SI

[SI @SI27T]

@fosWtoz @bennpeifert @karlrohe I don’t know the answer to this but I would love to ever see a chart of a 1m SPX skew fitted with 1) GVV compared to 2) GVV with additional skew Greeks (gamma and vanna)—->Benn :)

[SI @SI27T]

@rudyroggio @TheSpeculator0 Would I use this for calculate skew delta? No, because SSR typically is higher than 1 and smile dynamics should be looked in other dimensions than absolute strike (normalized strikes etc). But this is a different topic now.

[SI @SI27T]

@rudyroggio @TheSpeculator0 But that was my point?. That’s exactly I wanted to point out. My claim is that we observed sticky strike like behavior often for usual daily moves in the index. Just run some regressions to calculate SSR to see this for different indices.

[SI @SI27T]

@TheSpeculator0 @rudyroggio This should be clearly regime dependent. I remember regimes where even 1m Index vols had a skew stickiness ratio of 1 (sticky strike) or slightly above. A 10% move would probably have something like 2. So it’s tough to agree on a decomposition where realized SSR is not stable

[SI @SI27T]

@bennpeifert What about PL explain for a large book of plain options? What is appropriate? What are some pitfalls? Example: A simple Greek based PL explain by using implied vol for each option does not tell us about skew risk. How could we define a skew measure for a whole book of options?

[SI @SI27T]

@bennpeifert @RoniIsraelov @RyanMGavin He wrote so many articles about these concepts before this last great piece. Really, everyone should read these when interested in those strategies.

[SI @SI27T]

@bennpeifert When investing in dividend futures, are you actually calculate your own fair value of the future based on different estimates for the single name divs and compare this to the quoted div future? Or is it based on exit liquidity only? (E.g. in 2020)

[SI @SI27T]

@bennpeifert 1/2*Gamma*S^2/10000

[SI @SI27T]

@bennpeifert Wouldn’t this imply that the relative cost of owning one dollar of Volga downside should have a different cost than owning it for upside calls? Otherwise you would always construct a Volga long and gamma-vanna flat portfolio on the downside (GVV related).

[SI @SI27T]

@bennpeifert What about intraday action. Let’s say you bought variance and intraday (as in march 2020), SPX is already down 10% hours before the close. You are afraid that close-to-close may be lot less than 10%. How to lock in this level and what about hedging impacts on the seller side?

[SI @SI27T]

@bennpeifert Happy Birthday, Benn. Independent of the threads you post, it’s great to have you back here!

[SI @SI27T]

@bennpeifert At some point you listed a few concepts about dimensionality reduction (beta weighted Greeks etc). It would be interesting to learn more about the implementation. Length of historical data for beta calculations, when to update time series or even use implied betas etc

[SI @SI27T]

@bennpeifert @karlrohe Hi Benn, after going through these covariances I wonder what the corresponding Greeks in your linear regression equation would look like. Actually, we need to be able to calculate these skew greeks via BS formulas or some finite differences based on quoted option prices, right?

[SI @SI27T]

@platonovadim @bennpeifert What do you mean by „orthogonal Greeks?“? How do they differ from standard BS Greeks?

[SI @SI27T]

@bennpeifert Is there some finite difference definition for these two Greeks or some equivalent version to look at? So I just wonder about the exact mathematical definition in partial derivative terms

[SI @SI27T]

@bennpeifert Thanks, we are definitely on the same page here. This was very insightful. Maybe one last question: How does skew gamma and skew vanna relate to this concept? Are these separate Greeks in your Taylor expansion formulas and would add additional fit quality?

[SI @SI27T]

@bennpeifert The last time I followed the SPX vol surface, we needed approx 12-15 parameters even for 3m-9m options in order to fit the curve (in line with what vola dynamics guys are claiming).

[SI @SI27T]

@bennpeifert So I guess the costs are only representative for a strike range where GVV is a good fit for the market curve. For deep downside options the fit is not sufficient and we need different set of GVV costs

[SI @SI27T]

@bennpeifert Thanks, Benn. Great thread. Here is my issue with the GVV framework. When calculating the cost of GVV via three options per maturity (25d each and ATM) I get significantly different costs than doing the same exercise for three downside options only (5d put, 10d put, 25d put).

[SI @SI27T]

@bennpeifert Is there also a strike dependence on these costs? So, is the cost of owning one dollar of volga on the downside different than on the upside?