Is A Masters or PhD Overkill?
ฝัง
- เผยแพร่เมื่อ 2 ต.ค. 2020
- The question was asked, "Is a Masters or PhD Overkill?" This was in regards to working in quant finance. A lot of business students just can't grasp why you need a graduate degree. And even those with STEM under grad degrees think they know almost everything there is to know about math, statistics, engineering, and tech. The sad truth is an undergraduate degree barely teaches you anything when comparing it to what is required in quant finance or any technical job. Even a Masters or PhD barely gives you enough tools to start as a quant. Now of course you could just get the degrees, learn nothing, some how pass the interviews, and get a job. However no one will want to promote you as you won't be able to do the work. I see this a lot of undergrads who find a bank or firm who accepts undergrads for quant roles. They give them generic work with a hint of quant and the go out and hire PhDs for the "heavy lifting." Now these undergrads reach out to me asking why no one will hire them because they feel they have all the skills and experience of a quant. The sad truth is they aren't qualified. At many places I have worked, schools I have presented at, and conferences I have attended...most people look at me like I'm speaking another language. The truth is that most people are vastly underprepared to work in quant finance even with a PhD.
My Life as a Quant
amzn.to/2uOTxO8
Quant t-shirts, mugs, and hoodies:
teespring.com/stores/fancy-quant
Connect with me:
/ dimitri-bianco
/ dimitribianco
☕ Show Your Support and Buy Me a Coffee ☕
ko-fi.com/fancyquant
Whatever Dimitri. I took a "How to become a stock trading millionaire!?" online class by an 18yo billionaire and that's the only education I need.
lol
Haha! This is sarcasm, right? 😂😂😂
Applied math PhD student here. Putting in more effort now so that Dimitri can not run circles around me in an interview D;
That's much appreciated! I get tired after running so many circles.
13:23 A masters degree doesn't mean you will be able to do just the "bare minimum". Getting a PhD for a reason other than pursuing a career in academia or just personal accomplishment is counter-productive. While on average a person with a PhD is probably smarter, has better research skills and a better understanding of the scientific method it doesn't mean that they will perform better. As a PhD, interest in your research topic does not necessarily translate to interest in quant finance. Some of the greatest quants I've seen have only masters degrees. And I've also met incompetent people with PhDs.
A masters degree is more than enough so that you can productively read and understand textbooks and papers. Nowadays innovation in quant finance comes more from academia than from the industry. Unless you are one of the few firms (generally not banks) that want innovation and you need some PhD to implement some model or strategy based on their research that nobody else has, a PhD does not help.
Nothing beats industry experience. A masters at age 29 with 5 years work experience is much more valuable to anyone than person who just got their PhD. Both start out as associates.
13:38 This scenario in which a junior employee comes in, creates a whole model from scratch just does not exist. Nobody is ever going to ask someone who is entry level to create a model whether they have a masters or a PhD. But anyone would trust the masters with 5 years experience over the entry level PhD. If one person actually does create this whole model and then the whole rest of the team, validation, audit and regulators all miss the fact that it is a ridiculous model then the problem is probably management.
Baruch's 5 year report on grads here mfe.baruch.cuny.edu/5-year-report/ shows that there is no significant difference in average salary between people who had PhDs vs people who had bachelors prior to doing the program, though the sample size is small. Highest earners did not have PhDs prior to doing the program.
Exactly I think he goes a little overboard with the educational requirements for being a quant. I personally think it’s a lazy way to filter for candidates given how many applicants algorithmic trading firms get. But the amount of unpractical PhDs I have met is ridiculous.
Another thing I don’t understand is that he doesn’t have a PhD himself. His applied economics degree is barely mathematically rigorous. He clearly knows his stuff so he taught himself everything he needs to know plus got training when he needed it but tells others they need to do some next level degree in math to be even CONSIDERED a quant.
@@daudcodez3479 clearly you didn't understood what he said.
You are a far better face for the industry than anyone I've ever encountered. Keep it up!
Great information! Thank you!
Nice video Dimitri.
Greetings from Colombia!
Great video Dimitri! You’ve put the words to exactly how I think/feel about knowing/doing the real thing in a discipline. I definitely feel this way as I’m starting my Masters in CS and want to avoid “just passing” and focus on really excelling on concrete skills.
