Request for Statistics PHD Profile Evaluation

[ECE23]

Hello guys,

This is my first time posting on this site. I am looking at PhD programs for Fall 2018 in Statistics and I am not sure what schools I would realistically have a shot of being competitive for. Any suggestions on programs I would realistically have a chance of getting in to would be helpful. Obviously, some of my math grades are low, however my mathematical trajectory is upwards. Someone looking at my transcript would see that I took Cal II my first semester in college and that my grades improved each semester(with the exception of my senior year). Furthermore, I feel that an A in real analysis outweighs poor grades in other mathematical subjects due to the magnitude of greater difficulty.

 

My profile is the following.

 

Major: Electrical Engineering

Age:23 yrs old.

GPA: 3.52/4.00

Graduated in May 2016.

Math Courses: Cal 1(AP credit), Cal II(C), Cal III(B-), Diff Eq(A+), Linear Algebra(A), Theory of Probability(B+), Discrete Mathematics(A), Intro to Abstract Algebra(A), Intro to Real Analysis 1(A), Electromagnetics (cal III based) (A), Physics 1 w. cal 1(A-), Physics II w/calc II(A+). Self studied General Topology.

Furthermore, practically every single engineering course that I took involved differential equations, calculus 1, calculus II, and calculus III.

GRE - 155 Verbal/160 Quant(taken on 1 weeks notice, am studying for a higher score currently).

No research to date.

Programming - Taken 4 programming courses in college. Programming on par with what you would expect from an electrical engineering grad.

Work Experience: 1 year of engineering work experience at Chevron through internship. 1 year of engineering work experience at General Electric. Work experience is unrelated to statistics. 1 semester of tutoring calculus and physics.

Letters of Rec: Assume that they are reasonably strong.

I'm currently targeting schools that are 10-20 and I will add a few reaches that are in the top 10 like University of Chicago.

 

Let me know what you guys think

 

Thanks

 

 

 

Edited March 1, 2017 by ECE23

[GoPackGo89]

Why do you want to get a PhD in statistics? I see one semester of probability theory, surely engineers are required to take an engineering stats class? 

[ECE23]

Because I want to do research in machine learning applications and work as a researcher developing strategies for automated stock trading. Statistics is the course of study that best represents the area that I want to do research in. Furthermore, it is also the area of study that best matches my background (engineering with some pure mathematics)

 

 We weren't required to take any specific course on engineering statistics. However, one of my courses involved advanced use of probability theory applied to engineering problems. I am currently self studying stochastic calculus. 

 

[ECE23]

Also I would like to add on to my previous post about why I want to study statistics.

I have spent several months weighing different options and doing research on different course curriculums, cross referencing them with job posting descriptions in computational finance, and speaking with people who actually work in the industry.

 

Statistics is the best path for me to reach my goals. Furthermore, I love mathematics and I really want to study it full time and do research.

[chisquare]

1 hour ago, ECE23 said:

Because I want to do research in machine learning applications and work as a researcher developing strategies for automated stock trading. Statistics is the course of study that best represents the area that I want to do research in. Furthermore, it is also the area of study that best matches my background (engineering with some pure mathematics)

 

 We weren't required to take any specific course on engineering statistics. However, one of my courses involved advanced use of probability theory applied to engineering problems. I am currently self studying stochastic calculus. 

 

 

 

 

 

3

Based on what I know, probably Computer Science is more suitable for both "research in machine learning ... developing strategies for automated stock trading" and "engineering with some pure mathematics (background)" than statistics. If you are aiming for a job in private sectors, they would evaluate you more on "what you can do" rather than "what is your major". And if you search positions posted on glassdoor or linkedin, actually employers don't care whether you are stat PhD or cs PhD or ECE PhD (and I know a couple of ECE and physics PhDs working in Wall Street or HF in Boston now).

Another practical issue is that most professors in statistics departments are not focusing on financial statistics (which is why I had a hard time searching for potential supervisors). And as far as I know, among these professors, not many of them are implementing machine learning techniques, and vice versa.

