Loading ...
Sorry, an error occurred while loading the content.

Re: Restatement of ( Will a computer program ever be superior to humans in trading?)

Expand Messages
  • jk_stratton@stratton.org
    ... l-Finance@egroups.com ... program ... J: Maybe! It s not over yet! *g* ... of ... the issue ... systemless ... system ... before ... J: Hmm..there s the
    Message 1 of 20 , Jul 5, 2000
    View Source
    • 0 Attachment
      --- In Behavioral-Finance@egroups.com, "pgreenfinch"
      <pgreenfinch@w...> wrote:
      > -----Message d'origine-----
      > De : jk_stratton@s... <jk_stratton@s...>
      > À : Behavioral-Finance@egroups.com <Behaviora
      l-Finance@egroups.com>
      > Date : mercredi 5 juillet 2000 00:36
      > Objet : [Behavioral-Finance] Re: Restatement of ( Will a computer
      program
      > ever be superior to humans in trading?)
      >
      >
      > >Arguing with PG for the hell of it :)
      > >
      > Taken in my own game :)
      >
      J: Maybe! It's not over yet! *g*

      > >First (please see below), the game has gotten to the point where
      > >sucessful systems traders most likely must account for the limits
      of
      > >both humans' cognitiion and emotion and systems' "domain drift" to
      > >compete successfully. A new spin on Keynes's beauty contest?
      > >
      > Having no real answer to that, I will get away just by confusing
      the issue
      > and answerint questions with other questions :)))
      > Here it goes, I will call it the acid test.
      > What will happen to successful system traders when the music stop ?
      > Well, I guess you will answer, rightly, what happen then to the
      systemless
      > traders ?
      > To what I will reply : and to the systemless traders that followed
      system
      > traders excesses and biases (vwith the help of some gurus) ?
      > As only future will tell.. here we gained both some breathing time
      before
      > giving definite answer about the beauty contest winners :)
      >
      J: Hmm..there's the possibility that people who are smart enough to
      develop/use systems successfully can also realise when the game is no
      longer afoot and move on to something else. So my answer/question is
      whether past success and adaptability are in conflict.

      > >Here's what I think will happen: the market adjustment to systems
      > >trading overcrowding will not be smooth, but will be a sharp
      > >discontinuous process in which there will be a shakeout of the mass
      > >of systems traders, with a few players somehow adapting and riding
      it
      > >out and a few leaving the game to maybe come back later when the
      > >storm is over.
      > >
      > Well, supposing the market adjustment be violent, maybe system
      traders with
      > new kind of trading system will come. And new types of systemless
      traders
      > also.

      J: *nods* could be that some systems traders wil be able to
      successfully adapt systems to new conditions or will (successfully)
      go off-system?
    • Ronald Davis, CMT
      Should you really want some approximation of real world answers, I suggest that, rather than abstract argument, you get off your collective duff and do some
      Message 2 of 20 , Jul 5, 2000
      View Source
      • 0 Attachment
        Should you really want some approximation of real world answers, I suggest
        that, rather than abstract argument, you get off your collective duff and do
        some programming.
        Ronald Davis, CMT
        ----- Original Message -----
        From: <jk_stratton@...>
        To: <Behavioral-Finance@egroups.com>
        Sent: Wednesday, July 05, 2000 7:47 AM
        Subject: [Behavioral-Finance] Re: Restatement of ( Will a computer program
        ever be superior to humans in trading?)


        --- In Behavioral-Finance@egroups.com, "pgreenfinch"
        <pgreenfinch@w...> wrote:
        > -----Message d'origine-----
        > De : jk_stratton@s... <jk_stratton@s...>
        > À : Behavioral-Finance@egroups.com <Behaviora
        l-Finance@egroups.com>
        > Date : mercredi 5 juillet 2000 00:36
        > Objet : [Behavioral-Finance] Re: Restatement of ( Will a computer
        program
        > ever be superior to humans in trading?)
        >
        >
        > >Arguing with PG for the hell of it :)
        > >
        > Taken in my own game :)
        >
        J: Maybe! It's not over yet! *g*

        > >First (please see below), the game has gotten to the point where
        > >sucessful systems traders most likely must account for the limits
        of
        > >both humans' cognitiion and emotion and systems' "domain drift" to
        > >compete successfully. A new spin on Keynes's beauty contest?
        > >
        > Having no real answer to that, I will get away just by confusing
        the issue
        > and answerint questions with other questions :)))
        > Here it goes, I will call it the acid test.
        > What will happen to successful system traders when the music stop ?
        > Well, I guess you will answer, rightly, what happen then to the
        systemless
        > traders ?
        > To what I will reply : and to the systemless traders that followed
        system
        > traders excesses and biases (vwith the help of some gurus) ?
        > As only future will tell.. here we gained both some breathing time
        before
        > giving definite answer about the beauty contest winners :)
        >
        J: Hmm..there's the possibility that people who are smart enough to
        develop/use systems successfully can also realise when the game is no
        longer afoot and move on to something else. So my answer/question is
        whether past success and adaptability are in conflict.

