digital signal processing of stock prices

The May 2009 issue of Futures (as in financial derivatives) magazine
has a an interview with money manager Rafael Molinero, who has used
DSP algorithms to trade successfully. Based on what he says, I wonder
if someone can name the algorithms he may be using and provide
references to relevant books and/or papers.

Quoting the article:

He quickly became intrigued by DSP. [...]

Markets also move in waves similar to, but not identical to, the waves
of sound, leading Molinero to ask, "If markets have a non-random
component, can we extract the random one and keep the non-random one?"

This mean transforming existing algorithms, for a variety of reasons:
first, markets aren't sound waves, and second, DSP is designed to
rebuild something that previously existed (sound), while traders are
trying to project a bit into the future.

[...]

"We decompose the price into linear (trending) and non-linear
(cyclical) components, and then the algorithms identify which are
signals," Molinear explains., adding that what's left after filtering
the noise is probably the market's real pulse. Models also generate a
probability of success on each trade. "For example, if you can explain
a signal with very few components, it's probably a strong signal,
which we can use to project the next directional move."

This, he says, enables them to avoid getting whipsawed in choppy
markets and also to get into trending markets earlier.

"The models adapt in frequency based on the speed at which the market
is going," he explains. "By decomposing the market, we can look at the
underlying components. If these components change directions
frequently, our models will trade short-term, if not our models will
trade medium-term."

On May 6, 3:30&#4294967295;pm, Beliavsky <beliav...@aol.com> wrote:
> The May 2009 issue of Futures (as in financial derivatives) magazine
> has a an interview with money manager Rafael Molinero, who has used
> DSP algorithms to trade successfully. Based on what he says, I wonder
> if someone can name the algorithms he may be using and provide
> references to relevant books and/or papers.
>
> Quoting the article:
>
> He quickly became intrigued by DSP. [...]
>
> Markets also move in waves similar to, but not identical to, the waves
> of sound, leading Molinero to ask, "If markets have a non-random
> component, can we extract the random one and keep the non-random one?"
>
> This mean transforming existing algorithms, for a variety of reasons:
> first, markets aren't sound waves, and second, DSP is designed to
> rebuild something that previously existed (sound), while traders are
> trying to project a bit into the future.
>
> [...]
>
> "We decompose the price into linear (trending) and non-linear
> (cyclical) components, and then the algorithms identify which are
> signals," Molinear explains., adding that what's left after filtering
> the noise is probably the market's real pulse. Models also generate a
> probability of success on each trade. "For example, if you can explain
> a signal with very few components, it's probably a strong signal,
> which we can use to project the next directional move."
>
> This, he says, enables them to avoid getting whipsawed in choppy
> markets and also to get into trending markets earlier.
>
> "The models adapt in frequency based on the speed at which the market
> is going," he explains. "By decomposing the market, we can look at the
> underlying components. If these components change directions
> frequently, our models will trade short-term, if not our models will
> trade medium-term."

You have to be careful about which model you use for the Market. A
couple of Nobel Laureates were on the board in a company called Long
Term Capital Management. They devised some really neat analyses and
allowed them to reap 40% annual returns and this worked great until
the collapse of the Russian market. Their theory was good except it
didn't account for the herd effect in a bear market. They lost their
asses in 1998! And other stock firms ended up bailing them out.

We had a guy here who posted alot about fancy methods of analysis, but
the market crash in the early 2000s silenced him as well.

Often one finds a model that fits the market over a short period of
time, but noone seems to be able to fit the market over long periods
of time. It is hardly a stationary process.

Clay

On May 6, 3:52�pm, c...@claysturner.com wrote:

> Often one finds a model that fits the market over a short period of
> time, but noone seems to be able to fit the market over long periods
> of time. It is hardly a stationary process.

True, but the person being profiled uses methods that deal with the
nonstationarity, as described in the last three paragraphs of the
excerpt.

