DanielWarlock

I don't have a clear answer to this. But I just want to comment that it can become extremely broad. For example, I recently discovered that Chatterjee at Stanford even works on quantum field theory! (https://statweb.stanford.edu/~souravc/qft-lectures-combined.pdf) It makes sense in that a lot of statistics come from physics. But eventually I think people just do things because they find them interesting.

Yes I do think you have an excellent chance. At least at Harvard I know there are much interests about sports analytics. You will most likely get in to be frank if you elaborate that in your letter. I know Mark Glickman and Kevin Radar are involved with data science program at SEAS but I'm not sure to what degree. Take a look at their work and mention that you would be interested to work with them and the sports analytics club in your letter. I say you have very very good chance to get into Harvard. Stanford statistics master is hard to be admitted and the coursework is rigorous. ICME on the other hand is easier and has more flexibility in terms of choosing coursework, potentially more suitable if you would like to work on applied projects using ML, NN etc. ICME is similar to Harvard program on this aspect.

This is a very solid list, although I think MIT and Princeton OR programs, Stanford ICME, Michigan data science are worth consideration also.

I think the best first year plan is probably Columbia. They have 4 different tracks: probability, theoretical statistics, applied statistics and data science (joint with CS and managed by Blei himself). Students take different classes (with some overlap) and take different qual exam. This way, no one will waste time. Coursework looks to me very rigorous and in-depth within each track. For instance, if you specialize in probability, the probability sequence is 3 semester instead of 2. Other than this, it is also good to have a more hands-off approach such as Harvard where courses do not take much time and students can just arrange their own studies (perhaps in consultation to their supervisor).

We used Agresti foundation of linear and generalized linear model. I feel that it is a bit cumbersome and wordy but in general a good book in terms of mathematical rigour. A lot of details regarding different models are not useful for me but I did learn some very useful techniques and practiced calculation quite a bit.

To master a technique for me is very very hard. In fact, I often find that taking even a very solid class does not truly allow me to master a technique--in the sense that I can independently solve a problem using that technique. To give an example, I first learned the gaussian interpolation in a class in the context of Slepian lemma. Then I read Vershynin's book and learned it again, this time not only Slepian but also its extension such as Gordon's inequality. I even derived Gordon's inequality using interpolation as an exercise from the book. Now when I see it again in the context of spin glass (Guerra's work on existence of free energy and upper bound), I stumbled as a total novice. I tried to prove these two theorems on my own without looking at the proof. Again, it proves to be quite a challenge and I just can't do it. So I studied interpolation fourth and fifth times. Later the monograph poses an exercise to use interpolation--again this takes me hours to finally solve on my own. You could imagine that to apply interpolation in a research problem in a nontrivial way could be much more challenging. So I still have a long way to go before I can claim myself a master of interpolation. So in a sense, taking a class is as quick and efficient as it can get but in a way I also feel it is less "nutritious" a bit like junk food. Many classes (at my institution at least) feel like a guided tour around an amusement park. You see "prototypical arguments" of a lot of stuff in its simplest form, but it never gives you a feeling that you are "hitting it hard enough" by working out all the different variants.

Contrary to popular belief, I feel that 1st classes at my stats department uses very minimal real analysis. The prerequisite for almost any class is just linear algebra and calculus. You can literally know zero real analysis and do pretty well. But a level of mathematical maturity is always assumed. It is mostly about problem solving rather than actual knowledge. A CS major, if solidly done, should have absolutely no problem. A biology major will be more challenging (I'm not talking about "biologists" who are actually theoretical mathematicians or computer scientists in disguise).

Unlike place like Stanford, these programs mainly concern how well you are connected with Toronto risk management community, particularly your internship records. If you do not have necessary connection and experiences, they are indeed quite tough with just gpa and research projects. Regarding internship pay, it is very true, and apply to most technical master programs in US also. If someone does not follow up with a phd like myself, a master turns out to be a pretty sweet deal that pays back within a year or two. Of course this is conditional on that you have initial fund to pay tuition.

I'm very familiar with those programs. I know most people on MMF admission and worked with several people from MQF. First of all, MMF and MQF are both unfunded and has zero research component. They are very competitive because they are well-connected in industry and can guarantee 100% you will find a high paying job in finance after graduate. Your friends didn't get in because it is very hard for to get into MMF with no connection. If you don't know anyone on the committee or have already worked with Canadian banks such as RBC, scotia, or pension funds such as CPP OTPP, there is practically no chance. The reason is because many applications from uoft or waterloo like myself, who did internship with Canadian banks or funds, do already have such experiences and know people on the admission personally. MMF and MQF mostly concerns risk management and the circle is pretty small. There is no way you apply from a foreign country like US and get in over someone who had Canadian work experience. GPA requirement is also quite high, like 3.9+ is typical among a cohort.