        > >Here's what I think will happen: the market adjustment to systems
        > >trading overcrowding will not be smooth, but will be a sharp
        > >discontinuous process in which there will be a shakeout of the mass
        > >of systems traders, with a few players somehow adapting and riding
        it
        > >out and a few leaving the game to maybe come back later when the
        > >storm is over.
        > >
        > Well, supposing the market adjustment be violent, maybe system
        traders with
        > new kind of trading system will come. And new types of systemless
        traders
        > also.

        J: *nods* could be that some systems traders wil be able to
        successfully adapt systems to new conditions or will (successfully)
        go off-system?




        ------------------------------------------------------------------------
        CLICK HERE AND START SAVING ON LONG DISTANCE BILLS TODAY!
        http://click.egroups.com/1/4125/5/_/_/_/962808449/
        ------------------------------------------------------------------------

        To unsubscribe from this group, send an email to:
        Behavioral-Finance-unsubscribe@egroups.com
      • jk_stratton@stratton.org
        Who goes *off* their duff to write software? Jan ... suggest ... duff and do ... program ... to ... stop ? ... no ... mass ... riding
        Message 3 of 20 , Jul 5, 2000
        View Source
        • 0 Attachment
          Who goes *off* their duff to write software?

          Jan

          --- In Behavioral-Finance@egroups.com, "Ronald Davis, CMT"
          <rondavis@s...> wrote:
          > Should you really want some approximation of real world answers, I
          suggest
          > that, rather than abstract argument, you get off your collective
          duff and do
          > some programming.
          > Ronald Davis, CMT
          > ----- Original Message -----
          > From: <jk_stratton@s...>
          > To: <Behavioral-Finance@egroups.com>
          > Sent: Wednesday, July 05, 2000 7:47 AM
          > Subject: [Behavioral-Finance] Re: Restatement of ( Will a computer
          program
          > ever be superior to humans in trading?)
          >
          >
          > --- In Behavioral-Finance@egroups.com, "pgreenfinch"
          > <pgreenfinch@w...> wrote:
          > > -----Message d'origine-----
          > > De : jk_stratton@s... <jk_stratton@s...>
          > > À : Behavioral-Finance@egroups.com <Behaviora
          > l-Finance@egroups.com>
          > > Date : mercredi 5 juillet 2000 00:36
          > > Objet : [Behavioral-Finance] Re: Restatement of ( Will a computer
          > program
          > > ever be superior to humans in trading?)
          > >
          > >
          > > >Arguing with PG for the hell of it :)
          > > >
          > > Taken in my own game :)
          > >
          > J: Maybe! It's not over yet! *g*
          >
          > > >First (please see below), the game has gotten to the point where
          > > >sucessful systems traders most likely must account for the limits
          > of
          > > >both humans' cognitiion and emotion and systems' "domain drift"
          to
          > > >compete successfully. A new spin on Keynes's beauty contest?
          > > >
          > > Having no real answer to that, I will get away just by confusing
          > the issue
          > > and answerint questions with other questions :)))
          > > Here it goes, I will call it the acid test.
          > > What will happen to successful system traders when the music
          stop ?
          > > Well, I guess you will answer, rightly, what happen then to the
          > systemless
          > > traders ?
          > > To what I will reply : and to the systemless traders that followed
          > system
          > > traders excesses and biases (vwith the help of some gurus) ?
          > > As only future will tell.. here we gained both some breathing time
          > before
          > > giving definite answer about the beauty contest winners :)
          > >
          > J: Hmm..there's the possibility that people who are smart enough to
          > develop/use systems successfully can also realise when the game is
          no
          > longer afoot and move on to something else. So my answer/question is
          > whether past success and adaptability are in conflict.
          >
          > > >Here's what I think will happen: the market adjustment to systems
          > > >trading overcrowding will not be smooth, but will be a sharp
          > > >discontinuous process in which there will be a shakeout of the
          mass
          > > >of systems traders, with a few players somehow adapting and
          riding
          > it
          > > >out and a few leaving the game to maybe come back later when the
          > > >storm is over.
          > > >
          > > Well, supposing the market adjustment be violent, maybe system
          > traders with
          > > new kind of trading system will come. And new types of systemless
          > traders
          > > also.
          >
          > J: *nods* could be that some systems traders wil be able to
          > successfully adapt systems to new conditions or will (successfully)
          > go off-system?
          >
          >
          >
          >
          > --------------------------------------------------------------------
          ----
          > CLICK HERE AND START SAVING ON LONG DISTANCE BILLS TODAY!
          > http://click.egroups.com/1/4125/5/_/_/_/962808449/
          > --------------------------------------------------------------------
          ----
          >
          > To unsubscribe from this group, send an email to:
          > Behavioral-Finance-unsubscribe@egroups.com
        • Gord Cruikshank
          Thanks for the motivation Ron, Maybe you d like to visit the site patternrecognizer.com g.
          Message 4 of 20 , Jul 5, 2000
          View Source
          • 0 Attachment
            Thanks for the motivation Ron,
            Maybe you'd like to visit the site patternrecognizer.com
            g.
          • Ronald Davis, CMT
            OK. I did. ??? Are you involved? If so, good going. How does it test? Do you trade it? And so on. Ronald Davis, CMT ... From: Gord Cruikshank
            Message 5 of 20 , Jul 5, 2000
            View Source
            • 0 Attachment
              OK. I did. ???
              Are you involved? If so, good going.
              How does it test? Do you trade it?

