Implied Volatility autocorrelation

Discussion in 'Options' started by Rudolf13100, Aug 6, 2008.

  1. Hello,

    I have recently been trying to learn about trading the implied volatility.
    From what I read, IV is supposed to be serial correlated (or autocorrelated). From what I understand, it means that the movement IV will make in the next X days should closely match the mouvement IV has made during the past X days. Do I understand what autocorrelation means?
    This statement according to which IV is autocorrelated seems to me to be crucial as it leads to believe that predicting Implied Volatility is easier that predicting prices.
    Now, supposing that predicting IV is easier than predicting prices, then the serial correlation of IV is higher than the serial correlation of prices.
    I wish to know if some of you have actually calculated the autocorrelation of any IV data? If so, what measure of IV serial correlation have you found? Over what period of time IV serial correlation is highest? Have you compared IV serial correlation to prices serial correlation? What software do you use to do the measurement?
  2. Asking about it all. I don't think anybody but an institution can do what you're asking. There's no dataset I have access to that could do this.
  3. C99


    yes, Microsoft recently revised all their licensing agreements so that only MAJOR institutions have access to excel. Sorry, you're SOL.
  4. I've got several books withinin arms reach of my desk that discuss this at length plus a file folder on my desktop with a multitude of papers on it which I found by googling and then checking each papers' cross-referencing. The VIX is a great index to trade options on for a) an edge and b) a hedge

    "Volatility Trading" by Euan Sinclair
    This even includes a CD ROM with VIX data in an excel file where VIX data is analysed (this one is on my desk, I was playing with spreadsheets today).

    Next try "Options Pricing Models and Volatility" it too comes with a CD ROM with tons of spreadsheet anlyses and in particular one that addresses the VIX.

    Then there's Jeff Augen's new book and tons of pdf papers that you can find with a google. Look for the Credit Suisse paper December 2006. Do a search on ET too and you'll find a wealth of info on this. Then there's the blog "VIX and more".

    The gap bewtween the institutional investor's edge and the individual has narrowed incredibly. It's not hard, well, not really hard.

    Lastly, go to they know this subject really well.

    The Vix Index mean reverts over slightly longer time frames and it clusters (shows positive serial correlation) over short time frames. It spikes much higher to the upside than to the down side.
  5. Provided that you have IV historical data I would think that it should be easy to calculate IV's serial correlation on a specific period.
  6. Diamond Geezer,
    Thanks for the references.
  7. rosy2


    download R. its free and will do everything.

    > library("fImport")
    > query "s=^VIX&a=11&b=1&c=2006&d=0&q=31&f=20008&z=^VIX&x=.csv"
    > VIX= yahooImport(query)
    > pacf(VIX@data$Close, na.action = na.pass)

  8. I've heard some good things about R, but never took the plunge. Looks almost as good as matlab (at least for simple probability/statistics stuff). Was there a steep learning curve? Also, was the PACF function included in the package library, or did you have to write it?
  9. You don't need to be an implied volatility genius, with a Ph.D. in math to trade in volatility. Being a volatility Forest Gump will suffice i.e. knowing when IV is relatively high, or relatively low.

    However, if you get the direction of the underlying wrong, it will be hard to make money with options, even if you get IV right.
    #10     Aug 8, 2008