SERVING THE QUANTITATIVE FINANCE COMMUNITY

 
User avatar
JohnLeM
Topic Author
Posts: 380
Joined: September 16th, 2008, 7:15 pm

Quantitative Analysts witch hunt

October 31st, 2008, 3:40 pm

Since several months, rising murmurs, widely relayed by medias, are intimating that “Mathematicians”, “Scientist”, would be the people to blame for the current crisis.I was amazed this morning to hear, in France, the very serious public French radio France Inter, pod casting the very serious Eric Delvaux 7/10 emission, relaying this kind of information.I believe that some communication is needed here, to block this process. In the very hope to be useful, I wrote down a "Billet d'humeur", meaning a short note written purposely immediately in response to an anger feeling. You may find it here : Are Mathematicians Sons of Witches ?.Since I am not really a quantitative analyst, do not hesitate to react on the above forum, would you read something irrelevant or false. I will obviously correct it.
 
User avatar
quantmeh
Posts: 5974
Joined: April 6th, 2007, 1:39 pm

Quantitative Analysts witch hunt

October 31st, 2008, 4:16 pm

QuoteOriginally posted by: JohnLeMintimating that “Mathematicians”, “Scientist”, would be the people to blame for the current crisis.if not them, then who else? u can't say "everybody's t blame". that doesnt make any sense and conveys no information. blame must be localaized
 
User avatar
JohnLeM
Topic Author
Posts: 380
Joined: September 16th, 2008, 7:15 pm

Quantitative Analysts witch hunt

October 31st, 2008, 4:26 pm

who or what should we blame ? There is a proposition on the blog.More fundamentally, is this really impossible to forecast economical crisis ?
 
User avatar
quantmeh
Posts: 5974
Joined: April 6th, 2007, 1:39 pm

Quantitative Analysts witch hunt

October 31st, 2008, 7:02 pm

QuoteOriginally posted by: JohnLeMMore fundamentally, is this really impossible to forecast economical crisis ?impossible by definition.think of this. is it possible to forecast weather? suppose that we found a way to do it. would it influence the weather itself? no. regardless of our knowledge of the coming hurricane, it'll come.now if somehow we forecast the economic crisis it won't happen. because everybody will do something and our actions will affect the economy.if molecules in the air were intelligent and they knew about coming hurricane and didnt want it, then they'd do something that would prevent a hurricane
 
User avatar
JohnLeM
Topic Author
Posts: 380
Joined: September 16th, 2008, 7:15 pm

Quantitative Analysts witch hunt

October 31st, 2008, 7:49 pm

Jawabean,I do not agree with your analysis. We are not talking about physical phenomenas, we are talking about traders behaviors. The traders are the particles to consider in the forecast weather here. Usually, this population is not stupid at all, they obeys to quite simple laws.Should an organization be able to forecast a forthcoming crisis, then traders would probably include this forecast in their strategy as a risk. I am not talking about a self propheting theory, but about a quantified theory of crisis forecast.So what is the probability of having a strong decreasing prices at a future date ? Obviously I have no answer. But I may provide a hint to discuss.I am sure that such quantities can be computed, at least for simple economical models. For instance, consider a population of arbitragist traders, each of one believing that an asset follows a Brownian motion, or geometrical motion. Each of one trying to maximize their profits, placing bid and ask offer at random times, the whole defining a market of null sum over an asset.It is a striking fact to observe that such a simple economical model leads to an "economical" crisis, i.e. a crash: one of these trader will accumulate all profits. Then, not being able anymore to make profit, he will have to sell off massively, leading to a crash in asset prices. Numerically speaking, it is very easy to implement and to study. Almost surely, the expected time to crash can be computed explicitely.Tell me if it is clear enough. I may also write down a research note over this topic, would it sound interesting to the community.This is surely a very naive suggestion. However, it is a first model to study.
 
User avatar
quantmeh
Posts: 5974
Joined: April 6th, 2007, 1:39 pm

Quantitative Analysts witch hunt

October 31st, 2008, 8:09 pm

QuoteOriginally posted by: JohnLeMFor instance, consider a population of arbitragist traders, each of one believing that an asset follows a Brownian motion, or geometrical motion.why would they b trading then?IF they TRULY believed in GBM, they'd quit do something else. they trade precisely because they DONT believe in GBM
 
User avatar
spacemonkey
Posts: 443
Joined: August 14th, 2002, 3:17 am

Quantitative Analysts witch hunt

October 31st, 2008, 9:25 pm

There is one point that does not seem to have been made during the analysis of the totally predictable and avoidable 'credit crunch'. It would seem that many quantitative models have failed, but that would be a total misinterpretation of events. In fact, the quantitative models - and I am thinking particularly of risk models, but also the MBS/CDO pricing models - have succeeded spectacularly.Management and traders in investment banks needed risk models that supported an asset bubble. They didn't want risk models that told them not to trade, or models that suggested there might be a downside, because they were making money from the boom. They got exactly the models they needed which were exactly the models they paid for.
Last edited by spacemonkey on October 30th, 2008, 11:00 pm, edited 1 time in total.
 