              And so on.
              Ronald Davis, CMT
              ----- Original Message -----
              From: "Gord Cruikshank" <gordc@...>
              To: <Behavioral-Finance@egroups.com>
              Sent: Wednesday, July 05, 2000 9:38 AM
              Subject: RE: [Behavioral-Finance] Re: Restatement of ( Will a computer
              program ever be superior to humans in trading?)


              > Thanks for the motivation Ron,
              > Maybe you'd like to visit the site patternrecognizer.com
              > g.
              >
              > ------------------------------------------------------------------------
              > Accurate impartial advice on everything from laptops to table saws.
              > http://click.egroups.com/1/4634/5/_/_/_/962815196/
              > ------------------------------------------------------------------------
              >
              > To unsubscribe from this group, send an email to:
              > Behavioral-Finance-unsubscribe@egroups.com
              >
              >
              >
              >
              >
            • Gord Cruikshank
              Yes, I am. It tests really well. I am trading it, among others. The fundamental theory can be compared with weather forecasting. Imagine two groups of
              Message 6 of 20 , Jul 5, 2000
              View Source
              • 0 Attachment
                Yes, I am.
                It "tests" really well.
                I am trading it, among others.

                The fundamental theory can be compared with weather forecasting. Imagine
                two groups of people Meteorologists and Farmers. The Meteorologist creates
                complex of mathematical models to represent weather patterns. He then uses
                his model to predict what will happen tomorrow. (Note that he is
                occasionally still wrong). On the other hand, the Farmer can go outside,
                look at the sky, see thunderclouds and predict rain. In this example, the
                Farmer is using something akin to market sentiment to make a prediction
                which is correct surprisingly often (maybe more often than the
                Meteorologist's mathematical model!?). All the farmer does is watch for
                patterns that infer a tertiary result. He does not have to understand the
                inner workings of weather to make reasonable and relatively accurate
                "predictions". He simply picks his spots.

                The issue is not to promote this particular site, but to recognise that
                there are many VALID ways to trade. In this example, a nebulous and rather
                difficult task of quantifying market sentiment is in fact captured and then
                utilised in a predictable fashion. The applicable "group investor
                psychology" is interesting. Understanding the specifics of it, is
                unnecessary.

