A New Volatility Measure

Discussion in 'Technical Analysis' started by Joe Doaks, Aug 10, 2007.

  1. For nearly eight years I have sought a measure of volatility which was both mathematically rigorous and of practical value in trading. I scoured all the available books, and worked my fingers arthritic searching ET and the web at large, all in vain. My criteria for validity are:

    - at high values the measure must look as scary itself as the perception of volatility does in real time

    - on the one hand, high values must be so unambiguous that they give clear signals to stand aside

    - on the other hand, low values must correlate strongly with reliably tradeable trending.

    Like all truly insightful and revolutionary ideas, the solution came to me last night in a dream, specifically a wet dream (it is not well known that Kekule's dream of the benzene ring as a snake chasing its own tail is a politically correct fiction: it was a dream of Kekule himself chasing tail). It took merely 30 seconds to code it and five minutes to test it, and Voila! Tailatility! I will not reveal de-tails, but the clever trading algorithmicist can readily reproduce my results for himself. As always, my goal is to humble (or humbug) the Hershey crowd, who after 20 years still cannot code up SCT.
     
  2. A private correspondent (who shall remain unnamed to preserve his credibility on ET) has kindly suggested to me that, since I ultimately intend to trademark all my neo-TA, I should give Tailatility a more respectable name He suggested Volataility, and so shall it be. Another (who shall remain unnamed to preserve my own reputation) opined that I should market it for another use than I originally intended, that is, as a sediment indicator. Low readings indicate sediment settling down, high readings stirred up and muddy. I am not yet convinced, however, that conventional volatility reflects sediment, although Volataility might. Best regards to all, and thanks for the absence of comments. It reaffirms my faith in ET's distaste for originality. If the TA isn't at least 30 years old (and its guru dead or at least indistinguishable from it), ET isn't interested.
     
  3. Joe,
    Observing the recent market activity gave me the idea of measuring volatility based on bar height. For example, it's unusual to see a lot of bars on a 5 minute chart that are several points high. If the last several bars are like that, then I'd assume that trading with a 2 point stop might produce a lot of losers.
     
  4. MGJ

    MGJ

    Perhaps you would gain additional insight from drawing a Venn Diagram. Useful sets to represent on the diagram might include
    • The set of all trading ideas
    • The set of all good trading ideas
    • The set of all trading ideas that can be represented in computer code
    • The set of all trading ideas that ARE represented in computer code, today
    • The set of computer coded trading ideas which, when traded in real life with real money, result in extraordinary risk-adjusted profits out of sample (after they are coded and computer tested)
    You could put a dot on this diagram where VolaTailItty (note the two t's) resides, and another dot where SCT resides.
     
  5. I don't think that being a volatility guru is really required to make money. Being a volatility Forrest Gump will suffice. Really what you need to know about volatility is if it is relatively high or relatively low.

    The absolute value of your indicator is not what is important, because when your indicator is high, so will all the traditional ways of calculating volatility. When your indicator drops, so will all the other volatility indicators.
     
  6. 225, if you try that, please let me know the results. I have pustulated many possible definitions of volatility. Those based on bar height trend to discriminate against tends. Aside from that, I think your observation has merit in general. Trading NQ, 4-6 good sequential bars is about all you get (not unlike catting around on a Saturday night). When you get more, remember Seiki Shimizu's Rule of Ten!
     
  7. MGJ, a very asstoot observation. I like it! I might add another restriction: the set of trading ideas that is optimizable. That lets SCT out because it has far too many variables. Not to mention which it is qualitative. Volataility, however, is a simple measure which can be biary thresholded and optimized for a redlight/green light stand aside indication.
     
  8. Pantherman, I think we may have different uses for a volatility measure. I HATE volatility. I want slow tradeable trends. Hence my pecu-liar definition of it. For example, if NQ has a 5 point range in a one minute bar, like yesterday, that ain't my kinda market.
     
  9. There you go.
    Measure the bar range and smooth it.
    If that's increasing, step aside.
    When it 's decreasing, at some point you could enter.
     
  10. Thank you, 225. I will look into that.
     
    #10     Aug 11, 2007