Are the "Physics Models" of Financial Markets WRONG?

[xoxqun]

How does anyone developing keen interest in quantitative models hedge one’s bet on investing future life in the “Physics Models”?

The Stock Market is not Physics - Part I


The Stock Market is not Physics - Part II

 

[Sanket Patel]

I think it suffices to note that the paper was, after all, published in a journal entitled the "Elliot Wave Theorist." :D

 

[JacekJ]

I enjoyed the read. Thanks xoxqun.http://www.quantnet.com/forum/member.php?u=8167

 

[charlesdwright]

I think it suffices to note that the paper was, after all, published in a journal entitled the "Elliot Wave Theorist." :D

Granted, and while I haven't read them yet, my gut is still that "The stock [or any other] market is not physics."

It is a flaw to take fixed laws of nature that apply to material things and presume they apply equally reliably to human beings, which, while material in a sense, are quite a bit more than that.

Of course we are obligated to make our best efforts, using our intelligence and our wisdom, to understand our world and be creative within it. Intelligence helps us crunch the numbers. Wisdom tells us that man is not merely a molecule.

 

[cstassen]

To be honest, I think that article is crap, nothing more. It is clearly written by people having to take/retake introductory econ and finance classes. Just look at the way they interpret the first couple of graphs - idiots! Apologizes for my blunt language, but stuff like that just makes me mad.

Every person with just a little bit of common sense knows that the laws of physics are not applied directly to trading etc. I work at a trading floor, and I can tell you that the thing that traders spend the most time on is questioning the models they use - not following them!

 

[xoxqun]

Questioning the Models

Thank you everyone for sharing your comments.

cstassen, thank you for your insight that 'traders spend the most time on is questioning the models they use - not following them!' May I ask how do traders know when to follow the model and when to question it and how do they develop this expertise. Also, how do they recognize the boundaries of (assumptions and logic of) the model recognizing that the model may not apply to a specific event or specific events. I am referring to what NNT calls black swans. :tiphat:

 

[cstassen]

I do know what a black swan is, and that is exactly what traders are in for. Whether we will ever get to a point where models are so advanced that they can predict the pricemovements I do not know. But as for now, the traders role is to 1) try to be one step ahead of the market and 2) know when to follow the models and not.

An example: I HATE technical analysis - especially on an intraday market. But nonetheless, human herding behavior is proven time and resistance/support levels are very much present.

Fx today - I put on a eurodollar trade, betting that the rate would bounce of the support level. That is exactly what happened.

I'm not saying that this is rational, but that is whats happening in the markets. Just like humans are always celebrating ones 50th birthday etc - round, psychological levels will always (and have always) affected human minds! A model will not be able to capture effects like this, at least how the model looks now!!! So the answer to your first question is: they do it by experience, by gut feelings and by technical analysis! And then of course some times by superior product knowledge.

The way they recognize the boundaries - I believe that is actually also answered in the above. This is also why the big banks more and more often hire quantitative people that understands the models, as this is the only way the could ever give a qualified reason for not following it!

Thank you everyone for sharing your comments.

cstassen, thank you for your insight that 'traders spend the most time on is questioning the models they use - not following them!' May I ask how do traders know when to follow the model and when to question it and how do they develop this expertise. Also, how do they recognize the boundaries of (assumptions and logic of) the model recognizing that the model may not apply to a specific event or specific events. I am referring to 'black swans' - typically described as events with very low 'assumed' probabilities based upon the Gaussian (bell curve) frameworks. :tiphat:

 

[charlesdwright]

To be honest, I think that article is crap, nothing more. It is clearly written by people having to take/retake introductory econ and finance classes. Just look at the way they interpret the first couple of graphs - idiots! Apologizes for my blunt language, but stuff like that just makes me mad.

Every person with just a little bit of common sense knows that the laws of physics are not applied directly to trading etc. I work at a trading floor, and I can tell you that the thing that traders spend the most time on is questioning the models they use - not following them!

yeah but snazzy models help sell stuff.

whether it's a chart or a complex model, traders will still have a tendency to see what they want to see in them.

 

[xoxqun]

Thank you cstassen for sharing that experience, gut feelings and technical analysis play an important role in traders' decisions in knowing when to follow the models or not. After reading some criticism of TA was becoming a bit skeptical about it of late. You seem to have nailed the issue with your concluding observation that big banks more and more often hire quants who understand the models, as "this is the only way they could ever get a qualified reason for not following it!" :tiphat:

 

[cstassen]

That one is spot on! I once worked in a private equity company (in the quantitative team), and we held a meeting for potential clients. I joined the meeting to take any technical questions. We had given them charts about our performance, and calculated different metrics. A guy from a stat.arbitrage fund asked if I could perhaps help him understand the mathematics behind a certain metric, and I said "Do you know stochastic calculus? It can be interpreted as the inverse Girsanov Kernel" - after that, he was totally sold, not paying attention to everything but that!

yeah but snazzy models help sell stuff.

whether it's a chart or a complex model, traders will still have a tendency to see what they want to see in them.