                >OK. I did. ???
                >Are you involved? If so, good going.
                >How does it test? Do you trade it?
                >
                >And so on.
                >Ronald Davis, CMT
                >----- Original Message -----
                >From: "Gord Cruikshank" <gordc@...>
                >To: <Behavioral-Finance@egroups.com>
                >Sent: Wednesday, July 05, 2000 9:38 AM
                >Subject: RE: [Behavioral-Finance] Re: Restatement of ( Will a computer
                >program ever be superior to humans in trading?)
                >
                >
                >> Thanks for the motivation Ron,
                >> Maybe you'd like to visit the site patternrecognizer.com
                >> g.
              • Igor Lampin
                ... From: Gord Cruikshank [mailto:gordc@igs.net] Sent: Thursday, July 06, 2000 3:19 AM To: Behavioral-Finance@egroups.com Subject: [Behavioral-Finance] RE:
                Message 7 of 20 , Jul 6, 2000
                View Source
                • 0 Attachment
                  -----Original Message-----
                  From: Gord Cruikshank [mailto:gordc@...]
                  Sent: Thursday, July 06, 2000 3:19 AM
                  To: Behavioral-Finance@egroups.com
                  Subject: [Behavioral-Finance] RE: [Behavioural-Finance] Re: Restatement
                  of ( Will a computer program ever be superior to humans in trading?)

                  >The fundamental theory can be compared with weather forecasting. Imagine
                  >two groups of people Meteorologists and Farmers. The Meteorologist creates
                  >complex of mathematical models to represent weather patterns. He then uses
                  >his model to predict what will happen tomorrow. (Note that he is
                  >occasionally still wrong). On the other hand, the Farmer can go outside,
                  >look at the sky, see thunderclouds and predict rain. In this example, the
                  >Farmer is using something akin to market sentiment to make a prediction
                  >which is correct surprisingly often (maybe more often than the
                  >Meteorologist's mathematical model!?). All the farmer does is watch for
                  >patterns that infer a tertiary result. He does not have to understand the
                  >inner workings of weather to make reasonable and relatively accurate
                  >"predictions". He simply picks his spots.

                  I like your comparison and would like to add some "biological"
                  considerations.
                  The farmer's prediction is likely to be more accurate than the one from the
                  meteorologist due to natural selection:
                  - the meteorologist who is making a wrong prediction still gets his salary
                  at the end of the month
                  - the farmer who makes the wrong prediction will be soon out of the game
                  The decay rate of the two populations are very different so that the farmer
                  population will converge more faster to the right prediction.

                  Igor
                • Stefan Schulz
                  ... Two experiences in that regard: 1. A few weeks ago I attended a seminar by makers of stat software where one of the speakers was a quant from a local
                  Message 8 of 20 , Jul 6, 2000
                  View Source
                  • 0 Attachment
                    At 21:08 05/07/00 -0700, Larry Wright wrote:
                    >On Wed, 5 Jul 2000, H. Mark Hubey wrote:
                    > > If it can be explained to a human being clearly it can be programmed.
                    > ^^^^^^^^^^^^^^^^^^^^^^
                    >My point is it is very difficult to get a trader to do this so it would
                    >include *all* the things the trader takes into account - especially noting
                    >the 'clearly' part<g>.
                    >Has anyone done this, to include all the many nuances that go beyond the
                    >simple 'system' things?

                    Two experiences in that regard:

                    1. A few weeks ago I attended a seminar by makers of stat software where
                    one of the speakers was a quant from a local merchant bank. He explained
                    how he'd looked over the shoulder of a few interest rate traders for half a
                    year and then used cluster analysis to find what strategies they were
                    using, because they couldn't put it into words. He found that there were
                    six distinct strategies in use, only one of which was (marginally)
                    profitable. I didn't have the heart to tell him that he should have looked
                    at the traders' profitability before deciding to waste time on this
                    exercise ;-)

                    2. A friend of mine trades indices and bonds only, and he is pretty good
                    at it. Over the last year and a half he and I have tried to capture the
                    elements of his trading system and formalize them. (He uses bar charts
                    with a couple of indicators.) However, simply looking at levels of
                    indicators and/or price and trying to infer patterns from them led
                    nowhere. I found that he is taking in an awful lot of visual cues about
                    recent history and trending behaviour. Slopes of graphs, relative
                    distances, support and resistance levels all come into it. A simple graph
                    with bars and two indicators actually provides him with at least 20 pieces
                    of information! On top of that he also keeps abreast of fundamentals and I
                    guess this biases his inclination to take "technical" trades in particular
                    directions.

                    Looking at the sheer volume of information he processes I very much doubt
                    that this can be formalized - a large neural network may be able to take n
                    x 20 inputs, but that still doesn't look at the fundamental picture.

