standard financial instruments are broken

[1338 h4x0r]

and they keep doing it anyway

[Light Fantastic_]

didnt ltcm pop up out of nowhere on the back of the black-scholes model and rode it into ruin

like it was literally all they did

merrill lynch is an ok example though i'll read through it later

[1338 h4x0r]

Originally Posted by Light Fantastic_

didnt ltcm pop up out of nowhere on the back of the black-scholes model and rode it into ruin

like it was literally all they did

you can see an overview here and follow references for more details perhaps

http://en.wikipedia.org/wiki/Long-Term_Capital_Management#Trading_strategies

[Smokey D]

So... LTCM was ruined because it stopped doing what made it successful?

[Light Fantastic_]

from what i remember their glorious model failed to deal appropriately with tail risk

[Light Fantastic_]

which is what chris is whining about btw

[1338 h4x0r]

Mandelbrot uses the metaphor several times in his book that investors should be like shipwrights: build your ship for the 5% of the time when weather is really bad and not just for the remaining fair weather

[Smokey D]

Well he's entitled to whatever philosophy of investment theory he might have but the object of investment is to benefit the investors. Whether benefits accrue to society at large is, strictly speaking, beside the point. A firm doesn't need to be infallible for it to benefit its investors handsomely, it only needs to be pretty successful. This is a crucial point. Financial modelling isn't supposed to describe the world perfectly. It is only ever a heuristic with the much more limited goal of benefiting shareholders.

[1338 h4x0r]

Originally Posted by Smokey D

Well he's entitled to whatever philosophy of investment theory he might have but the object of investment is to benefit the investors. Whether benefits accrue to society at large is, strictly speaking, beside the point. A firm doesn't need to be infallible for it to benefit its investors handsomely, it only needs to be pretty successful.

he addresses that (bogus) point in his book too

[Smokey D]

It's not bogus.

[1338 h4x0r]

an investment scheme that takes heavy tails into account would also be only "pretty successful"

but it would avoid disaster far more effectively

the same goes for using the Gaussian in financial risk management as goes for rational choice theory in economics: it makes simplifications but is that an excuse for its uselessness?

[1338 h4x0r]

and for fuck's sake it's not like switching to empirically demonstrated distributions is that hard

[Smokey D]

I'm not saying LTCM used an optimal strategy. I'm saying that not using an optimal strategy doesn't necessarily prove the strategy is flawed for the purposes of the firm.

[1338 h4x0r]

Originally Posted by Smokey D

I'm not saying LTCM used an optimal strategy. I'm saying that not using an optimal strategy doesn't necessarily prove the strategy is flawed for the purposes of the firm.

apparently it was in their case

[Smokey D]

Well, they ended up going bust sure. The point is that "Mandelbrot uses the metaphor several times in his book that investors should be like shipwrights: build your ship for the 5% of the time when weather is really bad and not just for the remaining fair weather" is dumb and inconsistent with how the financial system works.

[1338 h4x0r]

Originally Posted by Smokey D

Well, they ended up going bust sure. The point is that "Mandelbrot uses the metaphor several times in his book that investors should be like shipwrights: build your ship for the 5% of the time when weather is really bad and not just for the remaining fair weather" is dumb and inconsistent with how the financial system works.

explain

what about the financial system necessitates grossly underestimating risk

[Light Fantastic_]

he's just saying you can do a less than ideal job and still get a return

[Smokey D]

The only question that's ever relevant is whether the firm makes money for its investors over the medium term. If it does, it's irrelevant that it won't survive a 5/100 event.

[1338 h4x0r]

chad:

you can also play Russian Roulette while assuming that 1/9001 chambers are loaded and live

Originally Posted by Smokey D

The only question that's ever relevant is whether the firm makes money for its investors over the medium term. If it does, it's irrelevant that it won't survive 5/100 event.

why not make better decisions when it wouldn't cost much to do so?

[Light Fantastic_]

Originally Posted by 1338 h4x0r

you can also play Russian Roulette while assuming that 1/9001 chambers are loaded and live

not really apt since theres no incentive to play russian roulette in the first place

and bankruptcy is way nicer than dying