GRE Math Subject test for admission to statistics grad programs?

Question

I am an international student from an unknown school with a sub 3.0 gpa. I am a CS major but I certainly have a deep interest for statistics grad programs. In the future I want to work on algoritmic trading and want to involve with topics such as time series, machine learning and other statistical techniques used in finance.

Some people I have talked with about this issue recommended me to get a PhD from a top 20 statistics program. But I don't have much math classes on my transcript, actually numerical analysis and ODE's are the most advanced math classes which I had. Also I had a traditional probability and statistics for engineers class.

Currently I am self studying through Walters Rudin's principles of mathematical analysis book and plan to involve further with math. I plan to self study undergrad level topology from munkres, abstract algebra from artin, and some advanced linear algebra and functional analysis may be some measure theory based probability etc. I can attend a university for these classes but it would be extremely difficult for me to manage it just because money constraints which I have. I have to admit that I learn better when I self study and I usually try to attempt most of the exercises in the books.

The problem is that I can't prove that I studied these topics except a good score in math subject gre test. But this test is not a good indicator of abstract math knowledge and most of the test is about calculus.

Does a good score from Math GRE carry a value in MS level admissions to a decent thesis based statistics program with some funding which will help me to get a PhD from a top 20 school later. I don't have any intention of applying to CS grad school because I do not want to get any systems related course as a requirement and majority of the classes I am interested in are mostly offered by math or statistics departments except some machine learning classes offered by CS departments.

So what is your point of view for GRE Math subject test results ? What score you love ? May be I can pay to a college some money for attending to one or two advanced courses such as differential geometry and a grad level real analysis course. If I do well on these courses in what level will they help for admission with some aid ?

Answer 1

I would say that a good score on the Math GRE subject test is always a good signal. That being said, test scores in general have relatively little importance in PhD admissions -- they can disqualify you, but barring that, won't do much to get you in the door. Its something to optimize only after you have optimized the other, more important parts of your application. In your case, it might be helpful to alleviate any concerns someone might have about your math background.

By the way, you might want to check out CMU's PhD program in Machine Learning: http://www.ml.cmu.edu/ -- none of those pesky systems course requirements!

Answer 2

You should check with the statistics departments in question about their admission requirements. In some cases, it may be required; in others, it may not be.

As for recommended scores, that's even harder to say. Different schools will expect different results, depending on the caliber of students they attract. In any case, though, you should aim to get the highest score you can, rather than aiming for a particular target. But at a minimum, you won't want to show a score that results in a below-average score; that probably won't help you at any competitive program.

Answer 3

1) Most stats programs do not require math GREs.

2) The majority of math GRE is not from advanced mathematics. I strongly suggest you read up on it.

3) Stats programs generally do not require a lot of math background. A solid background in linear algebra, calculus, and multivariate analysis, along an upper level stats sequence is more than sufficient.

4) Your references and any research you've done are going to be the important factors. Did you distinguish yourself in your math classes? Were you the top student? If not, applying to a top 20 school may be an unrealistic goal for you.

Answer 4

A GPA of less than 3.0/4.0 will make you ineligible for any top 20 grad school in the US (and Canada), no matter if you apply for a Master's program or a PhD. The lack of math is not as big of a problem as your GPA. If a school accepts you but doesn't think you have enough math in your background, they will put you in a qualifying year instead of the grad program directly.

A good GRE score is good to have, however the cut-off for considering your application is 3.0. If you don't have a minimum 3.0, chances are your application will be disqualified no matter what your GRE score is. These cutoffs are stated clearly at each school's admission requirements page.

For admission consideration, your only option is to talk to the supervisor at your current school and ask if he/she knows anybody at a US school and could put in a good word for you. Otherwise, I am afraid that you are wasting your time.