                    Regards,

                    Stefan Schulz
                    Suaviter Limited
                    programming@...
                  • Peter GREENFINCH
                    ... from the ... salary ... game ... the farmer ... I like this biological approach, as it enrich behavior analysis Some time that phenomena can become tricky:
                    Message 9 of 20 , Jul 6, 2000
                    View Source
                    • 0 Attachment
                      --- In Behavioral-Finance@egroups.com, "Igor Lampin" <igor@c...>
                      wrote:
                      >
                      > The farmer's prediction is likely to be more accurate than the one
                      from the
                      > meteorologist due to natural selection:
                      > - the meteorologist who is making a wrong prediction still gets his
                      salary
                      > at the end of the month
                      > - the farmer who makes the wrong prediction will be soon out of the
                      game
                      > The decay rate of the two populations are very different so that
                      the farmer
                      > population will converge more faster to the right prediction.
                      >
                      I like this biological approach, as it enrich behavior analysis
                      Some time that phenomena can become tricky:
                      The fund manager will lose its salary, not if it makes a wrong
                      prediction, but if it made at prediction that is, at the
                      same time wrong and differing from those of the bulk of other
                      fund manager.
                      If it loose money at the same time tnan others, no problem,
                      he has an excuse (Your hounor, it was "unpredictable",
                      you see, as nobody predicted it). But if it loose money and
                      the others have good performance, then he/she is in trouble.
                      Another illustration of mimetic rational behavior.
                      In Rome, do like the Romans do...
                    • Igor Lampin
                      Furthermore, a big part of the information processed by the human machine is probably sub-conscious information. Something like the memory of thousands of
                      Message 10 of 20 , Jul 6, 2000
                      View Source
                      • 0 Attachment
                        Furthermore, a big part of the information processed by the human machine is
                        probably "sub-conscious" information. Something like the memory of thousands
                        of old patterns that the trader has viewed in the past and which intefere
                        strongly and in a complex way with the rational analysis.
                        May be the solution to build a good program is to work with a schizoprenic
                        trader whose "reflexive brain" is always analysing the "active brain". But I
                        do not know if the performances of schizophrenic traders are statistically
                        better than the ones of so-called normal or paranoid traders.

                        Igor

                        -----Original Message-----
                        From: Stefan Schulz [mailto:programming@...]
                        Sent: Thursday, July 06, 2000 3:10 PM
                        To: Behavioral-Finance@egroups.com
                        Subject: Re: [Behavioral-Finance] Re: Restatement of ( Will a computer
                        program ever be superior to humans in trading?)


                        At 21:08 05/07/00 -0700, Larry Wright wrote:
                        >On Wed, 5 Jul 2000, H. Mark Hubey wrote:
                        > > If it can be explained to a human being clearly it can be programmed.
                        > ^^^^^^^^^^^^^^^^^^^^^^
                        >My point is it is very difficult to get a trader to do this so it would
                        >include *all* the things the trader takes into account - especially noting
                        >the 'clearly' part<g>.
                        >Has anyone done this, to include all the many nuances that go beyond the
                        >simple 'system' things?

                        Two experiences in that regard:

                        1. A few weeks ago I attended a seminar by makers of stat software where
                        one of the speakers was a quant from a local merchant bank. He explained
                        how he'd looked over the shoulder of a few interest rate traders for half a
                        year and then used cluster analysis to find what strategies they were
                        using, because they couldn't put it into words. He found that there were
                        six distinct strategies in use, only one of which was (marginally)
                        profitable. I didn't have the heart to tell him that he should have looked
                        at the traders' profitability before deciding to waste time on this
                        exercise ;-)

                        2. A friend of mine trades indices and bonds only, and he is pretty good
                        at it. Over the last year and a half he and I have tried to capture the
                        elements of his trading system and formalize them. (He uses bar charts
                        with a couple of indicators.) However, simply looking at levels of
                        indicators and/or price and trying to infer patterns from them led
                        nowhere. I found that he is taking in an awful lot of visual cues about
                        recent history and trending behaviour. Slopes of graphs, relative
                        distances, support and resistance levels all come into it. A simple graph
                        with bars and two indicators actually provides him with at least 20 pieces
                        of information! On top of that he also keeps abreast of fundamentals and I
                        guess this biases his inclination to take "technical" trades in particular
                        directions.

                        Looking at the sheer volume of information he processes I very much doubt
                        that this can be formalized - a large neural network may be able to take n
                        x 20 inputs, but that still doesn't look at the fundamental picture.

