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

Re: AI and Investing

Expand Messages
  • carlosuv7403
    Hi Jerry, I tried some years ago, I did a software for modelling and try to guess future market values of shares, commodities, etc. It didn t work, I
    Message 1 of 5 , Oct 20, 2008
    • 0 Attachment
      Hi Jerry,

      I tried some years ago, I did a software for modelling and try
      to "guess" future market values of shares, commodities, etc.

      It didn't work, I abandoned, but Im still convinced that such a system
      could be good for investment.

      The problems that I found are:
      - a result depends on the amount of data that you have, the more years
      of stock-commodities-currencies you have, the better. The information
      is available on internet (to pay for it)
      - Ia models need time to develop a good investment behaviour, and, with
      the great amount of data, it will take more than a few hours to get
      some convincing result
      - markets are more unpredictable that I thought. I read a japanese
      analist who said something like markets are not moved by intelligence,
      but are moved by ignorance and fear. I made me think a lot !
      - keep the system simple, that was I think my big mistake, I included
      more that 1000 prices of stocks, transactions, commodities prices,
      bonds, etc. etc., with so much data, and a little horizon of data, IA
      would get results that appear to be good, but in fact are a disaster.

      I am convinced that it can work (with some considerable error margin)
      because I remember from my finance courses there was people who made
      lot of money analysing graphs, yes market graphs curves, they
      classified different kind of curves and they could predict some
      behaviour. A software can analize a graph.

      Carlos



      --- In artificialintelligencegroup@yahoogroups.com, "G. Fredrick
      Nowatzke" <nowatzke60130@...> wrote:
      >
      > I've been researching short term investment stratigies using a number
      > of soft computing approaches. The resutls appear to be very positive.
      > Is anyone else looking into this type of application?
      >
      > Jerry
      >
    • james.bliss@comcast.net
      There are plenty of these types of investment programs which exist in proprietary environments. They are at the various investment houses. They are good to
      Message 2 of 5 , Oct 20, 2008
      • 0 Attachment
        There are plenty of these types of investment programs which exist in proprietary environments. They are at the various investment houses. They are good to an extent but are completely incapable of handling the events which we just went through across the globe since that type of event last occurred in 1929 with some minor events similar to it in the 60s and 80s.


        -------------- Original message ----------------------
        From: "carlosuv7403" <carlosuv7403@...>
        > Hi Jerry,
        >
        > I tried some years ago, I did a software for modelling and try
        > to "guess" future market values of shares, commodities, etc.
        >
        > It didn't work, I abandoned, but Im still convinced that such a system
        > could be good for investment.
        >
        > The problems that I found are:
        > - a result depends on the amount of data that you have, the more years
        > of stock-commodities-currencies you have, the better. The information
        > is available on internet (to pay for it)
        > - Ia models need time to develop a good investment behaviour, and, with
        > the great amount of data, it will take more than a few hours to get
        > some convincing result
        > - markets are more unpredictable that I thought. I read a japanese
        > analist who said something like markets are not moved by intelligence,
        > but are moved by ignorance and fear. I made me think a lot !
        > - keep the system simple, that was I think my big mistake, I included
        > more that 1000 prices of stocks, transactions, commodities prices,
        > bonds, etc. etc., with so much data, and a little horizon of data, IA
        > would get results that appear to be good, but in fact are a disaster.
        >
        > I am convinced that it can work (with some considerable error margin)
        > because I remember from my finance courses there was people who made
        > lot of money analysing graphs, yes market graphs curves, they
        > classified different kind of curves and they could predict some
        > behaviour. A software can analize a graph.
        >
        > Carlos
        >
        >
        >
        > --- In artificialintelligencegroup@yahoogroups.com, "G. Fredrick
        > Nowatzke" <nowatzke60130@...> wrote:
        > >
        > > I've been researching short term investment stratigies using a number
        > > of soft computing approaches. The resutls appear to be very positive.
        > > Is anyone else looking into this type of application?
        > >
        > > Jerry
        > >
        >
        >
        >




        [Non-text portions of this message have been removed]
      • Unmitigated Gall
        ... in proprietary environments. They are at the various investment houses. They are good to an extent but are completely incapable of handling the events
        Message 3 of 5 , Nov 15, 2008
        • 0 Attachment
          --- In artificialintelligencegroup@yahoogroups.com, james.bliss@...
          wrote:
          >
          > There are plenty of these types of investment programs which exist
          in proprietary environments. They are at the various investment
          houses. They are good to an extent but are completely incapable of
          handling the events which we just went through across the globe since
          that type of event last occurred in 1929 with some minor events
          similar to it in the 60s and 80s.