User avatar
JohnLeM
Topic Author
Posts: 380
Joined: September 16th, 2008, 7:15 pm

Quantitative Analysts witch hunt

November 1st, 2008, 8:55 am

SpaceMonkey, I agree with you for the models. Partially on your remark concerning Manager and Trader. Normal Managers and Traders, I mean no crook ones, obeys and are constrained by the overall bank risk management. An I am quite confident that financial organizations are looking forward to a fiable tool to measure the bubble risk, in order to include it into their current valorization, through an extra independant organism. At least, I bet that such an organism would have been useful to bear stern, AIO and others.Jawabeam, maybe should I try to explicit a little bit this idea. It means a mathematical formalism. Let's go to it.Suppose that I am the i-th trader, among a population of N trader. Today time is t. I own Ni(t) shares, and Ci(t) in cash. The price of the share today is S(t). So that my portfolio is valorized as Ni(t)S(t)+C(t). I have some information about this company, and I believe that from today t to a future date T, this share will follow a Geometric Brownian Motion, of rate ri(t,T), and volatility Vi(t). I have to place bid order b, and ask order a, for the period ranging from time t to time T, because I proposed a ride to a nice girl into my ferrari car.The orders are :i) if the share price crosses my ask price a, I will buy C/a shares (omiting the integer part to simplify)ii) if the share price crosses my bid price b, I will sell Ni(t) shares, having an added cash of Ni(t)*b.iii)if the share price remains between a and b, then nothing happens. Thus the expected returned value of my portfolio should be something likeC/a E(S(T)-C(t) + Ni(t)S(t) | S(t)<a) ) (i)+ Ni(t) E(S(T)/b + C(t) | S(t)>b) (ii)+ (Ni(t)S(t)+C(t)) E(b<S(t)<a) (iii)To optimize, I select a and b in order to maximize the previous fonctional. Since I suppose the underlying motion for this period, it involves mainly Brownian Bridge like computations, and I should be able to explicit them.From my side, I implemented this very simple model some four years ago. Numerically, it leads invariably to a bubble. I did it for fun, but I could look into my archives to see if I still have the corresponding code and a quick analysis if there is some interest of course.
 
User avatar
JohnLeM
Topic Author
Posts: 380
Joined: September 16th, 2008, 7:15 pm

Quantitative Analysts witch hunt

November 1st, 2008, 9:08 am

ok, the computations of the expected return has to be reworked a little bit, it is clearly false. Maybe this one is a little bit betterC/a E(S(T)-C(t) + Ni(t)S(t) | inf_{t<s<T} S(s) <a) ) (i)+ Ni(t) E(S(T)/b + C(t) | sup_{t<s<T} S(t)>b) (ii)+ (Ni(t)S(t)+C(t)) E(b<S(s)<a, for t<s<T) (iii)
 
User avatar
spacemonkey
Posts: 443
Joined: August 14th, 2002, 3:17 am

Quantitative Analysts witch hunt

November 1st, 2008, 12:27 pm

QuoteOriginally posted by: JohnLeMSpaceMonkey, I agree with you for the models. Partially on your remark concerning Manager and Trader. Normal Managers and Traders, I mean no crook ones, obeys and are constrained by the overall bank risk management. Of course, I was not suggesting that there was a grand conspiracy to deliberately build bad models. However, if you consistently advocate risk models that stop individuals from earning large bonuses, which in an investment bank means taking huge leveraged risks, then you are unlikely to find yourself in a position of authority. A perfect risk model would probably have suggested that you stopped trading MBS sometime in say 2005. If I had built a risk model that suggested we should stop trading MBS in 2005, do you think that model would have been adopted? Would I have been promoted for my hard work?
 
User avatar
ppauper
Posts: 70239
Joined: November 15th, 2001, 1:29 pm

Quantitative Analysts witch hunt

November 1st, 2008, 12:53 pm

QuoteOriginally posted by: jawabeanQuoteOriginally posted by: JohnLeMintimating that “Mathematicians”, “Scientist”, would be the people to blame for the current crisis.if not them, then who else? the social engineers of the BJ Clinton administration.It has been much publicized that BJ Clinton forced banks to lend money to people they would rather not have lent money too and forced Fannie&Freddie to insure those loans. That led to the subprime crisis
 
User avatar
Cuchulainn
Posts: 62602
Joined: July 16th, 2004, 7:38 am
Location: Amsterdam
Contact:

Quantitative Analysts witch hunt

November 1st, 2008, 12:57 pm

QuoteOriginally posted by: ppauperQuoteOriginally posted by: jawabeanQuoteOriginally posted by: JohnLeMintimating that “Mathematicians”, “Scientist”, would be the people to blame for the current crisis.if not them, then who else? the social engineers of the BJ Clinton administration.It has been much publicized that BJ Clinton forced banks to lend money to people they would rather not have lent money too and forced Fannie&Freddie to insure those loans. That led to the subprime crisisOn the other hand, everyone was for it at the time? After all, interest rates were real low?A number of issues came together as it were to make it possible. Again, they forget to do their 'what-if' scenarios (aka risk analysis)
Step over the gap, not into it. Watch the space between platform and train.
http://www.datasimfinancial.com
http://www.datasim.nl
 