                        Regards,

                        Stefan Schulz
                        Suaviter Limited
                        programming@...
                      • Stefan Schulz
                        List.members, There was a notable spike in the E-mini Nasdaq last week, going down to 3620 when the pit-traded big Nasdaq only went to 3675 or so. Today the
                        Message 11 of 20 , Jul 6, 2000
                        View Source
                        • 0 Attachment
                          List.members,

                          There was a notable spike in the E-mini Nasdaq last week, going down to
                          3620 when the pit-traded "big" Nasdaq only went to 3675 or so. Today the
                          Bund went from 104.7 to 104.31 and back to 104.65 in about five trades, at
                          one stage moving 23(!) basis points between trades within the same(!) second.

                          In both cases the explanation given afterwards was that sell stops had been
                          hit, reinforcing each other due to the speed of the move involved. Those
                          sell stops had been set by *people*, the speed of the response and hence
                          the violence of the moves was provided by machines.

                          Also in both cases the contracts concerned made new lows for the day, and
                          naturally these will influence volatility calcs and other metrics on which
                          future stops will be based - and so on.

                          I wonder whether this type of interaction between peoples' perceptions and
                          the parameters they set on the one hand and computers' execution of trades
                          based on those parameters on the other hand will lead to different and so
                          far unseen patterns of behaviour. Certainly it is a new type of
                          interaction because the large-scale conversion of futures exchanges to
                          electronic trading has only been going on for a few years.

                          Opinions ?

                          Regards,
                          Stefan Schulz
                          Suaviter Limited
                          programming@...
                        • matteoricci_limatuo@yahoo.com
                          ... hehe :-) ... The fundamental picture is a bit difficult. My bet would be that a lot of the 20 items are redundant. Maybe you should get him to theories and
                          Message 12 of 20 , Jul 6, 2000
                          View Source
                          • 0 Attachment
                            >profitable. I didn't have the heart to tell him that he should have looked
                            >at the traders' profitability before deciding to waste time on this
                            >exercise ;-)
                            hehe :-)


                            >distances, support and resistance levels all come into it. A simple graph
                            >with bars and two indicators actually provides him with at least 20 pieces
                            >of information! On top of that he also keeps abreast of fundamentals and I
                            >guess this biases his inclination to take "technical" trades in particular
                            >directions.
                            >
                            >Looking at the sheer volume of information he processes I very much doubt
                            >that this can be formalized - a large neural network may be able to take n
                            >x 20 inputs, but that still doesn't look at the fundamental picture.

                            The fundamental picture is a bit difficult.
                            My bet would be that a lot of the 20 items are redundant. Maybe you should
                            get him to theories and scale the degree of definitiveness step by step, check
                            for contradictions, etc. I betcha he isn't using 20.
                            Think of it as a work in progress. You can see this happening when you study a
                            philosopher who is particular prolific and watch him tighten his thoughts up,
                            throwing away stuff. Philosophy might bore some so you could try to observed
                            Robert Kiyosaki and see how his writing out of his ideas have firmed up
                            over the
                            years.
                          • Stefan Schulz
                            ... Hold on - he isn t using 20 items on a chart as chart items per se. When you look at say, a bar chart of prices, a moving average of the price and two
                            Message 13 of 20 , Jul 6, 2000
                            View Source
                            • 0 Attachment
                              At 23:24 06/07/00 +0800, matteoricci_limatuo@... wrote:
                              >My bet would be that a lot of the 20 items are redundant. Maybe you should
                              >get him to theories and scale the degree of definitiveness step by step, check
                              >for contradictions, etc. I betcha he isn't using 20.

                              Hold on - he isn't using 20 items on a chart as chart items per se. When
                              you look at say, a bar chart of prices, a moving average of the price and
                              two indicators, then you've got *four* items.

                              However, each of these carries several visual cues. A bar chart carries
                              price, range, an estimate of recent volatility, an estimate of speed of
                              price movement, etc. A moving average carries level, slope, distance to
                              price, etc.

                              Enumerating all these distinct types of information is what got me to the
                              figure 20. And yes, I do believe that most of them are taken in and
                              "processed" by him. There's actually very little redundancy and no
                              duplication in the count.

                              Regards,

                              Stefan Schulz
                              Suaviter Limited
                              programming@...
                            • jk_stratton@stratton.org
                              I think the major factor is the evolution of electronic markets, although fast market conditions (explosions?) are not new. I m curious if there have been any
                              Message 14 of 20 , Jul 6, 2000
                              View Source
                              • 0 Attachment
                                I think the major factor is the evolution of electronic markets,
                                although fast market conditions (explosions?) are not new. I'm
                                curious if there have been any studies on the effects of "circut
                                breaker" trading halts - whether they've injected as pause for calm
                                or have just delayed a bad event.