          Yes. I took a class in securities and market analysis once and was a
          bit surprised by technologists. The cup and saucer. It would come
          close to vodoo economics, but there are trends you can study.

          For example you have interest paying treasuring bills. Something that
          guarantees a certain return on a $1000 bill. Prime rates effect their
          value. So does inflation.

          Interest bearing bonds performance effect non interest bearing. So by
          entering numbers you should be able to make some predictions about
          categories of stock in general.

          But considering the value of this type of thing - billions, if not
          trillions, it is like the gambling industry. Just a slight advantage
          is huge. Massive, eleboarate casinos, free drinks, rooms, cheap air
          fair, paif for with only a slight statistical advantage at the
          tables.

          The issue is, computers can't read the news. If I read abotu Enron
          going broke on Yahoo 5 minutes after it happend, I would know before
          a machine to short sell Enron.

          It would be interesting to see some of these systems and what they
          are storing or doing. With enough stored historically, you could look
          at how this does against that. Company against company, industry
          against industry, investment type against investment type and make
          some predictions.


          >
          >
          > -------------- Original message ----------------------
          > From: "carlosuv7403" <carlosuv7403@...>
          > > Hi Jerry,
          > >
          > > I tried some years ago, I did a software for modelling and try
          > > to "guess" future market values of shares, commodities, etc.
          > >
          > > It didn't work, I abandoned, but Im still convinced that such a
          system
          > > could be good for investment.
          > >
          > > The problems that I found are:
          > > - a result depends on the amount of data that you have, the more
          years
          > > of stock-commodities-currencies you have, the better. The
          information
          > > is available on internet (to pay for it)
          > > - Ia models need time to develop a good investment behaviour,
          and, with
          > > the great amount of data, it will take more than a few hours to
          get
          > > some convincing result
          > > - markets are more unpredictable that I thought. I read a
          japanese
          > > analist who said something like markets are not moved by
          intelligence,
          > > but are moved by ignorance and fear. I made me think a lot !
          > > - keep the system simple, that was I think my big mistake, I
          included
          > > more that 1000 prices of stocks, transactions, commodities
          prices,
          > > bonds, etc. etc., with so much data, and a little horizon of
          data, IA
          > > would get results that appear to be good, but in fact are a
          disaster.
          > >
          > > I am convinced that it can work (with some considerable error
          margin)
          > > because I remember from my finance courses there was people who
          made
          > > lot of money analysing graphs, yes market graphs curves, they
          > > classified different kind of curves and they could predict some
          > > behaviour. A software can analize a graph.
          > >
          > > Carlos
          > >
          > >
          > >
          > > --- In artificialintelligencegroup@yahoogroups.com, "G. Fredrick
          > > Nowatzke" <nowatzke60130@> wrote:
          > > >
          > > > I've been researching short term investment stratigies using a
          number
          > > > of soft computing approaches. The resutls appear to be very
          positive.
          > > > Is anyone else looking into this type of application?
          > > >
          > > > Jerry
          > > >
          > >
          > >
          > >
          >
          >
          >
          >
          > [Non-text portions of this message have been removed]
          >
        • G. Fredrick Nowatzke
          It s retively useless to have AI that deals with a 1929 event. That is a Black Swan event. You d be waiting decades for such an event. Much more profitable to
          Message 4 of 5 , Nov 16, 2008
          • 0 Attachment
            It's retively useless to have AI that deals with a 1929 event. That
            is a Black Swan event. You'd be waiting decades for such an event.
            Much more profitable to predict what a financial instrument will do
            in the next few days or the next few hours. That gives you a bunch of
            trades a week instead of a few trades every 100 years.