On 6 Mai, 21:30, Beliavsky <beliav...@aol.com> wrote:
> The May 2009 issue of Futures (as in financial derivatives) magazine
> has a an interview with money manager Rafael Molinero, who has used
> DSP algorithms to trade successfully. Based on what he says, I wonder
> if someone can name the algorithms he may be using and provide
> references to relevant books and/or papers.

I have noted that a couple of books on Kalman filters
have used econometrics as the application for the theory.
This book is quite good on the Kalman filter part:

http://www.amazon.com/Analysis-Methods-Oxford-Statistical-Science/dp/0198523548/ref=sr_1_1?ie=UTF8&s=books&qid=1241640404&sr=8-1

Don't know how well eceonomy part works, though. I have
noted that the examples are based on reviewing historic
data to see effects of manipulations, like what effect the
requirement to wear seat belts in cars had on fatality
rates in road accidents. The example in the book showed
data from a couple of years before and after the
legislation. I can't remember having seen examples of
predictions.

But then, analyzing data after the fact is trivial
compared to predicting future behaviour.

Rune

On Wed, 6 May 2009 12:52:33 -0700 (PDT), clay@claysturner.com wrote:

>You have to be careful about which model you use for the Market. A
>couple of Nobel Laureates were on the board in a company called Long
>Term Capital Management. They devised some really neat analyses and
>allowed them to reap 40% annual returns and this worked great until
>the collapse of the Russian market. Their theory was good except it
>didn't account for the herd effect in a bear market. They lost their
>asses in 1998! And other stock firms ended up bailing them out.
>
>We had a guy here who posted alot about fancy methods of analysis, but
>the market crash in the early 2000s silenced him as well.
>
>Often one finds a model that fits the market over a short period of
>time, but noone seems to be able to fit the market over long periods
>of time. It is hardly a stationary process.

http://en.wikipedia.org/wiki/Renaissance_Technologies

-- 
Torbj�rn Svensson Diaz

On May 6, 4:42&#4294967295;pm, Torbj&#4294967295;rn Svensson Diaz <tobit...@gmail.com> wrote:
> On Wed, 6 May 2009 12:52:33 -0700 (PDT), c...@claysturner.com wrote:
> >You have to be careful about which model you use for the Market. A
> >couple of Nobel Laureates were on the board in a company called Long
> >Term Capital Management. They devised some really neat analyses and
> >allowed them to reap 40% annual returns and this worked great until
> >the collapse of the Russian market. Their theory was good except it
> >didn't account for the herd effect in a bear market. They lost their
> >asses in 1998! And other stock firms ended up bailing them out.
>
> >We had a guy here who posted alot about fancy methods of analysis, but
> >the market crash in the early 2000s silenced him as well.
>
> >Often one finds a model that fits the market over a short period of
> >time, but noone seems to be able to fit the market over long periods
> >of time. It is hardly a stationary process.
>
> http://en.wikipedia.org/wiki/Renaissance_Technologies
>
> --
> Torbj&#4294967295;rn Svensson Diaz


Thanks for the link. RenTech's Medallion Fund is only open to former
and current Renaissance employees. Their other funds are down.
Madoff's fund looked good too until!! I'm not saying the RenTech is a
Ponzi scheme, but it needs to be properly scrutinized and they keep
quite a bit of stuff quiet. They even petitioned the SEC about not
revealing their short positions. It will be interesting to see how
they fare with the current economic situation.

Clay

Beliavsky wrote:
> On May 6, 3:52 pm, c...@claysturner.com wrote:
> 
>> Often one finds a model that fits the market over a short period of
>> time, but noone seems to be able to fit the market over long periods
>> of time. It is hardly a stationary process.
> 
> True, but the person being profiled uses methods that deal with the
> nonstationarity, as described in the last three paragraphs of the
> excerpt.

It will be a long time before technology is able to predict as much as a 
month into the future of the stock market. Wishful thinking often buoys 
what amounts to get-rich-quick schemes.

