Paul Wilmott: most quants are stupid

[Daniel Duffy]

I suppose one of the issues that Paul Wilmott is referring to is that many (PDE) models are linear and do not reflect real market behaviour. Almost all BS PDEs are linear. Other alternatives are uncertain volatility models and models with memory.

Very few processes are linear.

 

[DominiConnor]

One thing I'd note is how little quants get out. I am a professional attender of conferences, seminars, etc and the pool of Quants who talk to to peers outside their team is a lot smaller than it should be.
Although one needs to concentrate on the work, you can too easily lose perspective, yes reading The Economist, Quantnet etc will poke you a bit, but a critical part of being good at your job is talking to people who don't share your assumptions. A team mostly have the same view, or (at least argue over the same points) and discussing with someone who doesn't share the same group view helps you get it right.
Also chats at meetings don't matter...
If you same something that makes you look dumb over a coffee at a conference, no one really cares which frees you up to think and say things you might keep closed when at work where it matters what people think about you a lot more.

Paul uses the word "stupid", I use the word "algebraists", everyone here can do algebra and express themselves this way, some people are better than others and of course some here are better than me. The problem is where this is your only way of thinking.

And of course most quants have only a shallow knowledge of economics, that has led to big problems in the past, yet I see little effort to fix it.

 

[Daniel Duffy]

// my 2 cents
Conferences are very expensive!!! And it's difficult to learn much from a 45 minute presentation. What is the percentage of traders at these conferences?

Speakers also tend not to give source code, so that attendees can kick the tyres after the event. Speaking from a numerical analysis viewpoint, many methods (e.g. CN, ADI) are getting long in the tooth (for some time), even though there are many better methods. Maybe time for an update here as well.

It seems to me that if you have quants that they get trained in Economics. Not sure if reading the Economist will fill this gap, however useful it is.

 

[bigbadwolf]

Paul uses the word "stupid", I use the word "algebraists", everyone here can do algebra and express themselves this way, some people are better than others and of course some here are better than me. The problem is where this is your only way of thinking.

And of course most quants have only a shallow knowledge of economics, that has led to big problems in the past, yet I see little effort to fix it.

It's inevitable. You've got people coming from the narrow and cloistered disciplines of physics, math, engineering and computing, often not even comfortable with English, and you're expecting them to have an interest in, and understanding of, Western economic structure, finance, and politics. It takes years of reading and moving in the right intellectual circles for this to come about. An Oxford PPE would probably be a good starting point. A subscription to the Economist or FT isn't going to help much. Being able to move back and forth between the world of equations and programming on the one hand and the world of people, politics, and real-world economics on the other is a very rare skill -- and it's not even prized. My own theory is that a technical education -- physics or computing -- shapes a person and he cannot then be much of anything else; he always interprets the world through the prism of his technical training.

 

[bigbadwolf]

It seems to me that if you have quants that they get trained in Economics.

I don't know what is meant by "getting trained in economics" just as I don't know what people here mean by "being good at coding." These are vague and general statements. Surely it doesn't mean working through compendious and flawed texts like Samuelson and Lipset? Is what is meant "a framework for thinking about real-world economic phenomena?" If so, how exactly is it to be acquired?

 

[Daniel Duffy]

I don't know what is meant by "getting trained in economics" just as I don't know what people here mean by "being good at coding." These are vague and general statements. Surely it doesn't mean working through compendious and flawed texts like Samuelson and Lipset? Is what is meant "a framework for thinking about real-world economic phenomena?" If so, how exactly is it to be acquired?

My remark was a bit vague. What I meant what that if an understanding of Ecnomics is so important then someone shoudl train them.
Would such knowledge have prevented the housing or internet bubbles?

A knowledge of history is a good asset.

 

[psuhoops]

Here's an example of why I think within 15-20 years, Financial Engineering will be a concentration within MBA programs rather than a degree unto itself. During my risk management class, the professor talked about how VaR is useless when need most, in a crisis; its main purpose is to give the company's Board of Directors and shareholders a vaguely defined "risk number/level" during "normal" course of business. Doesn't matter what version of VaR you use, a back of the envelope calculation with some assumed correlations (all risk assets goto 1, -1 vs US Treasuries) and haircuts will produce a much closer value of portfolio loss. Students with some market/finance background understood this (I appreciated the frankness of the professor); most with an engineering background and no market experience seemed disappointed. In this case, as in others, they wanted an equation or series of equations to solve that would given you the precise answer to the question.

