What dimension do you model your volatility curve?

Discussion in 'Options' started by 76132, Jan 3, 2019.

  1. 76132

    76132

    When trading, I don't know if you guys also try to plot out the market implied volatility curve using a trading software. There are different types of curves that can be used to try to fit the market implied volatility, but most models have different points on the curve that you can move around to fit your model. For example, the points furtherest away from the ATM option are usually there so you can fit the wings of the vol curve.

    What dimension are you using to locate these points? Obviously, the Y axis is IV, but what about X-axis? It wouldn't make sense to set points on the model based on strike as unless you trade a product that doesn't move, you'll have to adjust the strikes constantly to keep your vol curve in the same shape.

    Other common ways to dictate the location of the points on the curve include some measure of moneyness, standard deviation, or delta.

    I'm reading a doc and I'm getting the sense that delta may be the way to go. Standard deviation is what I am currently kinda using but I'm starting to realize that has issues if the product you trade has an IV that changes a lot.

    What do you guys use to dictate where the points on your vol curve lie?
     
  2. Peter8519

    Peter8519

    I take the long view i.e. the daily closing price absolute % change over the previous day. Categorize the data into some % change interval. Plot the data into bar charts. Volatile period will have a fat tail.
     
  3. Fx: usually IV(OTM put delta, OTM call delta)
    Equity: usually IV(OTM put strike, OTM call strike)

    The real question before plotting IV is whether the forward is right.
    Fx forward curves are straightforward.
    Equity forward curves are trivial without any dividend. In case of dividend, a discrete dividend model matches the reality. For American options, the volatility curve should be such that ATM call IV = ATM put IV using the put call parity as additional condition to get the implied discrete dividend.
     
  4. newwurldmn

    newwurldmn

    I looked at fixed strike and floating strike.

    I felt that standard dev and delta were odd because they required an input from the model you were parameterizing and atm vol.

    There is no right answer and it depends on what you are looking to isolate.