            --- In artificialintelligencegroup@yahoogroups.com, "Unmitigated
            Gall" <Spammastergrand@...> wrote:
            >
            > --- In artificialintelligencegroup@yahoogroups.com, james.bliss@
            > wrote:
            > >
            > > There are plenty of these types of investment programs which
            exist
            > in proprietary environments. They are at the various investment
            > houses. They are good to an extent but are completely incapable of
            > handling the events which we just went through across the globe
            since
            > that type of event last occurred in 1929 with some minor events
            > similar to it in the 60s and 80s.
            >
            > Yes. I took a class in securities and market analysis once and was
            a
            > bit surprised by technologists. The cup and saucer. It would come
            > close to vodoo economics, but there are trends you can study.
            >
            > For example you have interest paying treasuring bills. Something
            that
            > guarantees a certain return on a $1000 bill. Prime rates effect
            their
            > value. So does inflation.
            >
            > Interest bearing bonds performance effect non interest bearing. So
            by
            > entering numbers you should be able to make some predictions about
            > categories of stock in general.
            >
            > But considering the value of this type of thing - billions, if not
            > trillions, it is like the gambling industry. Just a slight
            advantage
            > is huge. Massive, eleboarate casinos, free drinks, rooms, cheap air
            > fair, paif for with only a slight statistical advantage at the
            > tables.
            >
            > The issue is, computers can't read the news. If I read abotu Enron
            > going broke on Yahoo 5 minutes after it happend, I would know
            before
            > a machine to short sell Enron.
            >
            > It would be interesting to see some of these systems and what they
            > are storing or doing. With enough stored historically, you could
            look
            > at how this does against that. Company against company, industry
            > against industry, investment type against investment type and make
            > some predictions.
            >
            >
            > >
            > >
            > > -------------- Original message ----------------------
            > > From: "carlosuv7403" <carlosuv7403@>
            > > > Hi Jerry,
            > > >
            > > > I tried some years ago, I did a software for modelling and try
            > > > to "guess" future market values of shares, commodities, etc.
            > > >
            > > > It didn't work, I abandoned, but Im still convinced that such a
            > system
            > > > could be good for investment.
            > > >
            > > > The problems that I found are:
            > > > - a result depends on the amount of data that you have, the
            more
            > years
            > > > of stock-commodities-currencies you have, the better. The
            > information
            > > > is available on internet (to pay for it)
            > > > - Ia models need time to develop a good investment behaviour,
            > and, with
            > > > the great amount of data, it will take more than a few hours to
            > get
            > > > some convincing result
            > > > - markets are more unpredictable that I thought. I read a
            > japanese
            > > > analist who said something like markets are not moved by
            > intelligence,
            > > > but are moved by ignorance and fear. I made me think a lot !
            > > > - keep the system simple, that was I think my big mistake, I
            > included
            > > > more that 1000 prices of stocks, transactions, commodities
            > prices,
            > > > bonds, etc. etc., with so much data, and a little horizon of
            > data, IA
            > > > would get results that appear to be good, but in fact are a
            > disaster.
            > > >
            > > > I am convinced that it can work (with some considerable error
            > margin)
            > > > because I remember from my finance courses there was people who
            > made
            > > > lot of money analysing graphs, yes market graphs curves, they
            > > > classified different kind of curves and they could predict some
            > > > behaviour. A software can analize a graph.
            > > >
            > > > Carlos
            > > >
            > > >
            > > >
            > > > --- In artificialintelligencegroup@yahoogroups.com, "G.
            Fredrick
            > > > Nowatzke" <nowatzke60130@> wrote:
            > > > >
            > > > > I've been researching short term investment stratigies using
            a
            > number
            > > > > of soft computing approaches. The resutls appear to be very
            > positive.
            > > > > Is anyone else looking into this type of application?
            > > > >
            > > > > Jerry
            > > > >
            > > >
            > > >
            > > >
            > >
            > >
            > >
            > >
            > > [Non-text portions of this message have been removed]
            > >
            >
          Your message has been successfully submitted and would be delivered to recipients shortly.