Help! Still trying to get my head around it

SkepticOne · 01-02-2009 10:43PM

nesf wrote: »

It's not a credibility thing, I'm only asking because the attitude you describe doesn't fit with the kind of people I've met in the empirical side of academic mathematical modelling in economics and finance who routinely reject mainstream beliefs and try other ideas when they don't fit with the data. The people at the coalface of social science disciplines (i.e. testing the theories against the data) tend to be the least likely to consider the theory as gospel because every day they see how it doesn't square up exactly or even close sometimes.

Edit: I can see how someone might think that of quants after reading the headlines of their misbehaviour in the markets over the past few months but I can assure you that among academics interested in modelling reality you'll tend to find little dogmatism about theory because, as they say, the data is king.

I'm hesitant to interject in a very interesting discussion but, with respect, I think some of the attitude JDLK is talking about (if I understand correctly) comes across in a lot of the posts here. It is the attitude of not taking full responsibility, not a belief that a theory is gospel. If a model fails, or there is failure to fully account for uncertainty or uncertainty about uncertainty then a person fails, not merely the model.

JDLK · 01-02-2009 10:49PM

SkepticOne wrote: »

I'm hesitant to interject in a very interesting discussion but, with respect, I think some of the attitude JDLK is talking about (if I understand correctly) comes across in a lot of the posts here. It is the attitude of not taking full responsibility, not a belief that a theory is gospel. If a model fails, or there is failure to fully account for uncertainty or uncertainty about uncertainty then a person fails, not merely the model.

This is in essence what Im trying to say about the intangibale value of uncertainty which I beleive is misrepresented in finacial advice.

Specifically, I think it comes across as the kind of attitude form analysts which readily accepts praise for correct predictions yet just as readily distances themselves from inaccurate predictions, but it is not within the analysts which I think the blame lies- I think the system which predicts certainty is not a mirror for uncertainty and that uncertainty is being undervalued- this has to be a logical conclusion to the current events- even if probabilities were accurate given the amount of available data we cannot deny their inaccuracy or explain them away mathmatically in the face of the system failure to which the models sought to serve

nesf · 01-02-2009 10:54PM

SkepticOne wrote: »

I'm hesitant to interject in a very interesting discussion but, with respect, I think some of the attitude JDLK is talking about (if I understand correctly) comes across in a lot of the posts here. It is the attitude of not taking full responsibility, not a belief that a theory is gospel. If a model fails, or there is failure to fully account for uncertainty or uncertainty about uncertainty then a person fails, not merely the model.

I wouldn't disagree with any of that, I'd consider the people involved in the modelling to have messed up big time. It might have been a bunch honest mistakes (I don't know) but they are still responsible for what happened.

What worries me though is that areas of quantitative finance that have worked fine, like futures, Foreign Exchange markets etc are getting tarred with the same brush even though these were a world away from CDOs.

SkepticOne · 01-02-2009 10:58PM

This video I think illustrates something or other

http://ie.youtube.com/watch?v=eKPQLl5rupg

Now the thing about this is, the meteorological office didn't go into a big huff after their (I think it is important to emphasise "their") failure. They admitted they got it wrong and moved on.

We all know it is very hard to predict the whether and it was even harder back in 1987. But the point is they did not say that their models were sound but the weather failed them or that the weather system went outside the parameters of their model. This would have been a cop out.

nesf · 01-02-2009 10:58PM

JDLK wrote: »

Specifically, I think it comes across as the kind of attitude form analysts which readily accepts praise for correct predictions yet just as readily distances themselves from inaccurate predictions, but it is not within the analysts which I think the blame lies- I think the system which predicts certainty is not a mirror for uncertainty and that uncertainty is being undervalued- this has to be a logical conclusion to the current events- even if probabilities were accurate given the amount of available data we cannot deny their inaccuracy or explain them away mathmatically in the face of the system failure to which the models sought to serve

I'll give you an example that's oft quoted in academic circles to show the madness that was going on in some of the banks (this is from back in September 2007):

David Viniar, the bank’s chief financial officer, denied at the time that Goldman was rescuing the fund, saying that it saw a “good investment opportunity”.

