Time for a Change in Structure of MFE Programs

[psuhoops]

I received my MSFE recently; while I learned a great deal, my intent was to focus on algorithmic trading/AI/data mining. Unfortunately, the lack of flexibility in the program did not allow me to focus on this area until the last (third semester). Most programs also appear to just as, if not more inflexible when it comes to choosing electives.

I look towards my MBA, where I had an opportunity to choose an area of concentration (finance), while still taking classes in other areas to get a broader base of understanding. While acknowledging that MSFE programs are shorter by a semester, much has changed since the first programs were launched in the 1990s. There are many "specialties" in the MSFE area such as algorithmic trading, risk management, quant/modeling, pure coding, etc. I think the time has come have different "tracks" or concentrations within MSFE programs. As an example, those wanting to work on a derivatives desk for an IB would take a different set of electives than someone who will work in risk management at an insurance company. They would both take a core set of courses that include programming, stoch calculus, and others that comprises about 40% of there classes, with the other 60% focused on their area of concentration. The "quant" area has changed a great deal since the 90s and continues to evolve, yet the structure of most programs has not adapted.

How do others feel?

 

[Andy Nguyen]

In an ideal market, those programs that "have not adapted" would be out of business. Truth is they are not and won't be simply because there are just many people willing to pay them. At the end, these customers will be left holding the bag because they are taught outdated things with no relevant real-life skills, such as programming.
Some programs have began to offer various tracks (Columbia MFE) or finding a niche to focus on. Those programs will continue to evolve and grow.

In the end, it's prudent to really do your homework about each MFE program. People don't take that point seriously, it seems.

 

[DominiConnor]

I wish you were wrong Andy, if we were to take at face value the questions on forums like this, the primary criterion for evaluating an MFE program seems to be the reputation of the university as a whole based upon how many people it rejects.

 

[Daniel Duffy]

The history of software is very interesting; its development was mostly GBM interspersed with Poisson jumps (birth-death processes) every [10,15] years. Some jumps are big, some small and quite a few (the smaller ones) are predictable.

30 years ago a PhD in engineering could get a job writing FEM s/w for reservoir engineering (THE hot topic in those days). Now FEM packages are almost free.

 

[psuhoops]

Andy, I was happy with the education I received, but given my background (15 years in Fixed Income), I was very focused on what my goal was and did not want to take classes that focused on areas I had no interest in. I know I am an outlier from an age/experience standpoint. However, here are a couple more observations/suggestions for the programs:

1) Like most MBA programs, most students should have at least a couple of years of work experience, particularly in finance/capital markets. Students from an engineering/coding background with little or no work experience don't seem to fully grasp how markets work, differences between markets (e.g. equities vs FI or corporates vs MBS) and other aspects that come into play (e.g. behavioral finance). They seems to think that memorizing Shreve and excelling at C++ qualifies them to be a trader. While those skills will get them a job, it will never get them into a trading role. I come from an Econ/MBA/CFA background...while learning the math was more difficult for me others with this background than an Engineer fresh out of undergrad, I feel we have a better understanding of how and when to apply models, copulas, etc. in real world situations. Based on resumes from current students, I think Berkeley does a very good job of this; others not so (FYI - I did not go to UCB).

2)Programs need to introduce alternate theories and encourage exploration of them. While I understand that the MFE is a professional degree, I don't think students are encouraged to think outside the box. I personally think that fractal theory and power laws are better at explaining markets than stochastic processes; most programs would hardly mention them (mine did). Anyone hoping to become the next Jim Simons will have to think outside the box. This goes hand-in-hand with the first point; with no experience in the markets, most students have no idea what to question when being taught (academic theory vs real world application/relevance). Schools need to hire more professors such as Mike Lipkin, who encourages this type of thinking.

 

[Andy Nguyen]

The more we talk about this, the more it stays the same.
This was 3 years ago http://www.quantnet.com/state-of-financial-engineering/
http://www.quantnet.com/hollow-men-of-financial-engineering/
This is recently http://www.quantnet.com/five-ways-to-improve-quantitative-finance-curricula/

So let me ask you this, did this become hindsight to you now or did you try to push for changes while still in your program? My experience is that MFE programs evolve because of both internal forces (students and alumni fight for changes) and external one (market changes). So if you wait for your program to become what you like of it, it will not happen without your input.