Certain funded masters or phd programs in Canada explicitly ask you to contact advisor before application, because funded means mandatory research. The program wants you to have a match with faculty before they take you. See for example uoft mathematics phd or MASc (if I recall correctly). Sometimes this is not written explicitly on the website but is assumed. You still didn't specify which funded master programs at UofT and waterloo your friends applied to so it is hard for me to judge the truth. That said, I do feel Berkeley is easier to get into compared to Chicago (I'm specifically referring to the statistics master not mfe, mfe at chicago is much easier) and certainly not comparable to Stanford statistics or Princeton ofre (btw I got in berkeley as well). I don't know about Cornell but I assume it could be easier as well. Such schools including columbia as the prof has metioned are a sort of "grey" area between the most elite and the your run of mill "cashcowy" master programs.

I get downvoted for a million times here for saying some elite master programs are very competitive. The dogmatic opinion on this site is that master programs are easy to get into, often described as much much easier than most phd programs. But it is simply not true. The aim of my previous post is to explain why. Also note that most posters here are direct biostats/stats phd applicants so it makes sense there is a bias. I also want to add that elite master programs like Stanford tend to be very competitive is a fact that bolstered by their admission stats. Not everyone wants to do or dead set on PhD. Some are very solid candidates for PhD but choose not to--this is paraphrasing from Stanford MS's own application guide.

It depends on the program. Some Canada programs require networking beforehand. If you just apply out of blue, you get rejected 100%. Can you give specific situation of your "counterexample"? Are those same programs OP mentioned because I applied to the programs OP mentioned.

US masters at prestigious schools can be harder to get into than funded masters such as those in Canada. I was accepted into UBC and Waterloo but was rejected at Chicago and Stanford--both are self-funded programs. The reason for this is simple: most graduates from these schools manage to land jobs in highly prestigious firms such as Google, Facebook, Apple, Citadel, Goldman Sachs. The school itself provides lots of opportunities in the form of alumni and career fairs. Some even find full time jobs in the same group that hires PhDs and working on exactly the same stuff (stuff that one may consider "fun")! So by paying somewhere around 100K intuition, you land a job that pays all tuition back in the very first year and 500K+ a year in 5 years and potentially much more in the long run. And you can form relationship with people who will "take off" in potentially highly distinguished careers, which could turn into a resource if you care to exploit it in some way (e.g. marriage, comradeship). No need to say, these type of "elite" master programs, albeit self-funded, have values beyond education, coming with boat load of life-changing opportunities for especially international students including visa. As a result, they are intensely competitive. I think programs like Stanford MS, Princeton OFRE MFin can be even more competitive than a majority of PhD programs (definitely outside of top 20 range). This will be controversial on this site but I still believe in this firmly. But again, the caveat is that these programs do NOT warrant a good job for an inexperienced and incapable person: you need to still perform superbly on your interviews and most likely obtain a large portion of such ability on your own (e.g. leetcode, side-projects) as opposed to just taking classes offered by the program. People who go to place like Harvard (for example) can end up in entirely different situations--for some, the degree just does not make a big difference before and after.

I did not contacted any profs from UBC but was still admitted as an international student. The same goes to waterloo. UofT on the other hand is entirely different. I graduated from UofT undergrad and did pretty well on courses from UofT stats department, did research with prof from stats department but was still rejected. I don't know if explicitly asking my prof to "secure me a spot" would make any difference but I simply asked for a reference, which was not enough on hindsight. So it goes with many people including someone who managed to get into phd program at Princeton. The (unofficial) consensus is that the quota for internationals is very very limited. In fact, most international people I know are not able to gain admission to that program. Waterloo and UBC are very good otoh. I got admission from both, except that stipend from UBC is on the low end if I recall correctly.

This is exactly what I thought. The recent hires at my department all work on the "modern statistics topics", especially concerning high-dimensional problems and statistical learning. Some senior professors who used to work on "outdated" stuff have also "switched field" and started to publish in some of these emerging subfields. Even so, when I look at MIT, our department still doesn't quite measure up on this kind of research.

I must confess that I downvoted some of your posts, but you also downvoted all of my posts in retaliation. All of this is a bit childish, but also largely benign. I trust that both of us are in good faith, especially in terms of contributing to this forum which we all love. OP obviously has interest in theory research and has taken very theory-leaning classes and is even considering applying to math. That's why I picked that particular list of professors. What's great about MIT is the range of options you have beyond statistics in a traditional sense--there are areas where MIT does not focus (e.g. biostatistics) but you have to admit a lot of research happening there is cutting edge, and cannot be find anywhere else, not to mention most of the resources at harvard is available to mit students also. Why would it be off-beat to consider applying to MIT as one of the "reach schools"?

OP is very strong and is considering pursuing a math phd. There are many other people here like OP who are very interested in the kind of research happening at MIT, e.g. see MIT's statistics and data science institute: https://stat.mit.edu/people_categories/core/ We always suggest them to try Stanford, UCB or even CMU. But no one mentions MIT ever. I think this is an important omission for our community. No one even mentions MIT as one of the "reach schools" when they can totally apply to EECS or Math stating that they are interested in statistics. Yes, a lot of them will not be good enough--but the same applies to Stanford etc. For those who DO have a chance, e.g. OP, MIT deserves serious consideration.