Edited March 1, 2017 by chisquare

[ECE23]

11 hours ago, chisquare said:

Based on what I know, probably Computer Science is more suitable for both "research in machine learning ... developing strategies for automated stock trading" and "engineering with some pure mathematics (background)" than statistics. If you are aiming for a job in private sectors, they would evaluate you more on "what you can do" rather than "what is your major". And if you search positions posted on glassdoor or linkedin, actually employers don't care whether you are stat PhD or cs PhD or ECE PhD (and I know a couple of ECE and physics PhDs working in Wall Street or HF in Boston now).

Another practical issue is that most professors in statistics departments are not focusing on financial statistics (which is why I had a hard time searching for potential supervisors). And as far as I know, among these professors, not many of them are implementing machine learning techniques, and vice versa.

Chrisquare,

 

Thank you for the reply. I was looking at statistics because I am strong in mathematics(as it is my real passion) and not as strong in computer programming. After speaking with many different programs CS and Statistics, I felt that the coursework in the Statistics programs more closely represented my career goals and what I am actually interested in learning. The idea of taking a bunch of theoretical probability courses and learning stochastic differential equations is exciting to me. Computer science courses, not so much

Additionally, If you look at quant shop job postings they are typically pretty similar. They all say they want people with deep knowledge of mathematics and data science with specific attention given to time series analysis, signal processing, stochastic calculus, machine learning. That's pretty much what statistics is all about from what I've read. Not to mention Quant algo jobs are not computer science jobs, they are really mathematics and statistics jobs focused on developing innovative ideas and backtesting them, with many shops hiring developers to do the leg work and fully flesh out ideas.

 

 Furthermore, Statistics programs prioritize candidates with strong theoretical mathematics backgrounds to the point where this is, actually, the number 1 criteria for most programs. This is not true for CS programs. For example, I called Statistics @ Texas ATM the other day, the first question the lady asked me was "how much real analysis do you know". CS departments could care less if I've taken real analysis. They are more interested in my portfolio of programming projects(which is practically non-existent). For those reasons, I more closely resemble the target candidates that statistics programs are looking for. This means I have a better chance of getting into a good program in statistics vs CS. 

So, Statistics, its what I'm the most interested in, a program in it would give me the skills I am looking to gain, and its the area that I have the best chance of being accepted to a good school in.

I think you have a good point about the difficulty of finding professors doing research in financial statistics. However, some schools that have good business programs, Rice for example, have computational finance centers and do interdisciplinary research. I've noticed that a lot of these programs that have strong finance programs or strong math programs try to boost their statistics programs by having the programs become cross-collaborative. Furthermore, just because a professor doesn't explicitly do research in statistical finance doesn't mean that I can't do research on my own. I already know my way around balance sheets and am a very independent thinker and a creative person. 

As you can see, I've already given this some thought. I didn't just wake up one day and decide I was going to do a PhD in statistics.

 

But back to the original question, if anyone has recommendations as to potential good fits that might be helpful. I'm asking on here because many of the school I talk to are very vague and tell me "oh yea, you should go ahead and apply". But when that's coming from University of Chicago, a top school, I am very skeptical of my odds of admission. 

 

EDIT: On machine learning, you are probably right that pure machine learning research is best done in a CS department. However, given my specific interests and the nature of jobs that I am pursuing. A CS PhD focusing on machine learning is going to be mainly computer science some statistics. I want to do mainly statistics some computer science. I'm more interested in the data side of things and want to take courses in time series analysis, stochastic calculus, measure theory, linear regression, advanced probability theory. I don't think I'll have the same opportunities in the CS department.

Edited March 1, 2017 by ECE23

[Robbentheking]

Well, my general impression of quantitative finance (disclaimer: I have't worked in it), is that if you want to just jump from theoretical math/statistics to getting hired, you need to be a beast at your subfield. Otherwise, I think it's much better to be able to get shit done (read: be good at programming). 

What is the approximate rank of your undergrad institution? what textbooks did you use for analysis/algebra? what did you cover? I agree that you seem to be better suited to statistics departments than CS departments, but just know that there will be people with stronger math backgrounds as you applying, especially at somewhere like Chicago. 

