Quants: The Alchemists Of Wall Street

[Harry Sacks]

You're gonna love this guys!

Have a nice day! :-)

Edit: Spelling.

Edited September 2, 2011 by dom

[Dorkins]

Thanks for that, it was a good watch.

[The Spaniard]

Thanks for the post, dom.

Good to hear that some quants are questioning their contribution to society at large, in particular Paul Wilmott's realisation that all he does is move numbers around.

Maybe next they'll start to question the debt-based money system that enabled the blowing of the unrepayable credit bubble upon which they feasted.

What a crazy parasitic monster banking has become.

[Englebert]

Watched the first 5 minutes, and found it quite boring to be honest. It was like stating the bleedin' obvious. It didn't alter my view one iota on what qcunnts these guys are. All those financial gurus....I don't really want to see their faces. I don't want to hear them talk, don't want to personalise them. To me, they are not real. If this is a video of those that have fecked up the system, now professing a regret at what they were complicit in, then I'm outta here.

[richrich]

Watched the first 5 minutes, and found it quite boring to be honest. It was like stating the bleedin' obvious. It didn't alter my view one iota on what qcunnts these guys are. All those financial gurus....I don't really want to see their faces. I don't want to hear them talk, don't want to personalise them. To me, they are not real. If this is a video of those that have fecked up the system, now professing a regret at what they were complicit in, then I'm outta here.

You probably don't even understand half the things he's talking about.

[wonderpup]

You probably don't even understand half the things he's talking about.

The problem is neither did he.

[Bloo Loo]

You probably don't even understand half the things he's talking about.

that was the WHOLE idea.

It is the ESSENCE of the confidence tricksters game.

[Si1]

17 minutes in the english quant guy says that quants modelled CDO risk based on previous 5 years of housing market data, ergo on the basis that house prices don't fall

[fluffy666]

17 minutes in the english quant guy says that quants modelled CDO risk based on previous 5 years of housing market data, ergo on the basis that house prices don't fall

Clearly they need to pay more, the guys who did that clearly weren't the sharpest knives in the rack.

(Unless, of course, the question was 'How do we make these CDOs look low-risk', not 'How risky are these CDOs?')

[Si1]

Clearly they need to pay more, the guys who did that clearly weren't the sharpest knives in the rack.

(Unless, of course, the question was 'How do we make these CDOs look low-risk', not 'How risky are these CDOs?')

might be simpler - one more responsible quant says I can only do this with 150 years' data, it will take me 8 months to normalise it

sh*t quant says I can do it with 5 years non-normalised data

2nd guy gets the job because he effectively costs half as much as he claims can do the same job in half the time

[newbie]

Quants: turning gold into lead.

[Englebert]

You probably don't even understand half the things he's talking about.

Too right I don't! I do understand though what a marvellous job these qcunnts have done to Feck up the Country...

[200p]

How do i become a quant then?

[A.steve]

How do i become a quant then?

Answer: http://www.cqf.com/

I watched the film (again - I think - I had a sense of deja-vous) in the background... One thing struck me as being specially relevant, but was something I didn't quite understand... in LaTeX notation:

F_t = | F_{XX} |

Obviously, I understood what he said about the absolute value being non-linear, that's obvious. What I'm unclear about is the meaning of the formula. I'm guessing that F_t relates to some quantity or vector that varies with time... but what is "Fxx"? It makes no sense if it's a scalar (there'd be too little meaning) and a vector gives little more meaning. If it were a square matrix the absolute value would be synonymous with the determinant... but... then I don't understand the XX subscripts... One guess is that they relate to the row/column indices being identical - perhaps suggesting a Hermitian (symmetric) matrix? Does this formula indicate a system of linear algebra to value a portfolio? I'm not sure - but assume someone might recognise it.

What made me focus on this formula is that what was said about it corresponds with an idea I've had about formulas themselves. He said "Beauty is having the right level of complexity, but not too much" - which, I think, might be profound - if interpreted in a particular way - or, to be specific, substitute "Profitability" for "Beauty" to switch to a money oriented mind-set.

While I'm not intending to suggest that quants approached things this way, consider this story: The idea is that a 'beautiful' model is one that people accept... this requires plausible sounding assumptions and enough symbolic transformation to bog-down the audience questioning the (absolutely provably correct) symbolic manipulations rather than the initial assumptions. In this sense, I think that a beautiful equation might be synonymous with a successful illusion. I'm wondering if it's possible to measure the complexity of models to establish an estimate for when they'll be broken... a bit like analysing conjuring tricks to establish how long they're likely to remain without imitation. Linear models are easily imitated - assuming similar input data... so I don't see a future for them... but I do wonder if the next bubble can be identified just by looking for the right level of sophistry? Am I reading too much into the comment about "beauty"? Perhaps... but it's an interesting concept - I think - and it makes me curious about where the formula originated. Has anyone seen it before?