He blamed the losses on “25 standard deviation moves, several days in a row”.

http://www.ft.com/cms/s/0/bbcf8df0-5cb1-11dc-9cc9-0000779fd2ac.html?nclick_check=1

25 standard deviation moves, seven days in a row. i.e. seven days in a row an event, whose probability means that barely one of them should have happened since the beginning of the universe at most, happened.

To say that this pronouncement is held in contempt in the mathematical modelling communities is being kind...

JDLK · 01-02-2009 11:02PM

SkepticOne wrote: »

This video I think illustrates something or other

http://ie.youtube.com/watch?v=eKPQLl5rupg

Now the thing about this is, the meteorological office didn't go into a big huff after their (I think it is important to emphasise "their") failure. They admitted they got it wrong and moved on.

We all know it is very hard to predict the whether and it was even harder back in 1987. But the point is they did not say that their models were sound but the weather failed them or that the weather system went outside the parameters of their model. This would have been a cop out.

Maybe its about time we regulated the weathermen:D

While I accpet the point the implications of global financial collapse are alot more unaccptable. Many of the worst politcial movements were born in recession- though that is going slightly off topic

SkepticOne · 01-02-2009 11:03PM

nesf wrote: »

What worries me though is that areas of quantitative finance that have worked fine, like futures, Foreign Exchange markets etc are getting tarred with the same brush even though these were a world away from CDOs.

Is it unfair to tar them with the same brush? How do we know that? Mathematical modeling?

SkepticOne · 01-02-2009 11:04PM

JDLK wrote: »

Maybe its about time we regulated the weathermen:D

Or the weather.

nesf · 01-02-2009 11:06PM

SkepticOne wrote: »

Is it unfair to tar them with the same brush? How do we know that? Mathematical modeling?

Yup. We're self regulating.

JDLK · 01-02-2009 11:13PM

Its hard to justify any part of a financial system which has almost completely failed, a system which is vehement about free trade yet ultimately forces the governement to nationalise institutions. It reminds me of the communists defending the USSR in the face of its collapse

The original point is that a revision of independant regulation must be put in place- even if this was superficial but enough to restore confidence in the short run with a "lesson learned" attitude for the long term then that would be something- though I dont think confidence will be restored too soon and if it isnt then people wont be using these models, and if the models arent being used then it doesnt matter how sound they are, they become redundant

01-02-2009 11:13PM

SkepticOne wrote: »

Is it unfair to tar them with the same brush? How do we know that? Mathematical modeling?

Forex futures markets, despite not being Keynesian, seem to have worked well since their inception. CIP holds for the majority of cases and most deviations are marginally profitable. Same goes for a lot of commodity market futures, again not being Keynesian in nature. They work, simple as that. There is sound (and reasonably simple theory) behind a lot of futures market.

I am sure that JDLK, with his masters in business, knows the benefits of locking in future prices and I again reiterate - these markets work.

nesf · 01-02-2009 11:18PM

SkepticOne wrote: »

Is it unfair to tar them with the same brush? How do we know that? Mathematical modeling?

It helps to understand CDOs in more detail to see why they failed so utterly (it wasn't just the quants at fault). If you look at other markets like Forex Futures, you'll see that they function very differently.

This Economist article is a decent jargon-free look at CDOs which I'd recommend everyone to read: http://www.economist.com/specialreports/displaystory.cfm?story_id=12957753*

*Note: It's both critical of the investment banks, the quants and mathematical modelling itself when poorly applied so it's no apology for the professional wing of the discipline.

nesf · 01-02-2009 11:19PM

JDLK wrote: »

Its hard to justify any part of a financial system which has almost completely failed, a system which is vehement about free trade yet ultimately forces the governement to nationalise institutions. It reminds me of the communists defending the USSR in the face of its collapse

I don't agree with your logic here. Just because one aspect of a system fails doesn't necessarily mean we should throw away the system. We most certainly do need to regulate the banks better, that isn't up for debate and should be accepted as a given.

JDLK · 01-02-2009 11:25PM

Deleted User wrote: »

Forex futures markets, despite not being Keynesian, seem to have worked well since their inception. CIP holds for the majority of cases and most deviations are marginally profitable. Same goes for a lot of commodity market futures, again not being Keynesian in nature. They work, simple as that. There is sound (and reasonably simple theory) behind a lot of futures market.

