistics and your C++. Unfortunately for the typical MFE grad, they could care less if you know a half-dozen different ways to derive Black-Scholes; stochastic calculus never even came up.
 
Joe, what do they usually ask? Stat Arb is one of the areas I consider getting into :) and I think I've mentioned adding a Stat Arb course to our MFE curriculum like a dozen times.
 
It is so good to get so many replies and suggestions. I love this board!
 
Joe, what do they usually ask? Stat Arb is one of the areas I consider getting into :) and I think I've mentioned adding a Stat Arb course to our MFE curriculum like a dozen times.

Hi Yuriy,

I think a solid handle on financial time series analysis would be good foundation for a career in stat arb. If you understand things like cointegration, serial correlation, vector autoregressive models, you are headed in the right directions. So there are courses that are indeed stat arb courses, just not dressed up as such ;).
 
John, thanks for your pointers :) It seems to me that AR, MA, ARIMA, GARCH and other similar things are "baby stuff" compared to what is needed for stat arb. It would be great to have a course that would help to sort out things needed for stat arb, just like you pointed out. I know one stat arb course offered at NYU and another at CMU, any others?
 
Hi Yuri,

I know those sorts of models may sound like baby stuff... however, in the interviews the thing I kept hearing about repeatedly was execution time. If that is the case, it could very well be that the underlying models are just simple, it's just that they need them to run extremely fast. Of course, I could've been mislead since these were just interviews - the interviewers certainly wouldn't have spilled the beans on what they're doing in such a context. And since my resume plays up my programming skills, there's a definite selection bias going on.

I should also warn you that some of the interviewers for these companies didn't even know what an MFE is! There also seems to be plenty of demand in this field for people with electrical engineering backgrounds in signal processing. I'd be surprised if any MFE curricula covers that.

As far as your question on what you get asked about - presuming my experience is typical, expect questions on your background in statistics, a lot of C++ questions (multithreading is a plus, STL is de rigeur), and questions on whether or not you've used R/SAS or something similar. Plus the usual brain-teasers.
 
Joe, thanks a lot for your explanation. Now, I'm starting to get the whole picture.
When I said "baby stuff", I meant intruductory time series models like AR, MA, ARIMA. They are nowhere near to what John mentioned in his post. I took STA9701 - Time Series - at Baruch where we covered introductory models, but we were not even close to cointegration and vector autoregressive models. And your post confirms that there is much more to learn.
 
Thanks for the R Ref Card Yuriy.

So that you know, we used R extensively in my group. It is primarily used to calibrate factor models to calculate Risk of portfolios. We do a lot with R so I'm getting used to it now. Not so much TS analysis as regression analysis, principal component analysis and a lot of cooking with Matrices (covariance matrices, exposure matrices and some other stuff).
 
Alain mentioned another possible application of statistics in finance. Someone who has taken a course in Multivariate Statistics knows regression (simple, multivariate, multinomial), factor analysis, principal component analysis and can be of great help in portfolio management or risk management.

In fact, there is STA9705 offered at Baruch titled Multivariate Statistical Methods, which I took a while ago. It includes a project where students are supposed to produce something meaningful from what they have learned in class.
 
Back
Top