 

[cstassen]

TA alone is crap. But I must admit TA does have something to it. My point was more or less, that no trader relies on "physics models" alone when trey trade! And the way that article presented its points was even more hopeless - I mean, the fact that the market sometimes reacts positively to bad news is quite easy to understand, so the author must really lack knowledge!

Thank you cstassen for sharing that experience, gut feelings and technical analysis play an important role in traders' decisions in knowing when to follow the models or not. Before becoming interested in quantitative models, I was primarily dependent upon TA for trading my portfolio. After reading some criticism of TA was becoming a bit skeptical about it of late. You seem to have nailed the issue with your concluding observation that big banks more and more often hire quants who understand the models, as "this is the only way they could ever get a qualified reason for not following it!" :tiphat:

 

[chrisd]

That one is spot on! I once worked in a private equity company (in the quantitative team), and we held a meeting for potential clients. I joined the meeting to take any technical questions. We had given them charts about our performance, and calculated different metrics. A guy from a stat.arbitrage fund asked if I could perhaps help him understand the mathematics behind a certain metric, and I said "Do you know stochastic calculus? It can be interpreted as the inverse Girsanov Kernel" - after that, he was totally sold, not paying attention to everything but that!

Slight threadjack here: PE shops have quant teams? I had no idea. I thought they were entirely fundamentals-driven.

 

[xoxqun]

cstassen, thank you for sharing your experiences. May I request your thoughts about the role of formal academic training versus experience, gut feelings in becoming good at stochastic modeling and trading (more below). May be both are necessary, maybe not. Your response to the following questions will be greatly appreciated.

Do you think that most successful pros in trading / modeling typically get formal academic training (e.g. MFE / PhD in stochastic modeling) before applying their skills at stochastic modeling / trading? Do you know of self-trained successful pros who learn it on the job, or on their own and are able to do equally well professionally? :tiphat:

That one is spot on! I once worked in a private equity company (in the quantitative team), and we held a meeting for potential clients. I joined the meeting to take any technical questions. We had given them charts about our performance, and calculated different metrics. A guy from a stat.arbitrage fund asked if I could perhaps help him understand the mathematics behind a certain metric, and I said "Do you know stochastic calculus? It can be interpreted as the inverse Girsanov Kernel" - after that, he was totally sold, not paying attention to everything but that!

 

[cstassen]

Chrisd: that is how it used to be - things are changing. The need for proper statistical analysis etc has become present in the mind of fund managers. Instead of just "guessing" as they did before. "peer group valuation" - come on guys, that's b...****!

xoxqun: many succesfull traders are indeed selftaught / taught on job - that was how the business was earlier. I doubt, however, that you will be able to get a good (with good being a very relative word) trading job without either 1) a background in maths/quant.econ/engineering/physics or 2) an MFE or another finance related master or 3) a phd (which is, however, not usual for traders).

It sounds a little bit like you're not totally aware of the difference between quants and traders - and also between different types of traders. FX traders need a gut feeling and a gut feeling only, whereas structured credit traders need a quantitative background.

Maybe you could tell me your background in some detail, then I can more easily "advice"??

 

[xoxqun]

cstassen, thank you for taking the time to respond and allowing me to ask further.

>I doubt, however, that you will be able to get a good (with good being a very relative word) trading job without either 1) a background in maths/quant.econ/engineering/physics or 2) an MFE or another finance related master or 3) a phd (which is, however, not usual for traders).

> Maybe you could tell me your background in some detail, then I can more easily "advice"??

I agree with you; the three criteria may apply to some extent as I have undergraduate and graduate degrees in all three areas above, but not in FE, computational finance, or financial modeling.

:tiphat:

 

[cstassen]

With your background, you shouldn't have problems getting a trading job.

Regarding the understanding, how much maths have you done? Phds generally have developed an ability to sit down, read a book and UNDERSTAND it. If I were you, I'd try to read the two excellent books by Steven Shreve, and really try to understand them. Meanwhile, read some stuff about technical analysis, and create an "imaginary portfolio" and try to come up with trade ideas, both with a technical underlyer and a fundamental. Trade both long and short term.

The optimal will of course be to do one or more courses in financial modelling, but with your background you should be able to go to an interview and get the job - after that, you "just" have to fight while on job.

An understanding is a relative term - I'm not sure you'll be able to achieve the same understanding as a MFE student, but that's probably not necessary for a trading job!

With regards to the roles: it's crucial that you understand the difference between the areas. You can't say that trading/structuring/quant is more or less the same, cause it isn't

 

[xoxqun]

cstassen, thank you for your advice. :tiphat: Any suggestions on good reads on TA in your opinion. Also, any thoughts on playing with an 'imaginary portfolio' versus 'real money', and when / how to make a transition from the first to the second say if you want to build and trade your own fund.

I agree with you that the roles in trading/structuring/quant are quite different and am aware of it. Thank you!

 

[cstassen]

If you plan on doing intraday trading, then get like 3 or 4 years of experience before you risk your own **** - one gets greedy all out of a sudden. If you invest long term, go right ahead! If somewhere in the middle, I'd say you should probably do some reading first, experiment a little, and then go for it. Very difficult to say!