This fits into the conversation above that MFE students should have an understanding of econ/finance before they are admitted to a program. The original purpose of a MFE (coding equations and methodologies for pricing derivatives) has been nearly completed. Lots of hedge funds and other writing proprietary code would prefer the programmers know only what little finance they need to get the job done (less chance to duplicate once the non-compete expires). A quantitative MBA would produce a more well rounded product who could communicate better with management (or eventually move into a mgmt position), marketing, etc, have a better understanding of econ/finance and general management skills, while also having the quant skills needed to compete in a changing marketplace.

 

[Barny]

It's inevitable. You've got people coming from the narrow and cloistered disciplines of physics, math, engineering and computing, often not even comfortable with English, and you're expecting them to have an interest in, and understanding of, Western economic structure, finance, and politics. It takes years of reading and moving in the right intellectual circles for this to come about. An Oxford PPE would probably be a good starting point. A subscription to the Economist or FT isn't going to help much. Being able to move back and forth between the world of equations and programming on the one hand and the world of people, politics, and real-world economics on the other is a very rare skill -- and it's not even prized. My own theory is that a technical education -- physics or computing -- shapes a person and he cannot then be much of anything else; he always interprets the world through the prism of his technical training.

lol @ the English comment.

That's a nice little theory you have - where's your evidence? Hopefully you'll see that the prism through which I view the world is one of empiricism and the scientific method, which is hands-down less opaque than the prism you'll look through as an economist..

The problem is not what discipline one is trained in - it is merely one's quality and depth of thought. There are people who want to understand the way the world works, and there are people who just want to learn and apply algebra, or learn and apply economic theories without asking themselves vital questions along the way. It takes considerable effort to think in the right way when it comes to solving problems, and it's very easy to take the path of least action. (See for instance, Dan Kahneman's book "Thinking, Fast and Slow").

Here's an example, taken straight from Dan's book:

A bat and ball costs £1.10. The bat costs £1 more than the ball. How much does the ball cost?

50% of people from Harvard, Princeton, Yale et. al. get this question wrong. Up to 80% at "lesser" universities get it wrong. It's nothing to do with intelligence (the problem is trivial), it's about whether or not you can take a step back from your intuition and think through something more deeply. Most can't, dubbed "lazy thinkers" by some psychologists, and they're not the sort of people you want working for you in a bank, whether they have a degree in physics or economics.

 

[bigbadwolf]

Hopefully you'll see that the prism through which I view the world is one of empiricism and the scientific method, which is hands-down less opaque than the prism you'll look through as an economist..

I'm a disciple of Paul Feyerabend so I don't believe in "scientific method." I'm also a disciple of Bruno Latour, so I don't believe in "empiricism" either.

The problem is not what discipline one is trained in - it is merely one's quality and depth of thought.

But the thing is each discipline trains thought differently. Doubtless you've seen people who are subtle and profound in one area who are prosaic and devoid of insight in another. We talk of "transferable skills" but I don't know if they exist. I've seen excellent programmers who can't get their heads around basic math ideas and vice versa; I've seen mathematicians who can't understand the simplest ideas in chess. The quality and depth of their thought is good in one or two areas only. I'm sure versatile polymaths exist -- but I haven't seen many of them. Maybe someone like John Nunn, who got his DPhil in homotopy theory and went on to become a FIDE 2600 GM. Or Ken Rogoff, who's an economist and also a chess GM.

 

[Barny]

I'm a disciple of Paul Feyerabend so I don't believe in "scientific method." I'm also a disciple of Bruno Latour, so I don't believe in "empiricism" either.



But the thing is each discipline trains thought differently. Doubtless you've seen people who are subtle and profound in one area who are prosaic and devoid of insight in another. We talk of "transferable skills" but I don't know if they exist. I've seen excellent programmers who can't get their heads around basic math ideas and vice versa; I've seen mathematicians who can't understand the simplest ideas in chess. The quality and depth of their thought is good in one or two areas only. I'm sure versatile polymaths exist -- but I haven't seen many of them. Maybe someone like John Nunn, who got his DPhil in homotopy theory and went on to become a FIDE 2600 GM. Or Ken Rogoff, who's an economist and also a chess GM.

