a typical scenario is the expectation of an extreme move that elevates short maturities above longer ones. You want to open a calendar with strikes in the direction of the move, e.g stock is at 100$, you expect a move to 80$ so you buy the 80$ calendar. Typically skew is also high to the side of the expected move. When the move comes, your front month rolls down skew and IV deflates while the backmonth also loses IV but is slower. You want to ratio your calendars, usually you're selling less front month options than back month options in a way so that your theta is zero but you get a lot of gamma. The sweet spot is usually when your weighted vega is zero, your theta is zero and you get a lot of gamma basically for free...and then the market nukes Long calendars are really tricky
In 2008 there were some occasions where you got the 1m6m calendar for free...of course you bought as much as you could EDIT: I just realized that in the screenshot I violated the term structure no arb criterium. Front month should never be more premium than backmonth....I hope you get the point anyways^^
Hah - you can say that again! And yes, that definitely helps. I didn't even think about ratioing them (should have, given your earlier fly example...) - that's super useful. I'm still having a tough time figuring out how to neutralize specific greeks in different types of trades, but I'm slowly getting there. Thanks again!
Thank you MrMuppet for this concrete example. Maybe I got this wrong, but I thought when the move comes in, the IV for both front and back months should increase, not reduce? I'm thinking in the case of an index and it drops suddenly. I agree the skew should flatten. Also, I'm curious how you would manage the vega risk in this example. I understand you construct it to be vega neutral at inception. But there is still the risk that front and back month IVs do not change according to the ratio (sqrt time or other), and it will make the calendar not vega neutral any more.
Like a proper troll, I was about to call you a dumb muppet without reading your whole post, including your edit.
@jamesbp sorry to be dense, but would you mind clarifying this formula for net normalized vega? Say I have: long option 1 with 80 DTE and initial vega of 0.86 short option 2 with 17 DTE and initial vega of 0.40 Is net normalized vega equal to: 0.86 * sqrt(17/80) - 0.40 * sqrt(17/17) = 0.86 * 0.46 - 0.40 * 1 = 0.396 + 0.40 ~ 0 In general terms, inside the square root, am I dividing the minimum DTE of all options in the spread (17 in this case) by the DTE of a specific option (e.g., 80 DTE in the case of option 1)? It looks like the equation that you wrote might be trying to do something entirely different than what I was thinking...
Thanks. I originally had this formula in my Franken-sheet, but the normalized vega values so often seemed to be zero, that I thought I had an error. If I want to understand the sensitivity of a complex option position to volatility, is this normalized vega "the answer", or is it more of a "it depends"?
standard dev isn’t additive. Variance (which is standard dev^2) is. you can view an options implied variance as the sum of all the expected daily realized variances.
When you use option combinations for ‘directional’ trading, isn’t it ‘better’ to shift the strike you are buying more atm? When your view is correct you lose less delta on your long side. Suppose you have a bearish view on spy, instead of: create a diagonal in the direction you are anticipating? So for example: The diagonal pays more (absolute) profit per combo IF it drops since you lose less delta on your longs calls. This is of coure offset to some extend having less delta when it rises (you are wrong).