Ever a good time to sell option premium on S&P 500?

Discussion in 'Options' started by short&naked, Dec 12, 2018.

  1. ktm

    ktm

    I would suggest you consider butterflies. :D
     
    #11     Dec 12, 2018
  2. sle

    sle

    I just wish people understand the product they are trading better before saying things like that. In the modern liquid vol market, an options model serves as a interpolation/comparison tool more than anything else. Black-Scholes framework, with some minor tweaks fits that purpose well enough. It's also a reasonable basic model for risk management, unless you are dealing with exotic derivatives of some sort.

    Back to the topic, there is a question of "when is a good time to sell options?" and then there is a question of "how do I prevent a blowup?". The answer to the first question is that you have one or several ways of forecasting that realized volatility would underperform the implied one. The second question, however, should be reformulated as "how much of your expected gains are you willing to spend to protect yourself from a disaster?" and "what exactly is a disaster scenario?".
     
    #12     Dec 12, 2018
    yc47ib, tommcginnis and MACD like this.
  3. JSOP

    JSOP

    Well if you sell options like your username @short&naked, then you will blow up sooner or later; it would be just a matter of time. No. 2 above would help you a bit but only if the expiration date of the long put is farther away than your short leg.
     
    #13     Dec 12, 2018
    bikgup and tommcginnis like this.
  4. Overnight

    Overnight

    Hehe, a bit of the leg.

    Sorry, sorry, feeling sleepy...

     
    #14     Dec 12, 2018
  5. traider

    traider

    Any hints on how to go about creating a model that can forecast realized volatility?
     
    #15     Dec 13, 2018
  6. Agreed. However the point remains that the Black Scholes framework does not allow you to predict the future in any sense. The framework may indeed tell you that given assumptions in the market today regarding volatility the option is priced more or less "correctly".

    What is does not tell you is what those assumptions will be tomorrow. The VIX makes assumptions about future volatility which change violently and almost instantaneously. Black Scoles is not going to help you predict such change or trade on it except in the very shortest of time frames.
     
    #16     Dec 13, 2018
  7. What are these methods? One poster has suggested using the mean reverting nature of volatility to sell high and buy low. One problem with this approach is you have to wait for extremes or at least levels which, historically at least, look to be high or low. While implied volatility can always go higher however to irrational levels, at least there is a fairly reliable floor.

    Black Scholes is not going to tell you today's implied volatility will look cheap or dear tomorrow or the next day. The charts coupled with the mean reverting nature of volatility might.
     
    Last edited: Dec 13, 2018
    #17     Dec 13, 2018
    MACD likes this.
  8. TheBigShort

    TheBigShort

    Create a regression with a response variable log(RV_T1/IV_T0) add some explanatory variables. That should get you started down the rabbit hole.

    Edit** I read "Implied / Realized volatility". For index and very liquid equities, implied vol is your best bet in forecasting realized vol. Try regressing SPX IV 1 month ~ SPX RV Lagged 1 month and you will see how good the relationship is.
     
    Last edited: Dec 13, 2018
    #18     Dec 13, 2018
    Aged Learner and tommcginnis like this.
  9. Does this equate to "if realised volatility today is lower than forecast volatility at T+30 then sell the cash S&P" and vice versa? In back testing? On the grounds that higher realised volatility usually occurs in falling markets? And in answer to the question of the original poster, if this regression shows higher volatility forecast at day 30 than today then should the poster sell calls on the S&P on grounds that the index is likely to go down?
     
    Last edited: Dec 13, 2018
    #19     Dec 13, 2018
  10. TheBigShort

    TheBigShort

    In all fairness a regression is probably not the best tool to use for this, but it's a place to start. That is not what I was implying in my message. I'm to tired to explain. Interpret it as you wish, time for 3 hours of sleep goodnight.
     
    #20     Dec 13, 2018
    tommcginnis likes this.