You're familiar with the current dogma that excellence is achieved from many thousands of hours of practice? The number of hours required to be a modern day GM scarcely leaves one able to follow any other pursuit with anything other than superficial dedication. I think if you're searching for true polymaths, you'll be searching for a long time. The best most could hope for in the modern day and age - where in physics, for example, you're unable even to master your sub-sub-field in a lifetime - is to be moderately good in many disciplines.

 

[Daniel Duffy]

A bat and ball costs £1.10. The bat costs £1 more than the ball. How much does the ball cost?

x +y = ..
x - y = ..

This is school algebra,

Why is it so difficult anno 2012??

 

[Dialectician]

A bat and ball costs £1.10. The bat costs £1 more than the ball. How much does the ball cost?

x +y = ..
x - y = ..

This is school algebra,

Why is it so difficult anno 2012??

Arithmetic is sufficient:

(1.10 - 1)/2

 

[Daniel Duffy]

Arithmetic is sufficient:

(1.10 - 1)/2

how do you motivate this? it looks like a geometrical solution.

 

[Dialectician]

how do you motivate this?

A bat and a ball cost £1.10.

The difference in cost between the two items is £1.

The total minus £1 is £0.10.

There are two items, so we must divide by two.

The ball costs £0.05. The bat costs £1.05. The sum is £1.10

 

[Daniel Duffy]

A bat and a ball cost £1.10.

The difference in cost between the two items is £1.

The total minus £1 is £0.10.

There are two items, so we must divide by two.

The ball costs £0.05. The bat costs £1.05. The sum is £1.10

The bold text is the crunch. Is this why 80% don't understand it I wonder?

 

[DominiConnor]

Here's why I don't think Quant will only be an option in MBAs....

Firstly let's look at programming, now MBAs typiccally do a bit of this, but it's dressed as as what people here would see as naive modelling.
However there are vast legions of people still doing CS and being hired as such.

Accountancy and marketing (the core of MBAs) have not disappeared as separate subject or as motivations to hire people.

The whole point of an MBA is not to understand anything at all. It's administration, of spun as "leadership", "entrepreneurship" etc.
The idea that you'd let an MBA actually do anything where there it was possible to define their work by themselves as "good" or "useless" is fanciful, the point of MBAs is to get credit for other people's work, sell stuff and how to sound like you're talking about.

As it happens, I teach all of the above on the CQF, but that's (hopefully) not why people do it.

What I do see happening is MBAs successfully claiming to be the right person to lead a Quant function, it's a crap idea but again if we look at IT, the reason banking IT is usually below average, occasionally scary and sometimes just funny is that the vast majority of banks have top IT leadership who have never once written code, configured a router and would reagrd swapping out a hard disk with the same horror an evangelical might experience if asked what a gene actually was.

MBAs actually do "quant" stuff already, marketing done properly is a quant skill, it can involve serious stats, time series etc and a few MBAs and some marketing people refer to that work, quite rightly as quantitative. Most MBA programmes have some of what Quantnetters would regard as quant, but just enough to bullshit your way into being dangerous.

 

[Barny]

The bold text is the crunch. Is this why 80% don't understand it I wonder?

The problem is not understanding. It's trivial to solve. The point is that the immediate answer that comes to your head is 10p, but you have to resist your intuition and work it out. The first thing I did was to say "OK, if the ball really is 10p, then.." and you immediately see that it gives the wrong answer. From there it was pretty easy to find the answer, 5p, by trial and error. Alternatively, you can use algebra:

(1+b) + b = 1.10
2b=0.10
b=0.5

Again, finding the answer is trivial - it's testing character more than intelligence, but it turns out that people who answer this question correctly are more likely to get firsts etc. at college. It's a more powerful predictor of performance than IQ, in some cases.

Here's another one for you:

It takes 5 machines 5 minutes to produce 5 widgets. How long does it take 100 machines to produce 100 widgets?

 

[Daniel Duffy]

It takes 5 machines 5 minutes to produce 5 widgets. How long does it take 100 machines to produce 100 widgets?

Can the machines break down? Do all machines behave in the same way?

This question reminds me of this:

http://www.wilmott.com/messageview.cfm?catid=3&threadid=75525&FTVAR_MSGDBTABLE=

 

[Dialectician]

Barny

The question really asks "what is 0.10 divided by 2?" I don't see the need for algebra, there is no unkown to solve for. It's just simple logic.

 

[bigbadwolf]

It takes 5 machines 5 minutes to produce 5 widgets. How long does it take 100 machines to produce 100 widgets?

Clearly 100 minutes (just kidding).