                                Jan

                                --- In Behavioral-Finance@egroups.com, Stefan Schulz
                                <programming@s...> wrote:
                                > List.members,
                                >
                                > There was a notable spike in the E-mini Nasdaq last week, going
                                down to
                                > 3620 when the pit-traded "big" Nasdaq only went to 3675 or so.
                                Today the
                                > Bund went from 104.7 to 104.31 and back to 104.65 in about five
                                trades, at
                                > one stage moving 23(!) basis points between trades within the same
                                (!) second.
                                >
                                > In both cases the explanation given afterwards was that sell stops
                                had been
                                > hit, reinforcing each other due to the speed of the move involved.
                                Those
                                > sell stops had been set by *people*, the speed of the response and
                                hence
                                > the violence of the moves was provided by machines.
                                >
                                > Also in both cases the contracts concerned made new lows for the
                                day, and
                                > naturally these will influence volatility calcs and other metrics
                                on which
                                > future stops will be based - and so on.
                                >
                                > I wonder whether this type of interaction between peoples'
                                perceptions and
                                > the parameters they set on the one hand and computers' execution of
                                trades
                                > based on those parameters on the other hand will lead to different
                                and so
                                > far unseen patterns of behaviour. Certainly it is a new type of
                                > interaction because the large-scale conversion of futures exchanges
                                to
                                > electronic trading has only been going on for a few years.
                                >
                                > Opinions ?
                                >
                                > Regards,
                                > Stefan Schulz
                                > Suaviter Limited
                                > programming@s...
                              • Gord Cruikshank
                                Igor, Your expansion on my comments certainly take the idea into a more abstract realm. Interesting in theory. I personally would stop speculation at the
                                Message 15 of 20 , Jul 6, 2000
                                View Source
                                • 0 Attachment
                                  Igor,
                                  Your expansion on my comments certainly take the idea into a more abstract
                                  realm. Interesting in theory. I personally would stop speculation at the
                                  point where the two methods describe the application of technology to our
                                  "situation". Certainly we could continue to pin additional criteria onto
                                  the analogy until we develop a fractal effect where the entire description
                                  begins to build upon itself yet fails to add any useful information to the
                                  scenario.

                                  Regardless, I agree with your comment in principle. The issue remains
                                  whether a computer can replace a human and what effect this will have on the
                                  marketplace. I'm still waiting...and listening.

                                  gord
                                • Larry Wright
                                  ... How did he know? Must be a good mind reader :-). ... I presume the traders were making money on these no-good techniques . ... My experience exactly!
                                  Message 16 of 20 , Jul 6, 2000
                                  View Source
                                  • 0 Attachment
                                    On Thu, 6 Jul 2000, Stefan Schulz wrote:

                                    > one of the speakers was a quant from a local merchant bank. He explained
                                    > how he'd looked over the shoulder of a few interest rate traders for half a
                                    > year and then used cluster analysis to find what strategies they were
                                    > using, because they couldn't put it into words. He found that there were
                                    > six distinct strategies in use,

                                    How did he know? Must be a good mind reader :-).

                                    > only one of which was (marginally)
                                    > profitable.

                                    I presume the traders were making money on these 'no-good' techniques <g>.

                                    > 2. A friend of mine trades indices and bonds only, and he is pretty good
                                    > at it. Over the last year and a half he and I have tried to capture the
                                    > elements of his trading system and formalize them. (He uses bar charts
                                    > with a couple of indicators.) However, simply looking at levels of
                                    > indicators and/or price and trying to infer patterns from them led
                                    > nowhere. I found that he is taking in an awful lot of visual cues about
                                    > recent history and trending behaviour.

                                    My experience exactly!

                                    > distances, support and resistance levels all come into it. A simple graph
                                    > with bars and two indicators actually provides him with at least 20 pieces
                                    > of information!

                                    I tried a similar enumeration, with several charts, and came up with many
                                    more than 20 - it compounds because of the interactions between charts.

                                    > Looking at the sheer volume of information he processes I very much doubt
                                    > that this can be formalized - a large neural network may be able to take n
                                    > x 20 inputs,

                                    I'd be interested on how one gets the traders processing of, say, a
                                    'rising pennant' chart pattern into a NN. Seems as though we might get
                                    right back to the original problem :-).