Hey Dimitri, could you make a video addressing the younger group of your viewers, specifically those trying to obtain a Bachelors in CS/DS/Math/Stat and laying out the key steps and key positions to try and obtain to push towards industry experience in quant (just scratch the surface), before getting our masters?
It is awesome to have found this channel, thanks for sharing your opinion, comments and inspiring words. See ya in a next video.
Thanks for being a part of the channel!
Glad to hear the comments in this video about feeling like you've scratched the surface once you complete a step of your education and realising what you don't know.
Finished a late-in-life undergrad degree just over a year ago and was surprised to get a 'is that it? There's all these things I don't know!' after going as deep as the course allowed and feeling it was so much information at the time.
On hold now on my educational path due to a mix of life, planning my focus for masters/PhD and of course the global pandemic!
Hi Dimitri,
( If you've answered this question, please link me to it)
I'm looking for some introductory stuff to get started on ML+Finance (not exactly quant) and I bought the "Advances in Financial Machine Learning" book but I feel it's a bit on the advanced side, could you share some books or courses if possible for a beginner> (I have a basic understanding as I've done my undergrad in Computer Science?)
Thanks a lot. :)
To be honest I don't see much difference in using ML for finance versus any other problem. Not many others have this perspective however if you know how a model works it won't make much difference if it's financial data or manufacturing data as long as you understand the finance basics you are applying it to. For basic ML I like "Hands-On Machine Learning with Scikit-Learn." It's a good introduction book. Below is a book review I made for the book.
th-cam.com/video/vLuwxDySiLk/w-d-xo.html
Dr Dimitri I’m a fourth year undergrad physics student at Oxford. I suppose I need a stats masters and PhD to be a good quant. I wonder whether I need to go to another elite school or the prestige from Oxford is enough ?
I'm not super knowledgeable on the UK's quant schools and job market but in general a well known school like Oxford should be enough. Oxford's statistics graduate program usually ranks tops 10 globally which is really good.
Im sold on getting the masters now!
Hey Dimitri I know you don't like commenting on certain programs but I just want to ask you about the MFE at UCLA because although it is under the business school it is ranked 13th(quantnet) and I would like to hear your opinion about whether such a curriculum would be rigorous enough.
Core Curriculum:
Investments
Stochastic Calculus
Econometrics
Financial Accounting
Financial Decision Making
Fixed Income markets
Derivative markets
Empirical Methods in Finance
Computational Methods in Finance
Financial Risk Management
Applied Fin. Engineering Project
Electives( I picked only quantitative electives and not courses like financial regulations) :
Quantitave Asset management
Data Analytics & Machine Learning
Advanced Stochastic Calculus
Statistical Arbitrage
The program is 3 semesters long plus an internship. Would you consider it good in terms of how rigor it is? Thank you so much in advance for your help and your valuable videos!
Love these videos
Thanks
Hi Dimitri,
Really enjoyable video. I am a second year undergrad and was wondering if there are any good guides on how to successfully apply to masters programs? I have a very good GPA and do a lot of self-study on the side about many different topics (your book reviews are usually a driver for those, halfway through python for kids right now!), but I'm worried that that's not really something discrete I can use in an application to grad schools (I can't really write "read text explaining derivation and applications of black-scholes" on a resume). I really want to get into a good program just to quench that thirst for learning, but I'm unsure of exactly what I should to during undergrad to convert my self-teaching of different skills/concepts needed for quantitative finance into a strong application. In fact, the entire quant masters application process is a little fuzzy to me, and I can't find any good resources online. Any advice?
If you listen to the podcast (videos come out on Tuesdays on TH-cam), I will be interviewing a director from an MFE program and some of the questions are about the application process.
Now for you specifically, you can add in your personal interests and discuss the books you have read somewhat in your "statement of purpose" (SOP). Many programs want you to write an SOP and this is a huge part of the application. You need to answer why you want the degree and how will it help you. This is your chance to tell your story and in your case it sounds like self-studying is a big part of that.
@@DimitriBianco Awesome, thanks!
Dimitri, do you think it is necessary to always have the 'latest' edition of textbooks? This has to do with the fact that textbooks are very expensive especially in developing country.
Maybe you can share your opinion on this.
It rarely matters. I encourage people to buy which ever is the cheapest version. I like new books just so there are no other notes. I add notes and highlight my books.