 

[psuhoops]

Andy, I made these and other suggestions from nearly the minute I set foot on campus. Since UIUC was a new program, there were many small things that could be fixed/improved upon without much trouble. The program heads were very interested in our thoughts on how to improve the program. Most of the more "high profile" ideas will take more time to implement, if they get done. As an example, I know they want to enlarge the size of the incoming classes. Doing that will allow them to hire profs w/real world experience and "beef up" the electives that will be offered. However, by enlarging the class size, the majority of the class will continue to have little or no work experience. You can't always get what you want....

I do agree that working with the program heads while a student and as an alum is the best way to get things changed. Obviously, some programs directors will be more willing to listen than others. A new program is more likely to listen to get things right than an established one. One thing that would be interesting to see. A survey of alums at each program, 5 yrs after graduation and their opinion/satisfaction of their learning experience/degree.

 

[bullion]

Here's an alternative way to look at it.

1) MFE courses are mostly gap programs created to promote/package/market those without experience (or from non-trading roles). And direct capital market experience is truly a privilege. Bloomberg alone can take most of your tuition (also explains why most MFE graduates never touched REAL financial data in Aaron Brown's post). In fact, I think most traders don't fully utilized/understand all the financial information available either. The requirement for prior market experience will severely reduce the applicant pool (aka client base aka profitability) for these MFE programs.

2) There are 2 types of traders (sell-side and buy-side) requiring very different skills. Sell-side traders focus on teaching clients the unique features of financial products (mostly theory-based) and buy-side traders are interested in beating the market. Since most MFE graduates (often from oversea and just want a stable job with big fancy names to impress family friends) are aspired to join BBs (with global marketing), MFE programs is more interested in providing a set path for students earning 4.0, remembering Shreve, writing standardized pricing algorithms, and getting graduates into "high paying jobs" (Instead of thinking outside the box or Aaron's argument about MFE not being a real, rigorous science).

3) With the exception of a few schools, most faculty have no trading experience (and if you have a profitable trading system, why would you become a professor and teach ?) Most students simply underestimate the full details of trading system implementation (ranging from high level bar-based rules on TD Ameritrade to HF routing/latency/hardware-sensitive, book/tick/order-based, statistics/natural language/control theory/signal processing/machine learning-inspired strategies). So it is very difficult to provide trading training in school since most practitioner systems are results of joint efforts (of multiple PhDs from complementing disciplines) and impossible for one professor to master all details (explaining why Aaron's suggestion for trading simulation course is hard to implement). Since the professors cannot beat the market, how would you expect the students to have the confidence to back what he learnt with their own money? (therefore taking responsibility for his own calculation)

4) Since most banks have devoted vast resources to develop their trading strategies in-house with armies of PhDs and top-notch infrastructures, I wonder why you expect ANY school to provide the same calibre of graduate comparable to practitioners. MFE programs are great for training analysts/associates who can explain to clients the features/dynamics of a new products using STANDARDIZED models (because if you give them a high quote using an alternative model, no one will do business with you). But they are still newbies as students of the market, and I think we all need to recognize just how qualified/unqualified these graduates truly are as potential traders.

In all fairness, MFE program directors need to balance the needs/demands from various fronts including
- students (who want classes to be easy, get a high paying job, pay as little tuition as possible, and in shortest time)
- professors (who want the same class materials every year so they don't have to update/learn new things + easy grading scheme)
- recruiters/career services (who want the students to be able to answer standard questions from Shreve and C++, predictable quality/marketability)
- school administration/admissions marketing staff (who want to generate as much revenue as possible, as little on-going research as possible to reduce overhead, promote the program claiming "our gaduates have become [fancy title]")

So if you want someone who truly understand the challenges of prop trading, it's best to do it in-house and just have him following you around and listening to all your complaints about what's not working. If he voluntarily tries to learn NEW things to solve your problems and improve your trading, then you have just found your next new star trader/partner (And yes, I will love to intern with any one of you big dogs).

Don't hate the players. Hate the game.

 

[quotes]

There are two different expectations on an financial engineer:

Price and hedge derivatives as accurate as possible;
Make correct bets on the market through statistics and mathematics.

The first one is difficult yet accessible to fulfill,
but the second one is almost impossible to find someone to teach.

Why?
Why do I teach you how to forecast the price if I can make money myself?
I am better off playing in a hedge fund like Renascence instead of earning your tuition.

Yet the closest alternative is the statistic major.

I disagree with psuhoops because people expect a statistic quant to use maths instead of intuition to forecast the market. They provide another job called experienced trader without requirement on maths. Of course someone gains if he is both, but isn't it asking too much for the first to be the second as a prerequisite?