[ECE23]

6 minutes ago, Robbentheking said:

Well, my general impression of quantitative finance (disclaimer: I have't worked in it), is that if you want to just jump from theoretical math/statistics to getting hired, you need to be a beast at your subfield. Otherwise, I think it's much better to be able to get shit done (read: be good at programming). 

What is the approximate rank of your undergrad institution? what textbooks did you use for analysis/algebra? what did you cover? I agree that you seem to be better suited to statistics departments than CS departments, but just know that there will be people with stronger math backgrounds as you applying, especially at somewhere like Chicago. 

I went to The University of Alabama. Not a super prestigious university. The whole point of going to grad school is to give myself a chance to become a beast in my field. When I was in undergrad, I was an electrical engineering major, not a pure mathematics major. Most people take 5 years to finish the Electrical Engineering degree. I did it in 4 with a decent GPA and an extra semester of math crammed in there. I was trying to double major, couldn't quite squeeze it in though(2 courses shy). I was spread too thin in undegrad and was mathematically distracted by pursuing internships and worrying about getting a job, etc. At the time, I didn't realize that I wanted to study at the graduate level and pursued mathematics out of a motivation driven by personal interest. Now that I have graduated and worked in a full time role for 6 months, I have a better understanding of what it is that I want to do and am serious about devoting all of my time. The point is that I am not a beast in my field because I had too many other things on my plate such as an extremely time consuming engineering degree and internships. I am now looking for an opportunity to devote myself exclusively to studying mathematics, statistics and data science so I can give myself an honest chance to become extraordinarily good. 

Don't remember the book for Abstract Algebra, but I took the course as a sophomore. For Real Analysis we used Gaughan took it first semester of my junior year, the difficulty is probably close to what you would expect from Rudin's introductory analysis text. The main difference is that Gaughan doesn't develop his analysis on metric spaces and treats the topics in a less general, but still theoretical and rigorous light, but I now understand metric spaces through my self study of Topology. The course I was in contained graduate students and I did very well, one of the best students in my class as a matter of fact. I almost had an A+,was only 1 point shy, and was amazed by my own ability to solve difficult problems.  I basically took 2 years off from pure math after I got my internship at Chevron. One month ago I started again and I'm already deep into Mendelson's Introduction to Topology, which I am currently reading mainly due to interest and partially as a bid to raise my mathematical maturity. The book covers metric space, topological spaces, compactness, homotopy, the fundamental group, etc. Many of these topics are fundamental to Real Analysis and overlap. I have been studying several hours almost every day in addition to my full time job. I will finish the entire book in several months after which I plan on working through Rudin's Introduction to Mathematical Analysis, start to finish, just to give myself a very solid background in analysis for starting graduate school and to learn the material covered in a second Introductory Real Analysis course. At the rate I am working, I will finish both books and could be through another one before I would enter into a graduate program in Fall 2018. In addition to this, I am going to start studying for the GRE as well as I am going to need to get my score up to not be disqualified from some programs.

I think you're right about the programming typically being an easier entry point. Unfortunately, this is not an area that I am motivated to study. Additionally, you're right about Chicago, which is why I would consider that school a reach(by a good bit) and is why I am targeting schools in the 10-20 range mainly.

 

However, one of my best friends went from doing a theoretical math phd in Algebraic topology at a state university to being a quant algo analyst. He beat out guys from Stanford and MIT. He didn't know any programming at all but still got the job.

[machinescholar]

I did an internship at a quant desk in NYC and have many friends who go to top IB in Wall Street or hedge funds in Chicago. In my experience, the finance industry is very status obsessed and recruit heavily from the Ivies. So going through a solid Statistics or Mathematics PhD from a solid but non Ivy university may not get you there.

If being a quant is what you really want, there are friendlier ways such as getting a top-tier Masters in Financial Engineering, Operations Research or Mathematical Finance. The admission is often easier than a Math & Stat PhD and schools like CMU or Berkeley have great records of placing their students to top jobs.