[Scunnered]

Answer: http://www.cqf.com/

I watched the film (again - I think - I had a sense of deja-vous) in the background... One thing struck me as being specially relevant, but was something I didn't quite understand... in LaTeX notation:

F_t = | F_{XX} |

Obviously, I understood what he said about the absolute value being non-linear, that's obvious. What I'm unclear about is the meaning of the formula. I'm guessing that F_t relates to some quantity or vector that varies with time... but what is "Fxx"? It makes no sense if it's a scalar (there'd be too little meaning) and a vector gives little more meaning. If it were a square matrix the absolute value would be synonymous with the determinant... but... then I don't understand the XX subscripts... One guess is that they relate to the row/column indices being identical - perhaps suggesting a Hermitian (symmetric) matrix? Does this formula indicate a system of linear algebra to value a portfolio? I'm not sure - but assume someone might recognise it.

Partial derivatives? F could be some function of X and t, with F_t = dF/dt and F_XX = d²F/dX².

Edit. Google works! If you look at http://www.google.co.uk/webhp?q=f_t+f_xx you'll find people on a quant forum talking about pricing options or something. They're talking about PDEs and it looks as if the subscripts do denote differentiation. No idea what it all actually means though.

Edit again. I think the equation above is something like the one-dimensional heat equation: http://en.wikipedia.org/wiki/Heat_equation .

Edited September 2, 2011 by Scunnered

[A.steve]

Partial derivatives? F could be some function of X and t, with F_t = dF/dt and F_XX = d²F/dX².

Edit. Google works! If you look at http://www.google.co.uk/webhp?q=f_t+f_xx you'll find people on a quant forum talking about pricing options or something. They're talking about PDEs and it looks as if the subscripts do denote differentiation. No idea what it all actually means though.

Hmmm - yes, I didn't think of googling for that pseudo-latex... (it's wrong, as F_xx formats only the first x as a subscript - but let's drop the discussion about typography...)

On the forum (first result) they're discussing a formula to price (harmonic Asian) options - using a modified Black Scholles approach.

- f_t = 0.5 sig^2 x^2 f_xx - (r x + 1/T) f_x

-f_t => This looks like the rate of change of the price of the option over time

sig^2 => Standard deviation squared - the variance of the price.

x => The spot price of the underlying instrument

f_x => The first derivative (rate of change) of the price of the option with respect to spot price.

f_xx => The second derivative (rate of change of rate of change) of the price of the option with respect to spot price.

r => The relevant interest rate

1/T => This is stated to be the reciprocal of the time to maturity/expiry.

Absent from the standard Black Scholes formula is an additional term r*f - which relates to the interest on the price of the option. This kind-of makes sense if we assume they're talking about at-the-money options with small prices. They say they're concerned by Asian options - so prices would be smaller than for European or American options at similar nominal values. What makes very little sense at all is the 1/T (which is commented as making things hard...) It makes things hard because it's dimensionally incompatible with the rest of the formula... r*x is a ratio times a price (which has currency as a unit) - but 1/T has 'per day' as a unit - and adding them together is a syntactic error - even without trying to work out what it means. I suspect the error is with bracketing... but I'm not sufficiently confident to supply the correction... I believe that the bit that makes the option 'harmonic' is that the payout is proportional to the mean price during a window... which means the payout is zero if there's no change in price - and the payout is zero, too, if the price change traces a sinusoid during the window... essentially the window just tames volatility for the purposes of determining the "right price" for the underlying instrument... though this doesn't explain why the poster chose the maturity of the option as the window period... perhaps that's common practice - perhaps it's just a mistake.

I agree - however, it looks like f_xx is intended to mean second derivative of f with respect to x and f_t is intended to mean the first derivative of f with respect to t.

The other post talks about:

f_t = f_xx - c f^2

Here, I'm finding it harder to establish an interpretation. It says the rate of change with respect to time is the rate of change of rate of change of that thing with respect to something else less some constant factor times the square of the value of the thing. I can't think of anything that behaves like that... at least not off the top of my head. Perhaps that's the point, however... as they're talking about research problems... and an obscure system is likely to be one for which a solution isn't already published.

