Put Vega vs. Call Vega Risk Management

[Patrik]

I think you're overcomplicating things. If you're only looking at a single risk reversal where you're long the put and short the call, short underlying - then the only hedge that provides the kind of protection you're looking for is selling some puts and buying some calls. Need not be same strikes - you may see "value" in some other strikes, and we're back to taking a view on what is cheap and expensive.



No model in the world will tell you how the realized implied vol surface will move - hence you can estimate greeks and hedge to these all day long, but at the end of the day you're doing this IN RELATION TO YOUR MODEL, i.e. the actual outcome depends on how close your model is to reality.



So instead of thinking about mathematics here I'd recommend you start thinking about vol dynamics, what you're assuming, what you're not assuming, what seems reasonable, etc. Once you have actively decided what to believe about the world, then start work out how you want to manage RR risk in such a world/model.

[silverside]

Here's something for you to think about and maybe implement



So you take 3 points of the vol surface each day (to keep it simple assume just a single maturity)

For arguments sake, ATM, ATM *1.01, ATM *0.99



Now you can quite easily fit a curve through these 3 points

The constant represents your ATM vol

The linear parameter corresponds to your risk-reversal risk

And you could call the quadratic part your butterfly risk



Then of course the spot moves day to day



If you calculate simple Greeks to these 4 stresses

And use these to predict your PL each day (calculating the market move in a simple way )

You should gain a good understanding of what is driving your performance



If you think your are flat Vega overall then you should see this in your PL prediction

And then after tracking this for a while you can see pattern of delta profit, RR profit, etc.



Good luck with it!

[TakeItAndRun]

First, I think "FX options and toxic products", Wystup, is one the best book about convexity.



It reminds me a huge long call spread position I had on a tech stock, 6 months before expiration, spot between both strikes. The strategy I set up was simple: mark to market vol only when spot approachs the upper strike or the lower strike.



For instance, if the spot increases then

- upper strike vol decreases, less effect on negative gamma for this strike

- lower strike vol increases, more effect on positive gamma for this strike



I think to understand your options book, you need a good pricer (accurate forward model) and a portfolio manager with mark to market data (implied forward, implied volatilities obtained from the same pricer).

[silverside]

@TakeItAndRun : without wishing to divert too much from the main discussion in this thread : do you really think the implied forward (i.e. divs and financing costs) makes a big difference ? for a vol strategy, would it be enough to use some simplifications (e.g. LIBOR discounting and divi assumption updated for example once per day) ?

[TakeItAndRun]

As Patrik said, the actual outcome depends on how close your model is to reality.



Yes, for long-dated options it makes a big difference. If the forward is not adjusted during the day because of a sudden news then the model moves away from reality.



Moreover, one should bear in mind that, at constant forward, the forward model impacts implied volatilities even for European options.