Edited March 1, 2017 by machinescholar

[Robbentheking]

2 hours ago, ECE23 said:

basically took 2 years off from pure math after I got my internship at Chevron. One month ago I started again and I'm already deep into Mendelson's Introduction to Topology, which I am currently reading mainly due to interest and partially as a bid to raise my mathematical maturity. The book covers metric space, topological spaces, compactness, homotopy, the fundamental group, etc. Many of these topics are fundamental to Real Analysis and overlap. I have been studying several hours almost every day in addition to my full time job. I will finish the entire book in several months after which I plan on working through Rudin's Introduction to Mathematical Analysis, start to finish, just to give myself a very solid background in analysis for starting graduate school and to learn the material covered in a second Introductory Real Analysis course.

This sounds good, especially the drive to self study, although the deeper theorems in the Topology book are probably not super relevant to the courses you'd be taking as a stats grad. You probably be better off just mastering Rudin, if you can stomach it. Definitely work to improve the GRE score, and maybe take the math GRE too if you want to show off your ability.

2 hours ago, ECE23 said:

 The point is that I am not a beast in my field because I had too many other things on my plate such as an extremely time consuming engineering degree and internships.

This rubs me the wrong way a bit though. Seems like quite the assumption imo...

[ECE23]

 

1 hour ago, Robbentheking said:

This sounds good, especially the drive to self study, although the deeper theorems in the Topology book are probably not super relevant to the courses you'd be taking as a stats grad. You probably be better off just mastering Rudin, if you can stomach it. Definitely work to improve the GRE score, and maybe take the math GRE too if you want to show off your ability.

Thanks for the feedback. I will take that into consideration. I agree Rudin would be better preparation.

1 hour ago, Robbentheking said:

This rubs me the wrong way a bit though. Seems like quite the assumption imo...

Understood, I believe I misspoke. The way you are probably reading it is not how I meant it. Let me rephrase the statement in a way which closer represents what I was trying to communicate.

What I meant to say was. I have not personally invested myself enough into mathematics to find out what my full potential actually is, due to other obligations and distractions. Everyone who is a "beast" in their field has invested the majority of their time into their respective subject for years, without exception(if you know of an exception I'd be curious to hear about it). Whether or not I have the aptitude to reach that level remains to be unseen because I have not been committed to the study of mathematics.

[ECE23]

2 hours ago, machinescholar said:

I did an internship at a quant desk in NYC and have many friends who go to top IB in Wall Street or hedge funds in Chicago. In my experience, the finance industry is very status obsessed and recruit heavily from the Ivies. So going through a solid Statistics or Mathematics PhD from a solid but non Ivy university may not get you there.

If being a quant is what you really want, there are friendlier ways such as getting a top-tier Masters in Financial Engineering, Operations Research or Mathematical Finance. The admission is often easier than a Math & Stat PhD and schools like CMU or Berkeley have great records of placing their students to top jobs.

Appreciate the advice.

Would a Masters in those topics you mentioned be good for placement as someone wanting to work in automated trading?

I read this article recently

https://www.quantstart.com/articles/Why-a-Masters-in-Finance-Wont-Make-You-a-Quant-Trader

Not familiar with Operations Research, but isn't FE and Mathematical Finance primarily concerned with pricing of derivatives, hedging and things of that nature. I can understand how that would lead to a job at a IB as a "quant" or hedgefund that does more traditional investing and also studies derivatives. Not so sure that it would give the skills necessary to work for a quant hedgefund focused on automated trading.

I've heard the whole thing about status and what not, more than once and from someone in industry. Obviously, it is a steep path to get there due to the competition. But, I personally know someone who graduated with a PhD from a non-target school that got a job doing quantitative research for a hedgefund that does automated trading.

Edited March 1, 2017 by ECE23

[machinescholar]

3 hours ago, ECE23 said:

https://www.quantstart.com/articles/Why-a-Masters-in-Finance-Wont-Make-You-a-Quant-Trader

This is one of the most prevalent myth on Wall Street. Perhaps, NYU MathFin AdCom may give another take on the same topic:

https://web.archive.org/web/20160930183355/http://math.nyu.edu/financial_mathematics/content/02_financial/03.html

**Also note that while the author at quantstart offered some interesting ideas, he seems (at least from his profile) to never work in a Quant desk before.