Jerry
-- 
Engineering is the art of making what you want from things you can get.
�����������������������������������������������������������������������

Jerry Avins wrote:
> Beliavsky wrote:
>> On May 6, 3:52 pm, c...@claysturner.com wrote:
>>
>>> Often one finds a model that fits the market over a short period of
>>> time, but noone seems to be able to fit the market over long periods
>>> of time. It is hardly a stationary process.
>>
>> True, but the person being profiled uses methods that deal with the
>> nonstationarity, as described in the last three paragraphs of the
>> excerpt.
> 
> It will be a long time before technology is able to predict as much as a 
> month into the future of the stock market.

This may *never* occur. I don't think the stock market is
anything like analytic. Even applying heuristics requires
the applier to make assumptions.

Look at the CDS market failure - an assumption in the
model everybody used led to the downfall. There's no
reason to think that the set of such assumptions is
even countable, much less manageable. Once you
conquer one, it's hard to think that a reasonable
person can be confident that was the *last* one.

> Wishful thinking often buoys 
> what amounts to get-rich-quick schemes.
> 
> Jerry

--
Les Cargill

On Wed, 6 May 2009 12:30:26 -0700 (PDT), Beliavsky <beliavsky@aol.com> wrote:
> The May 2009 issue of Futures (as in financial derivatives) magazine
> has a an interview with money manager Rafael Molinero, who has used
> DSP algorithms to trade successfully. Based on what he says, I wonder
> if someone can name the algorithms he may be using and provide
> references to relevant books and/or papers.

I read this twice, and I'm still puzzled by the reference to "DSP".

The last time I read an article about someone's attempt to model the
ups and downs of the stock market, it was based on "time-series
analysis". True, a lot of DSP algorithms _are_ functions of time,
but by using the term "DSP" is Mr. Molinero saying anything more than
that?
 
> Quoting the article:
> 
> He quickly became intrigued by DSP. [...]

It's a nice buzzword, and its use makes him newsworthy.

> Markets also move in waves similar to, but not identical to, the waves
> of sound, leading Molinero to ask, "If markets have a non-random
> component, can we extract the random one and keep the non-random one?"

"If my signal has noise, can I extract the signal?"
 
> This mean transforming existing algorithms, for a variety of reasons:
> first, markets aren't sound waves, ...

(although I hear of frequent yelling and screaming among the individual
 traders <grin!>)

> ... and second, DSP is designed to
> rebuild something that previously existed (sound), while traders are
> trying to project a bit into the future.

I'm not a DSP expert -- I don't even play one on TV -- but something
about this description of DSP seems... limited.

I freely admit that I haven't read the article, and Mr. Molinero may,
indeed, have found a way of modelling the rise and fall of stock
prices. Based on the article, though, I'm a little cautious about
the quality of his tailoring.

Am I just missing something?


Frank McKenney
--
  Inborn desires are a nuisance to those with utopian and totalitarian
  visions, which often amount to the same thing. What stands in the
  way of most utopias is not pestilence and drought but human behavior.
                             -- Steven Pinker, "The Blank Slate"
--
Frank McKenney, McKenney Associates
Richmond, Virginia / (804) 320-4887
Munged E-mail: frank uscore mckenney ayut mined spring dawt cahm (y'all)

On 6 Mai, 22:42, Torbj&#4294967295;rn Svensson Diaz <tobit...@gmail.com> wrote:
> On Wed, 6 May 2009 12:52:33 -0700 (PDT), c...@claysturner.com wrote:
> >You have to be careful about which model you use for the Market. A
> >couple of Nobel Laureates were on the board in a company called Long
> >Term Capital Management. They devised some really neat analyses and
> >allowed them to reap 40% annual returns and this worked great until
> >the collapse of the Russian market. Their theory was good except it
> >didn't account for the herd effect in a bear market. They lost their
> >asses in 1998! And other stock firms ended up bailing them out.
>
> >We had a guy here who posted alot about fancy methods of analysis, but
> >the market crash in the early 2000s silenced him as well.
>
> >Often one finds a model that fits the market over a short period of
> >time, but noone seems to be able to fit the market over long periods
> >of time. It is hardly a stationary process.
>
> http://en.wikipedia.org/wiki/Renaissance_Technologies

http://en.wikipedia.org/wiki/Madoff