An important omission on the above suggestions is MIT. MIT does not have a statistics department but it is possible to study statistics there through EECS, math, OR, or CSE track. The matter fact is that when you talk about the "hot areas" such as statistical/machine learning, inference algorithms, high dimensional statistics, MIT is as strong as (or is probably stronger than) Stanford or UCBerkeley. A list of "emerging superstars" there: Elchanan Mossel, Sasha Rakhlin, Philippe Rigollet, Guy Bresler, David Gamarnik, Ankur Moitra, and many many more. I'm surprised that this forum doesn't even mention MIT when it has one of the most powerful stats communities there. Not to mention that when you get into MIT, you virtually get into Harvard because you can have supervisors/collaborators at both schools and take any courses, joins any reading groups you like at both places.

I think you definitely stand a chance. I think Cambrdige math tripo III publishes their offer rate which I believe is above 30% so your profile definitely is competitive. But of course they are still competitive.

Cambridge MASt is a strong program, so is the oxford one especially you have your eyes on PhD. My cohort this year has graduate from both programs. University of Bonn has a very strong master program in mathematics also where you can specialize in probability/statistics.

I'm a starting phd myself this year so I can't claim authority on this. But here is my two cents. Normally I would be tempted to say that you will have no trouble at top programs as you listed. But I have seen too many competitive people here and elsewhere this year to say that you are a shoe-in (several published at places like anal of stats/prob, jmlr, jams etc). I think one of the reasons is that the covid pushed a lot of elite people to grad school from industry. An argument can be made that there is less competition from international applicants, but as far as I know most competitive people will apply regardless. But it is really hard to say. So even if your profile is very strong, I'd be cautionary and apply broadly: this does not mean you should not consider top 10 schools. But rather to apply to some range 20-50 schools as well.

First of all, you don't have to be anxious or stressed out. I read your post. You seem to be concerned about a lot of things. I am not able to answer many of your very technical questions. But trust me there is no need to be worried about any of these things. Try to relax and enjoy your senior year! Do not let the pressure get to you. Have a social life, have fun, and talk to your friends (especially when you feel low or are stressed out). Here is another fact that may soothe you: most students applying to masters or phd programs in statistics don't have nearly as much experience as you do. You have done great work and should be proud of your achievements! Master programs are often less selective than PhD programs. Your profile has far exceeded requirements of many programs out there. I'm sure that you will get into some of the places on your list. Finally, good luck!

You should know that master programs do not rank exactly like PhD programs (as in US news). I would like to use this opportunity to give a TOP 10 Rankings for US Master Programs: ==================================== Tier 1: 1. University of Chicago (MA in Statistics) 1. Stanford (MS Statistics or Data Science) --------------------------------------------------------------------- Tier 2: 3. Stanford (ICME, ML/AI/Data Science track) 3. Princeton (OFRE, MSE, MFIN) 3. MIT (Masters at OR or EECS) --------------------------------------------------------------------- Tier 3: 6. UW 6. Duke 6. Upenn ---------------------------------------------------------------------- Tier 4: 9. Harvard 10. Berkeley ==================================== This list is based on personal opinion but is generally influenced by popular opinion. For example, Stanford and UChicago is considered best 2 programs (personally I think UChicago is better than Stanford with thesis option, scholarship and ~50% admission rate to its PhD program). The other overlooked program at Stanford is the data science offered under its ICME institute as opposed to the one offered at stats department, which is also very good--there is an option to just switch to PhD at ICME upon satisfying some basic requirement. The next in line must be the MFIN or MSE at Princeton ORFE--at first glance it is untraditional for statistics but there is a good chance to switch to PhD program under ORFE which gives you excellent chance to study under top probabilists and statisticians. If you orientation is industry then ORFE gives you the best platform. The other excellent option to study statistics at master level is at MIT: either through EECS or OR's master programs. There are a lot of statistics/probability going around at MIT and lots of classes even though it does not have a dedicated statistics department. Next, UW, UPenn, Duke all have very good, very solid master programs in statistics. The last category encompasses two better known schools. They are both good but curriculum is not as rigorous. For example, Berkeley only has only 8 months, no thesis, and classes are really watered down stuff compared to PhD. Harvard is autonomous --more like a self-tailored sort of program--so it is harder to say. On your profile: Your course work and research are really good. I think you have a shot at Top 10 (maybe outside of tier 1).

- I recommend retaking GRE because your verbal/writing is a bit too low. - If your main interest is EM, I recommend applying to Harvard also; there is considerable interest there in Bayesian computation and I think Xiao Li Meng is quite well known as far as EM is concerned. Also Columbia is also a good choice with well-known Bayesian people like David Blei and Andrew Gelman.

Your profile could become much stronger than mine. But most of your hardcore classes as well as research have not come out in time. And you have not done math GRE which is another problem. Gap one year will get you into Stanford level assuming optimal performance in those 3 graduate series (real analysis, probability, stats theory), math GRE (90%+) and research (to the point where your supervisor finds impressive). If no gap, then it is really hard to say. But still, you could get into some solid schools even with no gap. It is just hard to say if you need to get to the Stanford level. So I would personally choose to gap if I can.