mathematical modelling of stock markets is....

[garcia]

...utter bollocks, part 47

[Brandenburger Toro]

mathematical modelling of stock markets is....

It's not bollocks, it's modelling. Anybody who confuses a model with reality is... well, in disconcertingly wide company these days. But also an idiot. Modelling is modelling, it is what it is. It's useful in the real world only insofar as the model reflects the real world, which is a systematically different question.

[garcia]

mathematical modelling of stock markets is....

jesus, that must be one of the most pedantic responses i've ever seen.

It's useful in the real world only insofar as the model reflects the real world

i.e., it isn't useful in the real world. apologies if my crude shorthand of "bollocks" failed to communicate this obviously enough.

[Brandenburger Toro]

mathematical modelling of stock markets is....

Pedantic it may be, but you're still missing my point. There's nothing wrong with modeling, the problem arises if you take modelling to be some sort of secular prophecy. IE, if you don't understand it in the slightest.

[Ginger Yellow]

mathematical modelling of stock markets is....

Yeah, but the problem is that huge numbers of investors do (or did, and will again once the memories of the last year fade) treat it as secular prophecy. Indeed you could argue that this whole crisis was caused by mass map/territory confusion. There are (were) hundreds of billions of dollars in algorithmic hedge funds whose investment strategy is 100% model based. And beyond that, much of the financial system is (was) based on the overuse of models as a substitute for qualitative human judgements.

[The Awesome Berbaslug!!!]

mathematical modelling of stock markets is....

it would appear that that article is about someone slowly coming to terms with the idea that returns aren't in fact normally distributed, which is a rather large assumption to make in the first place, and then they go on to offer a load of evidence as to why they should have come to this conclusion years ago. Well done boys.

[Ginger Yellow]

mathematical modelling of stock markets is....

Indeed. A simple look at the two charts of what normally distributed returns might look like and what returns actually look like would suggest they didn't need last year's events to falsfify the hypothesis. The conclusion is a bit bizarre as well, given that it jumps from talking about how the returns of the simplest financial asset around don't follow the models to saying that banks shouldn't be allowed to hold complex financial assets. And it's not as if (most) banks have suffered huge losses from equity derivatives anyway, compared to their other losses. SG being the obvious exception.

[Ginger Yellow]

mathematical modelling of stock markets is....

While we're on the subject, there's a good piece on the foolish use of normal distribution assumptions in Value at Risk models here. And to back up Yves Smith's point about the flaws being well known, RBS put out a research report well before the recent stock market volatility pointing out that fat tails mean VaR drastically understates risk.