Edited March 2, 2017 by machinescholar

[Robbentheking]

56 minutes ago, machinescholar said:

This is one of the most prevalent myth on Wall Street. Perhaps, NYU MathFin AdCom may give another take on the same topic:

https://web.archive.org/web/20160930183355/http://math.nyu.edu/financial_mathematics/content/02_financial/03.html

**Also note that while the author at quantstart offered some interesting ideas, he seems (at least from his profile) to never work in a Quant desk before.

Not saying this is bs per se, but of course finmath department at NYU is going to say that, just from a marketing perspective. 

4 hours ago, ECE23 said:

What I meant to say was. I have not personally invested myself enough into mathematics to find out what my full potential actually is, due to other obligations and distractions. Everyone who is a "beast" in their field has invested the majority of their time into their respective subject for years, without exception(if you know of an exception I'd be curious to hear about it). Whether or not I have the aptitude to reach that level remains to be unseen because I have not been committed to the study of mathematics.

For sure. Didn't mean to come on too strong. 

Certainly in math I think it's near impossible to get to the top of a subfield without devoting your life to it. Anecdotally, I had a professor in undergrad that told me that he was spending 24 hours/week on each math course he took by the time he was a senior at Harvard. This guy was a seriously smart guy, had that level of commitment, and ended up struggling like everyone else in math to get a professorship. 

You likely know this given your interest in finance, but check out https://en.wikipedia.org/wiki/Michael_Burry if you haven't. Cool example of a smart guy dabbling in something at a high level. 

[ECE23]

1 hour ago, machinescholar said:

This is one of the most prevalent myth on Wall Street. Perhaps, NYU MathFin AdCom may give another take on the same topic:

https://web.archive.org/web/20160930183355/http://math.nyu.edu/financial_mathematics/content/02_financial/03.html

**Also note that while the author at quantstart offered some interesting ideas, he seems (at least from his profile) to never work in a Quant desk before.

To be honest, the word quant is used broadly and can mean a lot of different things. I'm not debating that MFE degrees will qualify you for certain jobs like the ones mentioned in the article.  "They are employed in trading, securitized and derivative financial modeling, quantitative support of traders, quantitative risk management, and portfolio management."

Rather, I am saying that if you are interested in automated trading, which in my opinion in the future of hedgefunds(granted there will always be a place for fundamentals). The MFE is going to help you understand derivatives. Its not going to help you give you skills you can use in automated trading, other than understanding derivatives type instruments.

[ECE23]

11 minutes ago, Robbentheking said:

Certainly in math I think it's near impossible to get to the top of a subfield without devoting your life to it. Anecdotally, I had a professor in undergrad that told me that he was spending 24 hours/week on each math course he took by the time he was a senior at Harvard. This guy was a seriously smart guy, had that level of commitment, and ended up struggling like everyone else in math to get a professorship. 

My buddy I mentioned earlier that did his PhD in Algebraic Topology used to study 10-14 hours a day. Obviously high level math takes natural ability, but every successful mathematician has to work hard to bring it out.

12 minutes ago, Robbentheking said:

You likely know this given your interest in finance, but check out https://en.wikipedia.org/wiki/Michael_Burry if you haven't. Cool example of a smart guy dabbling in something at a high level. 

I found out about Michael Burry when I watched the big short. A good example of why fundamentals will still have a place in investing. Basically does the same thing Benjamin Graham used to do, but in a modern context. Value will continue to be useful, especially on the small cap/micro cap side of things and on the private equity side of things. But, I think HFT and algo trading is going to be huge. I think its the future if you are someone that wants to be an innovator in this area. Besides, I'd rather study mathematics than pour through balance sheets. Successful fundamental investing is less quantitative and numeric, its more of a literary function guided by a logic and reasoning. Yes, you use models and what not, but its really more about how you interpret the model in the context of the business. It involves ferreting out your facts, understanding the business you are studying, maintaining emotional stability, and adhering to a well defined philosophy. Its part art, part science, despite what people think.