Hi Dimitri. Quick question, on min 7:16 you mentioned that “Technical Analysis was a massive joke and never worked”. Can you unpack this a bit more in another video, as there are still millions of retail traders that swear by it.
Signed Dr. Spence, Engineering Doctorate, and post doc (MBA) in finance. Finance Professor and Business Consultant.
Great video Dimitri. I'm a business analytics graduate student self-learning the maths and stats just because I enjoy them. While I'm always eager to learn, it takes a lot longer for me really understand and absorb the materials than others. Do you have any tips or advice for people without quantitative background such as me self-learning maths and stats?Many thanks.
Hi, for math and only basic stats, you may try the briliant. It is good in concept building and have lot of practice problems. If you practice consistently, you would gain much.
For stats, you may try online course from top schools like MIT on TH-cam or Edx. If you do on Edx, it would be graded with very good assignments and you will learn much.
I hope it helps.
Cheers!
@@poojanpujara4040 Thanks Poojan! That is really helpful!
Dimitri cracks me up, when he said "rigour, rigour, rigour!!!!". looks good on merch
I think the Fed should require banks to have bare minimum combo of masters CS + math / stats / econometrics.
Hi, Dimitri, I'm a long time fan, I see so many TH-camrs saying that, to model, you need no understanding of theory and a 3 month online course is all it takes, you're one of the rare who doesn't just say to people whatever they want to hear and I really appreciate that.
I don't know if you'll find the time to answer but I'd really appreciate it if you did as I can't think of anyone more qualified:
I'm now starting my second year of my Masters in Data Science with a minor in Mathematics at EPFL, Switzerland. It's a 2 year program but, as there is a lot of courses I want to cover before leaving, I'm probably going to do it in 3 years. I plan to take 24 courses (5 by semester during the 3 first semesters, 6 during the 4th, 3+12 credits project during the 5th and my Master Thesis in industry during the 6th):
-2 in Social And Human Sciences (Compulsory): Philosophy Of The Hard Sciences I and II
-1 in Business: Technology Ventures In IC (startup business plans for CS ventures)
-9 In Statistics:
-Linear Models
-Risk,Rare Events And Extremes (Extremal Modeling)
-Time Series
-Modern Regression Methods (Advanced Course On Linear Modeling)
-Multivariate Statistics
-Stochastic Simulations (Monte Carlo Simulations Implementation And Analysis)
-Statistical Theory (Foundational Theory Of Statistics)
-Statistical Machine Learning (Theory Of Machine Learning)
-Statistics For Data Science (Worthless, but took it last year...)
-4 In Other Math:
-Advanced Algorithms (Greedy Algorithms, Linear Programming, Randomized Algorithms...)
-Advanced Probability And Applications (Measure Theoretical Probability Theory)
-Optimization For Machine Learning (Optimization Algorithms For Convex, Smooth and/or Strongly Convex Functions)
-Foundations Of Data Science (Advanced Course In Information Theory)
-8 In CS:
-Machine Learning (ML Implementation)
-Distributed Information Systems (Information Retrieval, Data Mining And Knowledge Bases)
-Intelligent Agents (Software Agents, AI)
-Data Visualization (Front-End Implementation For The Efficient And Intuitive Display Of Information)
-Deep Learning (Deep Nets Implementation+Some Theory And Empirical Results)
-Applied Data Analysis (Course On The Software Tools To Work With Data)
-Systems For Data Science (Course On The Implementation Of Large Scale Databases)
-Information Security And Privacy (Implementation Of Information Security And Anonymization)
Here are my questions:
Do you think I'm missing something important or took something useless on the statistics side?
Do you think I'm spreading myself too thin and should further specialize or is this distribution ok?
If are not sure exactly what you want to do after graduation it's good to cover a variety of areas however at some point you need to narrow down what you want to do. The goal of a Masters is to get the fundamental skills for your area of expertise. You'll need to learn a lot more after you graduate however getting those key skills makes a big difference. Adding an additional year is really a personal decision. Remember that employers only care if you have the skills they want. I had 8 years of work experience at a startup and every quant firm I interviewed with thought it was worthless. It wasn't until I downplayed that experience and really highlighted my education before I got job offers. I can't really recommend any specific classes to cut as I don't know your end goal.
@@DimitriBianco Thank you for your input!