                                    Larry
                                  • Stefan Schulz
                                    ... Not a mind reader, only a statistician. He just used analysis techniques as above. ... Erm, no - that s what he rather sheepishly admitted at the end of
                                    Message 17 of 20 , Jul 7, 2000
                                    View Source
                                    • 0 Attachment
                                      At 21:31 06/07/00 -0700, Larry Wright wrote:
                                      >On Thu, 6 Jul 2000, Stefan Schulz wrote:
                                      > > year and then used cluster analysis to find what strategies they were
                                      > > using, because they couldn't put it into words. He found that there were
                                      > > six distinct strategies in use,
                                      >How did he know? Must be a good mind reader :-).

                                      Not a mind reader, only a statistician. He just used analysis techniques
                                      as above.

                                      > > only one of which was (marginally)
                                      > > profitable.
                                      >
                                      >I presume the traders were making money on these 'no-good' techniques <g>.

                                      Erm, no - that's what he rather sheepishly admitted at the end of his
                                      presentation. AIUI, the object of his exercise was to understand the
                                      motivations and reasons for trades in the first place. Profitability was
                                      second.

                                      >I'd be interested on how one gets the traders processing of, say, a
                                      >'rising pennant' chart pattern into a NN. Seems as though we might get
                                      >right back to the original problem :-).

                                      Funnily enough, I have thought about this one. The first issue is to get
                                      the data into an NN. Data representing a bar pattern like a pennant,
                                      however, doesn't fit the vectorized approach of merely taking snapshots of
                                      indicators. You need to do a fuzzy match to a pennant being built in
                                      recent bars, and then add two scalars to your observation vector: Nature
                                      of pennant and fuzzy strength of match. In other words, it takes some data
                                      transformation but seems doable in principle. The real problem is the
                                      number of inputs required.

                                      Regards,

                                      Stefan Schulz
                                      Suaviter Limited
                                      programming@...
                                    • matteoricci_limatuo@yahoo.com
                                      However, each of these carries several visual cues. A bar chart carries ... There maybe 20 or more that he could recognise as he uses. At a particular point
                                      Message 18 of 20 , Jul 7, 2000
                                      View Source
                                      • 0 Attachment
                                        However, each of these carries several visual cues. A bar chart carries
                                        >price, range, an estimate of recent volatility, an estimate of speed of
                                        >price movement, etc. A moving average carries level, slope, distance to
                                        >price, etc.

                                        There maybe 20 or more that he could recognise as he uses.
                                        At a particular point of time, maybe 5 or 6 are used (lesser number than 20)
                                        with the rest redundant except to give some history. This happen, that happen,
                                        but for now what is happening, "this" and "that" may be quite redundant.

                                        Perhaps, a better way of putting it, he's got multiple signals, each
                                        independent
                                        though because he follows them thru time, he sorts of links them up together
                                        and thinks what happen previously is relevant now. But when placed under
                                        the statistical scrutiny, they can be seen to be largely unrelated.
                                        (Multiple signals for exits and entries. No 2 entry and exit signals are
                                        linked
                                        together like some canned testing platforms make it so).

                                        Cognitive impenetrability makes knowledge extraction a lot of fun ...

                                        >Enumerating all these distinct types of information is what got me to the
                                        >figure 20. And yes, I do believe that most of them are taken in and
                                        >"processed" by him. There's actually very little redundancy and no
                                        >duplication in the count.
                                        >
                                        >Regards,
                                        >
                                        >Stefan Schulz
                                        >Suaviter Limited
                                        >programming@...
                                      • matteoricci_limatuo@yahoo.com
                                        ... Blush for that smile on my face when I read about it earlier. Reading this para made me realised that he deserves all our support. Where would we be
                                        Message 19 of 20 , Jul 7, 2000
                                        View Source
                                        • 0 Attachment
                                          At 06:36 PM 7/7/00, Stefan Schulz wrote:
                                          > > > only one of which was (marginally)
                                          > > > profitable.
                                          > >
                                          > >I presume the traders were making money on these 'no-good' techniques <g>.
                                          >
                                          >Erm, no - that's what he rather sheepishly admitted at the end of his
                                          >presentation. AIUI, the object of his exercise was to understand the
                                          >motivations and reasons for trades in the first place. Profitability was
                                          >second.

                                          Blush for that smile on my face when I read about it earlier.
                                          Reading this para made me realised that he deserves all our support.
                                          Where would we be without those individuals who do basic research!!!!
                                          I doubt the ex-Israeli paratroopers who started the thing on framing
                                          research intended to trade stocks.
                                        Your message has been successfully submitted and would be delivered to recipients shortly.