I am sure that JDLK, with his masters in business, knows the benefits of locking in future prices and I again reiterate - these markets work.

I knew that would come back to bite me, i did try and remain neutral on the whole credibility thing.

But I think this is the fundemental problem which those within quant cannot rise above, cannot see the wood for the tree's if you will, the system is only as good as the confidence of the people using it- this system has failed and within that very loss of confidence is a value which is only now being quantified. I can tell you my cafr is excellent but if you are too afraid to use it then it is redundant

The economic stimulus packages now being put in place are pure Keynsian.

When the ppropensity to hoard money takes grip there is no point throwing money at people to stimulate spending as people will simply continue to save- this is what Keyenes meant by fiscal policy, the real issue is confidence- market confidence is where the real trading takes place and defending a system which is bankrupting millions will not restore confidence

JDLK · 01-02-2009 11:30PM

nesf wrote: »

I don't agree with your logic here. Just because one aspect of a system fails doesn't necessarily mean we should throw away the system. We most certainly do need to regulate the banks better, that isn't up for debate and should be accepted as a given.

Im not talking about the system of capitalsim, Im talking about the corrupted system of deregulated free market capitalism, the belief in self correction and natural equilibrium

EDIT: anyway its getting late, thats it for me folks- good talk, some interesting perspectives! Take it easy

nesf · 01-02-2009 11:43PM

JDLK wrote: »

Im not talking about the system of capitalsim, Im talking about the corrupted system of deregulated free market capitalism, the belief in self correction and natural equilibrium

I would agree that this latest crash has highlighted the need for a serious root and branch examination of plenty of aspects of the financial system (especially how ratings are assigned, who assigns them and who pays the people who make the ratings! This would be one of my personal axes to grind on this topic.)

What I don't agree with that it has irrevocably ruined mathematical finance. The models never should have been treated as infallible and there should have been far more scepticism of the numbers they were pumping out. Mathematical finance does have a place in the world, especially in areas like derivatives and foreign exchange but the confluence of ineptitudes that created the sub-Prime/CDO mess can never be left to happen again.

SkepticOne · 01-02-2009 11:44PM

nesf wrote: »

It helps to understand CDOs in more detail to see why they failed so utterly (it wasn't just the quants at fault). If you look at other markets like Forex Futures, you'll see that they function very differently.

This Economist article is a decent jargon-free look at CDOs which I'd recommend everyone to read: http://www.economist.com/specialreports/displaystory.cfm?story_id=12957753

I'm sure they do, but as JDLK has pointed out, it is easy to see the flaws after the fact. Before the fact, these "flaws" would have been dismissed as remote theoretical possibities by people who thought they knew. Nevertheless I will read up further on these and other issues.

nesf · 02-02-2009 12:06AM

SkepticOne wrote: »

I'm sure they do, but as JDLK has pointed out, it is easy to see the flaws after the fact. Before the fact, these "flaws" would have been dismissed as remote theoretical possibities by people who thought they knew. Nevertheless I will read up further on these and other issues.

Read up on people using bell curves (normal distributions) and risk. It's something that academics have been ranting about for years (Mandlebrot being a major one) that the people in the markets were dismissing.

It's part of the reason why the quote from Viniar is considered to be so silly. He's still talking about a world where market events can be described by normal distributions...

This 1997 debate about VAR is interesting because Taleb highlights some of technical problems going on. We can't precisely know the tail risks simply because we do not have enough data. It was what I was getting at above with the determinism stuff. The system is fairly deterministic, the problem is the data needed to accurately predict things with a "25 standard deviation probability" according to the normal distribution would require thousands of years of data in some cases. This is a hard mathematical barrier that you can't get around* and which undermines much of the silliness that went on risk assessment by rating agencies and banks with respect to CDOs and it's why many academics in the area despair about some of the stuff that went on. They should have known better!

*In sciences like Physics you can get around it by running experiments over and over and create enormous data sets to analyse as well as trying to control all the influences on the system you're studying. You can't do that in Economics which is why people need to be aware of it.

Help! Still trying to get my head around it

Comments