I have a bachelor in CS, people doing this Data Science masters have a similar background, the program is mostly CS-oriented. I took the minor in Math so I could cover Pure Statistics, Statistical Modeling and Machine Learning in a more solid way, the main idea was to get a competitive edge when it came to model development and tweaking, also I really love Probability Theory (there's no subject I like more actually).
The original goal was to be full-stack, by covering the entire pipeline of Data Science, with extra focus on the algorithmic and theoretical side, but I realize that might not be the best strategy. If I really had to choose, I'd take theory over implementation, I enjoy developing models and algorithms more than implementing them, also I'm sure the skill is harder to find.
If my goal was to go into model or algorithm development rather than implementation, do you think I should drop implementation courses to put more stats and/or algorithms or is what I have enough?
@@allisterblue5523 you have more than enough classes. If you wanted to cut back though, I would cut one or two of the cs classes.
All I can say is since you are such an admirable situation( such affordable tuition) I would go as far as I can.
@@DimitriBianco Thank you for all your help!
Dimitri, could you be so kind as to put a list of all the topics I should know to be a fancy quant. Cuz my circumference isn’t really big enough to realize what those are. Please... (actually I’m sorry for asking such a large question, it’s ok if you don’t answer)
I'm actually working on this as a video now. It's a lot of work but there should be a video out are the beginning of next year.
That would be awesome, thnx!
What should a physics masters do to become a quant, should I go for a physics phd or try to get industry experience?
The country you are in will make a difference but do you want to work in quant research or more towards developing models used for daily uses? If you want to do research you'll want to get a PhD in something quantitative (it could be physics but it could be something else). If you are wanting to have more hands on experience, many firms hire Masters students for developing or validating models used by the businesses.
@@DimitriBianco I want to go for quant research, so phd would be more advised?
@@hawtsauce2471 yes, you would need a PhD. The research experience is what firms are looking for.
Hi Dmitri! Do you know of any methodologies or papers, which describe how to monetise an increase in discrmintaory power of a model at a bank? For example if the current model has a Gini of 60% and the new model has a Gini of 80%. How could one put a monetary effect on the 20% delta?
The Gini will give you the separation between goods and bads. To be able to calculate a financial difference or improvement you'll have to select a score threshold. For example if you are using a PD model you could use 50% as a cutoff. You then use the old score to assign accept and reject (call it "old_accept"). If the PD is >= 0.50 then mark as a rejected application and the everyone else would be an accept. Do the same thing for the new PD and create the variable "new_accept." Then calculate out the financial losses of your data based on those two variables. You'll get higher losses for the low Gini model and lower losses for the high Gini model.
If you want to be even more accurate you can ask the business you are modeling for what they currently use as a cutoff. Often the PD as an example will get converted into a score which is used by the business. So instead of using a PD of 0.50 as a cutoff you could use their actual score such as 475.
👍🏻
I'm interested to hear your take on the cryptoindustry which has some of the frontier innovation in the finance world. Maybe you could reach out to Sam Alameda for an interview. He is a physics master/phd who decided to start his own exchange (ftx.com) which has had great success. They also have a YT channel with some great interviews that you can check out.
“I’m offended” *waits a week, clicks on another one of your videos*
LMFAO HE KNEW HE WAS GONNA OFFEND PEOPLE 😂😂
Disclaimer: This response is riddled with anecdotes.
So I never got a graduate degree, only an undergrad (granted from a higher ranked school). It never really limited my career and I've worked at money center banks, regional banks, and currently run a model dev team. I don't think a graduate degree offers any intrinsic value beyond a piece of paper that says "i studied X" ... In my experience, most grad degree holders who kiss their degrees more than they kiss their spouse have ended up being rank disappointments and average at best. Hyped up, over-exaggerated, and often arrogant students.
However .. people who got an undergrad in something non-stem but a grad degree in stem are typically very attractive candidates. The economics bachelors with a statistics masters is one of my favorites. I've even hired philosophy undergrad/statistics masters students and been roundly impressed by their work (often, the best communicators i've ever had work for me are these types). I find them to be extremely motivated and interested in the discipline. I have to admit, its not what I expected when I started running my team .. its something that I've been sort of forced to accept in the past few years.
These of course are based on my generalizations and are by no means fact for the entire industry.
You must be hitting a niche area. I could see that having a more creative soft skill side combined with a more technical side could be a good blend. most firms aren't looking for those so you might be able to get the cream of the crop in that area.