Typography aside, while I have some handle on the equations in the forums... I'm still baffled by the equation on the blackboard in the film.

f_t = | F_xx |

The rate of increase of some quantity is equal to the absolute size of the rate of rate of change of the same quantity relative to something else? The only thing I can think of that fits that sort of model would be an orbiting object with a centripetal force... measured from the perspective of the orbit rather than the central point... the bigger the centripetal force, the faster the orbit.... but we need to look at the absolute value of the centripetal force as, the centripetal force is always orthogonal to the orbit's trajectory. So far, so good... this is equivalent to the idea that the right hand side might be the determinant of an hermitian matrix (the matrix that describes a circle has a specific hermitian form).

The bad news is three-fold. First, I don't see how this can be described as 'beautiful' - there are far more elegant ways to describe the dynamics of orbits. Second, he bangs on about it being 'non-linear' - and, indeed, absolute value (and determinant - in the general case) are non-linear... but orbits can be perfectly modelled using linear techniques - so the complexity is fake if this is the interpretation. Finally, I fail to see how this could be a 'result' in quantitative finance... something that only ever goes up... at a rate that's determined by it's rate of change relative to something else. (Please, goldbugs, don't you start with the ********, too.)

So, ultimately, I'm still confused... Perhaps that's the point though... all you need to do to create the illusion that you have something is to say a bunch of things that aren't wrong, but don't fit together... and refuse to answer questions about it. Kind of sums up how I see quant models to have been deployed. :-S

Edit again. I think the equation above is something like the one-dimensional heat equation: http://en.wikipedia.org/wiki/Heat_equation .

Hmmm - there's a connection... I think - but not a one-to-one correspondence. It doesn't need to be one-dimensional - as everything in the equation can be considered a vector quantity just as easily as a scalar quantity... hence dimensionality is up for us to choose.

The problem I have with the heat equation explanation (as interesting/compelling as it is) - is the fact that the rate of change of temperature is linearly proportional to the rate of change of rate of change of temperature. There's no absolute value in heat equations... as far as I can see... hence no non-linearity... and the non-linearity was (allegedly) the reason the presenter gave us the equation.

Edited September 2, 2011 by A.steve

[Bloo Loo]

Answer: http://www.cqf.com/

I watched the film (again - I think - I had a sense of deja-vous) in the background... One thing struck me as being specially relevant, but was something I didn't quite understand... in LaTeX notation:

F_t = | F_{XX} |

Obviously, I understood what he said about the absolute value being non-linear, that's obvious. What I'm unclear about is the meaning of the formula. I'm guessing that F_t relates to some quantity or vector that varies with time... but what is "Fxx"? It makes no sense if it's a scalar (there'd be too little meaning) and a vector gives little more meaning. If it were a square matrix the absolute value would be synonymous with the determinant... but... then I don't understand the XX subscripts... One guess is that they relate to the row/column indices being identical - perhaps suggesting a Hermitian (symmetric) matrix? Does this formula indicate a system of linear algebra to value a portfolio? I'm not sure - but assume someone might recognise it.

What made me focus on this formula is that what was said about it corresponds with an idea I've had about formulas themselves. He said "Beauty is having the right level of complexity, but not too much" - which, I think, might be profound - if interpreted in a particular way - or, to be specific, substitute "Profitability" for "Beauty" to switch to a money oriented mind-set.

While I'm not intending to suggest that quants approached things this way, consider this story: The idea is that a 'beautiful' model is one that people accept... this requires plausible sounding assumptions and enough symbolic transformation to bog-down the audience questioning the (absolutely provably correct) symbolic manipulations rather than the initial assumptions. In this sense, I think that a beautiful equation might be synonymous with a successful illusion. I'm wondering if it's possible to measure the complexity of models to establish an estimate for when they'll be broken... a bit like analysing conjuring tricks to establish how long they're likely to remain without imitation. Linear models are easily imitated - assuming similar input data... so I don't see a future for them... but I do wonder if the next bubble can be identified just by looking for the right level of sophistry? Am I reading too much into the comment about "beauty"? Perhaps... but it's an interesting concept - I think - and it makes me curious about where the formula originated. Has anyone seen it before?

looks a bit like a wabbit, or fork bomb

has the same effect of collapsing any system it runs in where other participants are kept in the dark.

){ :|:& };:

EDIT...a part of the code is turned into a sad face by the system editor....how appropriate!

Edited September 3, 2011 by Bloo Loo

[MrPin]

Maths can generalise common sense into gibberish!

[erranta]

that was the WHOLE idea.

It is the ESSENCE of the confidence tricksters game.

