Is Vanna relevant when Gamma scalping?

TskTsk · Jan 30, 2012

As you know, the P/L from a gamma scalp is 0.5*gamma*S^2 (where S is spot movement). However isn't delta affected by other factors such as Vanna, thus isn't the formula useless for predicting P/L per scalp (because it only takes gamma into account)?

tradingjournals · Jan 30, 2012

Quote from TskTsk:

As you know, the P/L from a gamma scalp is 0.5*gamma*S^2 (where S is spot movement).

Link to the proof?

PS: why a different handle?

newwurldmn · Jan 30, 2012

It's small unless you are running a very levered skew book.

TskTsk · Jan 30, 2012

Quote from newwurldmn:

It's small unless you are running a very levered skew book.

Alright thanks

Quote from tradingjournals:

Link to the proof?

PS: why a different handle?

Dont have any proof, it's a formula I picked up on the NP forums. I've seen it used elsewhere as well so I just assume its correct.

jsp326 · Jan 30, 2012

Only when she's drunkenly flipping letters on a game show.

newwurldmn · Jan 30, 2012

Quote from TskTsk:

Alright thanks

Dont have any proof, it's a formula I picked up on the NP forums. I've seen it used elsewhere as well so I just assume its correct.

Ignore TradingJournals. He knows little of what he speaks.
What is the NP forum?

sle · Jan 30, 2012

Quote from newwurldmn:

It's small unless you are running a very levered skew book.

The delta component would be small. The mtm component dVega/dSpot can put you in sufficient amount of pain, but I assume you are calculating as if you are holding it to expiry.

Martinghoul · Jan 30, 2012

Quote from newwurldmn:
Ignore TradingJournals. He knows little of what he speaks.
What is the NP forum?

NucPhy (Nuclear Phynance)...

And yes, the formula is a correct quick 'n dirty one, but, like all such formulae, it's an approximation. It's obtained simply from the Taylor series expansion, by discarding higher-order terms, because they're assumed to be less significant.

newwurldmn · Jan 30, 2012

Quote from sle:

The delta component would be small. The mtm component dVega/dSpot can put you in sufficient amount of pain, but I assume you are calculating as if you are holding it to expiry.

but even that would generally be small compared to dC/dsigma unless you were hedge in vega.

Assuming you don't have something crazy convex or a huge move in vol.

TskTsk · Jan 30, 2012

Quote from sle:

The delta component would be small. The mtm component dVega/dSpot can put you in sufficient amount of pain, but I assume you are calculating as if you are holding it to expiry.

What I'm looking for is basically a way to calculate P/L per scalp when you're long gamma and continuously hedging delta. As far as I understood, 0.5*gamma*S^2 is ~accurate, but I fear there are other factors at play that may influence my P/L per scalp, so my question is what would these other values be? I understand vanna is negligible...and I disregard any IV movement as I hold until expiry.