User avatar
Paul
Posts: 10789
Joined: July 20th, 2001, 3:28 pm

Quantitative Analysts witch hunt

November 1st, 2008, 1:06 pm

spacemonkey is perfectly correct. The mathematics and the models are irrelevant. All they are needed for is to justtify trading stupidly risky instruments in ridiculous amounts so that everyone gets an obscene bonus. This explans why quants are always using bad models when there are good models available. In this sense quants play the same role in getting us into this financial mess as lawyers and intelligence did in getting us into the Iraq mess.P
 
User avatar
JohnLeM
Topic Author
Posts: 380
Joined: September 16th, 2008, 7:15 pm

Quantitative Analysts witch hunt

November 1st, 2008, 1:17 pm

QuoteOf course, I was not suggesting that there was a grand conspiracy to deliberately build bad models. However, if you consistently advocate risk models that stop individuals from earning large bonuses, which in an investment bank means taking huge leveraged risks, then you are unlikely to find yourself in a position of authority. A perfect risk model would probably have suggested that you stopped trading MBS sometime in say 2005. If I had built a risk model that suggested we should stop trading MBS in 2005, do you think that model would have been adopted? Would I have been promoted for my hard work?Financial organizations are already quantifying credit risk in their pricers, VAR methods and Bale agreement. Why would not they consider bubble risk, that is clearly a risk.SpaceMonkey, I totally agree with you. I am obviously not interesting financial organizations with such ideas. However, recall that the world investment banking system is threatened to be under a heavily regulation process. Would we able to quantify the bubble risk with pertinent arguments, why not suggesting to regulators to include the bubble risk in their analysis ?By the way, I found some traces of this ancient work of mine in my archives. I am rewriting it a little bit as a working paper, since it is Mhh...let say..dusty. I propose to start another topic, which will be more technical, to discuss the mathematical ideas.Here are some related questions.Sound "Bubble formation in a simple Stochastic Market Model" a good title to you ?Is Wilmott forum the good place to discuss it ? If yes, is the general forum the good place for it ? Alternatively, I can put it on my company forum.
 
User avatar
spacemonkey
Posts: 443
Joined: August 14th, 2002, 3:17 am

Quantitative Analysts witch hunt

November 1st, 2008, 2:07 pm

QuoteOriginally posted by: JohnLeMQuoteOf course, I was not suggesting that there was a grand conspiracy to deliberately build bad models. However, if you consistently advocate risk models that stop individuals from earning large bonuses, which in an investment bank means taking huge leveraged risks, then you are unlikely to find yourself in a position of authority. A perfect risk model would probably have suggested that you stopped trading MBS sometime in say 2005. If I had built a risk model that suggested we should stop trading MBS in 2005, do you think that model would have been adopted? Would I have been promoted for my hard work?Financial organizations are already quantifying credit risk in their pricers, VAR methods and Bale agreement. Why would not they consider bubble risk, that is clearly a risk.SpaceMonkey, I totally agree with you. I am obviously not interesting financial organizations with such ideas. However, recall that the world investment banking system is threatened to be under a heavily regulation process. Would we able to quantify the bubble risk with pertinent arguments, why not suggesting to regulators to include the bubble risk in their analysis ?By the way, I found some traces of this ancient work of mine in my archives. I am rewriting it a little bit as a working paper, since it is Mhh...let say..dusty. I propose to start another topic, which will be more technical, to discuss the mathematical ideas.Here are some related questions.Sound "Bubble formation in a simple Stochastic Market Model" a good title to you ?Is Wilmott forum the good place to discuss it ? If yes, is the general forum the good place for it ? Alternatively, I can put it on my company forum.I suspect that regulators are under a similar pressure to risk managers, coming from politicians who's career rests on a bubble economy.Should regulators care about bubble risk models? Perhaps. Certainly academics should study what leads people to sign up to ridiculous fads and fashions, and how we can prevent people passing the costs of such behaviour onto the rest of us. On the other hand, I think some sort of 'Bubble Var' would probably be an example of the kind of pseudoscience that banks love, and hasn't helped prevent the current situation. So I am in favour of qualitative models that might teach us something about human nature, but not quantitative models that claim to make accurate forecasts, but don't really provide insight.Also, for the current crisis, we didn't need a bubble model. The bubble in houses and consumer credit (in the UK) was totally obvious, you didn't need a model to see that the crash was coming.
ABOUT WILMOTT

PW by JB

Wilmott.com has been "Serving the Quantitative Finance Community" since 2001. Continued...


Twitter LinkedIn Instagram

JOBS BOARD

JOBS BOARD

Looking for a quant job, risk, algo trading,...? Browse jobs here...


GZIP: On