And you MASONS should know ALL about that!

Just 'aving a look thru your Turk-Sufi roots coming out of Pre-Moslem/Muslim Afghanistan area!

Loads of Dragons involved!

Never realised "Hindu Kush" Mountains split east/west (not just the country!)

(Funny that woz Bin Laden hideout along with Taliban who constantly grow addictive-misery Opium Poppys with old Egyptian (Blue) Lapis Lazuli mines - It was used as the blue colour on 'old master' paintings and Temple buildings)

'HAL' pops up as part of the Shamanistic djinn system - remember Hal 9000?

(Odds on coincidence of that word being chosen & used pleeze)

Space Odyssey = Star Trek = Trek in Afghan Mts

(Od-yssey = Homer > Greek Homerus also means 'blind'

(highly elitist and related to Biblical stuff + dumbed-down cartoon 'blind' man 'Homer' Simp-leton-son)

You know the Satanists are up to no good - here's a link >>>

"Leonardo used oil and tempera paints on dry plaster, an experimental technique, and as a result, the Last Supper is now so faded and cracked it can't withstand exposure to bright light. To protect the painting, HAL9000 worked with restoration specialists at Rome's Istituto Centrale per il Restauro to develop a lighting system without the ultraviolet emissions and high thermal impact so hazardous to works of art. Shot with a Nikon D2X digital SLR

in just nine hours, the total impact of the digitization process was equal to just a few minutes of the soft lighting that normally illuminates the painting."

You would have thought - if GOD were pleased with the "Last Supper "painting he would not let it fade and fall apart whilst it is kept in semi darkness! The tempura is just an excuse!

Think about it - A message to all true believers throughout the World

Different types of tempera paint have been in use since ancient Egypt. Renaissance artists used egg yolk as binding agent, mixing in colored pigments dry gesso with charcoal made from burnt willow twigs

Lapis Lazuli can be ground into powder and mixed with binding agents to create watercolor, tempura and oil paints in a brilliant blue called ultramarine

A color known as hematite is red. This color is natural, and it is a very strong and solid stone. And it is so solid and perfect that stones and crooks are made of it for burnishing gold on panel; and they acquire a black and perfect color, dark as a diamond.?

- the Shamans make their initiates repeat and remember certain combinations of numbers to conjure up - be in touch with their djinns.

You would be shocked reading thru it and comparing that to Masonic 'Mysteron' higher degrees !

Appley the 'Math' to 'Quants' & the essence of their Banking = Djinn & Tonic?

Edited September 3, 2011 by erranta

[FIGGY]

Maths can generalise common sense into gibberish!

Can it turn an Erranta post into common sense thought, if it can all hail the maths?

[MrPin]

Can it turn an Erranta post into common sense thought, if it can all hail the maths?

I think A.Steve is aiming for the Erranta medal!

[algobet]

Hmmm - yes, I didn't think of googling for that pseudo-latex... (it's wrong, as F_xx formats only the first x as a subscript - but let's drop the discussion about typography...)

On the forum (first result) they're discussing a formula to price (harmonic Asian) options - using a modified Black Scholles approach.

- f_t = 0.5 sig^2 x^2 f_xx - (r x + 1/T) f_x

-f_t => This looks like the rate of change of the price of the option over time

sig^2 => Standard deviation squared - the variance of the price.

x => The spot price of the underlying instrument

f_x => The first derivative (rate of change) of the price of the option with respect to spot price.

f_xx => The second derivative (rate of change of rate of change) of the price of the option with respect to spot price.

r => The relevant interest rate

1/T => This is stated to be the reciprocal of the time to maturity/expiry.

Absent from the standard Black Scholes formula is an additional term r*f - which relates to the interest on the price of the option. This kind-of makes sense if we assume they're talking about at-the-money options with small prices. They say they're concerned by Asian options - so prices would be smaller than for European or American options at similar nominal values. What makes very little sense at all is the 1/T (which is commented as making things hard...) It makes things hard because it's dimensionally incompatible with the rest of the formula... r*x is a ratio times a price (which has currency as a unit) - but 1/T has 'per day' as a unit - and adding them together is a syntactic error - even without trying to work out what it means. I suspect the error is with bracketing... but I'm not sufficiently confident to supply the correction... I believe that the bit that makes the option 'harmonic' is that the payout is proportional to the mean price during a window... which means the payout is zero if there's no change in price - and the payout is zero, too, if the price change traces a sinusoid during the window... essentially the window just tames volatility for the purposes of determining the "right price" for the underlying instrument... though this doesn't explain why the poster chose the maturity of the option as the window period... perhaps that's common practice - perhaps it's just a mistake.

