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

Use of statistical tools to detect "irrational exuberance"

Expand Messages
  • michael.clemens
    As an equity analyst/portfoli manager, I analyze companies and make recommendations based on my research. However, being a believer in behavioral finance, I am
    Message 1 of 3 , Aug 2, 2006
    View Source
    • 0 Attachment
      As an equity analyst/portfoli manager, I analyze companies and make
      recommendations based on my research.
      However, being a believer in behavioral finance, I am of course
      aware of all possible sorts of traps you might fall into (biases).
      Therefore, I make use of statistical tools such as standard
      deviations in assessing whether or not the current P/E ratio is
      within +/-1 or +/- 2 standard deviations from its historical
      average. Other well known financial ratios such as the "Fed model",
      the implied risk premium etc. may also subject themselves to this
      statistical analysis.
      I have found that it works best on market level rather than on
      company level. Since there are demomstrated long-term mean reversion
      in many financial variables (e.g. Shiller), it is to me of
      lesser importance that the variables I analyze are
      not 100% normally distributed. My simple working rule is that
      anything within +/-1 standard deviation should be
      considered "normal" but anything outside +/-2 standard deviations
      should be considered "extreme" or a case of "irrational exuberance".
      Observations in between are "alert"-triggers, which should prompt an
      extra check on data/assumptions etc.
      Have anyone seen any work on the use of statistical tools as a
      practitioners use of behavioural finance or does anyone have any
      comments or suggestions?
    • Kwok Yeung
      As an economist and investment advisor, I think we are dealing with variables that change on an on-going basis. The statistical data only showed what happened
      Message 2 of 3 , Aug 2, 2006
      View Source
      • 0 Attachment
        As an economist and investment advisor, I think we are dealing with variables that change on an on-going basis. The statistical data only showed what happened in the past. They are useful as references, but their predicting power can be questionable very often. We have to be very careful when employing historical data and used them to assess what is happening now, considering that many variables have been changed over time. When we use the simple standard derivation to measure risk, we are assuming a static state. Considerations should be given to the possible case that there could a mega trend going on to affect the data. We should then consider such perceived change of mega trend is valid. (The mentality that "it is different this time" usually can be found in last stage of market rally. But sometimes, it is really different this time.)
         
        Rather than studying or measuring "irrational exuberance", I tend to feel that it may be more fruitful to study what drives investors to be so brave or so nervous. ( I would drop the word irrational here.)
         
        The newspaper headlines could be good indicators for maximum pessimism and optimism. My question is that: would the measure of "irrational exuberance"  give us further advantage in making better investment decision?


        "michael.clemens" <michael.clemens@...> wrote:
        As an equity analyst/portfoli manager, I analyze companies and make
        recommendations based on my research.
        However, being a believer in behavioral finance, I am of course
        aware of all possible sorts of traps you might fall into (biases).
        Therefore, I make use of statistical tools such as standard
        deviations in assessing whether or not the current P/E ratio is
        within +/-1 or +/- 2 standard deviations from its historical
        average. Other well known financial ratios such as the "Fed model",
        the implied risk premium etc. may also subject themselves to this
        statistical analysis.
        I have found that it works best on market level rather than on
        company level. Since there are demomstrated long-term mean reversion
        in many financial variables (e.g. Shiller), it is to me of
        lesser importance that the variables I analyze are
        not 100% normally distributed. My simple working rule is that
        anything within +/-1 standard deviation should be
        considered "normal" but anything outside +/-2 standard deviations
        should be considered "extreme" or a case of "irrational exuberance".
        Observations in between are "alert"-triggers, which should prompt an
        extra check on data/assumptions etc.
        Have anyone seen any work on the use of statistical tools as a
        practitioners use of behavioural finance or does anyone have any
        comments or suggestions?


      • Dalton Mota
        hi i think you missed the econometric point here. While you speak the truth about taking care in dealing with historical data, if you assume that some
        Message 3 of 3 , Aug 2, 2006
        View Source
        • 0 Attachment
          hi
          i think you missed the econometric point here. While
          you speak the truth about taking care in dealing with
          historical data, if you assume that some stochastic
          process is stationary (by taking the first difference
          or so) you have a valid claim on computing SDs.
          Personally i belive his approach to be to simplistic
          (not a bad think at all) to be reliable on its own.

          Dalton.

          --- Kwok Yeung <kc.yeung@...> wrote:

          > As an economist and investment advisor, I think we
          > are dealing with variables that change on an
          > on-going basis. The statistical data only showed
          > what happened in the past. They are useful as
          > references, but their predicting power can be
          > questionable very often. We have to be very careful
          > when employing historical data and used them to
          > assess what is happening now, considering that many
          > variables have been changed over time. When we use
          > the simple standard derivation to measure risk, we
          > are assuming a static state. Considerations should
          > be given to the possible case that there could a
          > mega trend going on to affect the data. We should
          > then consider such perceived change of mega trend is
          > valid. (The mentality that "it is different this
          > time" usually can be found in last stage of market
          > rally. But sometimes, it is really different this
          > time.)
          >
          > Rather than studying or measuring "irrational
          > exuberance", I tend to feel that it may be more
          > fruitful to study what drives investors to be so
          > brave or so nervous. ( I would drop the word
          > irrational here.)
          >
          > The newspaper headlines could be good indicators
          > for maximum pessimism and optimism. My question is
          > that: would the measure of "irrational exuberance"
          > give us further advantage in making better
          > investment decision?
          >
          >
          > "michael.clemens" <michael.clemens@...> wrote:
          > As an equity analyst/portfoli manager, I
          > analyze companies and make
          > recommendations based on my research.
          > However, being a believer in behavioral finance, I
          > am of course
          > aware of all possible sorts of traps you might fall
          > into (biases).
          > Therefore, I make use of statistical tools such as
          > standard
          > deviations in assessing whether or not the current
          > P/E ratio is
          > within +/-1 or +/- 2 standard deviations from its
          > historical
          > average. Other well known financial ratios such as
          > the "Fed model",
          > the implied risk premium etc. may also subject
          > themselves to this
          > statistical analysis.
          > I have found that it works best on market level
          > rather than on
          > company level. Since there are demomstrated
          > long-term mean reversion
          > in many financial variables (e.g. Shiller), it is to
          > me of
          > lesser importance that the variables I analyze are
          > not 100% normally distributed. My simple working
          > rule is that
          > anything within +/-1 standard deviation should be
          > considered "normal" but anything outside +/-2
          > standard deviations
          > should be considered "extreme" or a case of
          > "irrational exuberance".
          > Observations in between are "alert"-triggers, which
          > should prompt an
          > extra check on data/assumptions etc.
          > Have anyone seen any work on the use of statistical
          > tools as a
          > practitioners use of behavioural finance or does
          > anyone have any
          > comments or suggestions?
          >
          >
          >
          >
          >


          __________________________________________________
          Do You Yahoo!?
          Tired of spam? Yahoo! Mail has the best spam protection around
          http://mail.yahoo.com
        Your message has been successfully submitted and would be delivered to recipients shortly.