Volatility estimators

Discussion in 'Risk Management' started by bjelinski, Jan 24, 2011.

  1. bjelinski


    Hi all,

    in the attachement you can find a file , about volatility estimators. They are writen in mathematical terms, but as with any other math/statistics formula this should be explained in more simple way. Does anybody know where to find a resource that could be helpful in explaining this.

    I would say the 1st one (Close to close) is just plain standard deviation of returns, correct me if I am wrong?

    Thanks for any input
    • doc.pdf
      File size:
      116.9 KB
  2. I would suggest Google.

    The first one, as stated in the document, is the annualised standard deviation of log returns, squared, aka variance, if memory serves.
  3. Close-to-Close Estimator
    [*] It has well-understood sampling properties.
    [*] It is easy to correct bias.
    [*] It is easy to convert to a form involving typical daily moves.
    [*] It is a very inefficient use of data and converges very slowly.

    Parkinson Estimator
    [*] Using daily range seems sensible and provides completely separate information from using time-based sampling such as closing prices.
    [*] It is really only appropriate for measuring the volatility of a GBM process. In particular it cannot handle trends and jumps.
    [*] It systematically underestimates volatility.

    Garman-Klass Estimator
    [*] It is up to eight times more efficient than close-to-close estimator.
    [*] It makes the best use of the commonly available price information.
    [*] It is even more biased than the Parkinson estimator.

    Rogers-Satchell Estimator
    [*] It allows for the presence of trends.
    [*] It still cannot deal with jumps.