I agree - however, it looks like f_xx is intended to mean second derivative of f with respect to x and f_t is intended to mean the first derivative of f with respect to t.

The other post talks about:

f_t = f_xx - c f^2

Here, I'm finding it harder to establish an interpretation. It says the rate of change with respect to time is the rate of change of rate of change of that thing with respect to something else less some constant factor times the square of the value of the thing. I can't think of anything that behaves like that... at least not off the top of my head. Perhaps that's the point, however... as they're talking about research problems... and an obscure system is likely to be one for which a solution isn't already published.

Typography aside, while I have some handle on the equations in the forums... I'm still baffled by the equation on the blackboard in the film.

f_t = | F_xx |

The rate of increase of some quantity is equal to the absolute size of the rate of rate of change of the same quantity relative to something else? The only thing I can think of that fits that sort of model would be an orbiting object with a centripetal force... measured from the perspective of the orbit rather than the central point... the bigger the centripetal force, the faster the orbit.... but we need to look at the absolute value of the centripetal force as, the centripetal force is always orthogonal to the orbit's trajectory. So far, so good... this is equivalent to the idea that the right hand side might be the determinant of an hermitian matrix (the matrix that describes a circle has a specific hermitian form).

The bad news is three-fold. First, I don't see how this can be described as 'beautiful' - there are far more elegant ways to describe the dynamics of orbits. Second, he bangs on about it being 'non-linear' - and, indeed, absolute value (and determinant - in the general case) are non-linear... but orbits can be perfectly modelled using linear techniques - so the complexity is fake if this is the interpretation. Finally, I fail to see how this could be a 'result' in quantitative finance... something that only ever goes up... at a rate that's determined by it's rate of change relative to something else. (Please, goldbugs, don't you start with the ********, too.)

So, ultimately, I'm still confused... Perhaps that's the point though... all you need to do to create the illusion that you have something is to say a bunch of things that aren't wrong, but don't fit together... and refuse to answer questions about it. Kind of sums up how I see quant models to have been deployed. :-S

Hmmm - there's a connection... I think - but not a one-to-one correspondence. It doesn't need to be one-dimensional - as everything in the equation can be considered a vector quantity just as easily as a scalar quantity... hence dimensionality is up for us to choose.

The problem I have with the heat equation explanation (as interesting/compelling as it is) - is the fact that the rate of change of temperature is linearly proportional to the rate of change of rate of change of temperature. There's no absolute value in heat equations... as far as I can see... hence no non-linearity... and the non-linearity was (allegedly) the reason the presenter gave us the equation.

perhaps it was just edited in the program to refer to non-linear relationships. that was my take-away. good of you to spend time on it though.

[nmarks]

Partial derivatives? F could be some function of X and t, with F_t = dF/dt and F_XX = d²F/dX².

Edit. Google works! If you look at http://www.google.co.uk/webhp?q=f_t+f_xx you'll find people on a quant forum talking about pricing options or something. They're talking about PDEs and it looks as if the subscripts do denote differentiation. No idea what it all actually means though.

Edit again. I think the equation above is something like the one-dimensional heat equation: http://en.wikipedia.org/wiki/Heat_equation .

Its all greeks to me.

[A.steve]

perhaps it was just edited in the program to refer to non-linear relationships. that was my take-away. good of you to spend time on it though.

Perhaps... Non-linearity is very important - it's a big part of what makes Black Scholes bogus... The algorithm assumes that the rate of change of rate of change is constant... while it often appears to be the case, this is just an artefact of our attempt to measure it. In reality, a price only exists the instant at which a trade is made - and the price only applies to the traded units. The upshot of this is that, pretty much, all linear algebra models for prices are bogus - because prices simply aren't continuous. Initially, the model is 'forced' to fit the available facts - then... if it continues to hold as the future unfurls, it does so only because people defer to the model over other opinion, and it becomes a self-fulfilling prophecy. It's as true as predictions from astrology... and for the same reasons.

I thought it particularly interesting as it was a claim that they'd proved that something only ever goes up (all be it at a rate that might involve complicated calculation.)

[winkie]

-Banking Has lost touch with its purpose and has now become dangerous.....there are more people trading,speculation on food/commodities than involved in the production...betting on the future price, moving money around.

-The gamblers can't lose they get a bonus making other people money using their money... they also make a bonus and take a fee from losing other peoples money....the investment banker can therefore never lose.

Two important points made.....nothing has changed....not clever only blinding with science :-(