Hi,Three questions for improving my intuition if you may:1. When buying an option, what am I "paying for"? obviously it's not first order derivatives (delta - could have bought spot, vega - could have bought vol swap, etc).My answer is positive convexity, i.e. positive 2nd order derivatives (gamma, volga - dvega/dvol, etc.). But what about vanna (dvega/dspot) for example? Assuming a flat surface, I'm not paying for that...So I'm leaning towards 2nd order derivs of the same variable (time, vol, etc).My other prob with this thesis is that vega has a simple mathematical connection with theta, which is opposite to my conclusion that I'm not "paying for" 1st order derivs.Any thoughts pls?2. Using OIS discounting means that each swap leg isn't at par, because the forward curve isn't the discounting curve (say fwd libor 3m and ois dc). So let's say the floating leg is now worth 103 and the fixed leg is worth -103. Are there any practical implications for that fact in a single ccy swap?3. My workplace has CSAs with its counterparties, with different terms, such as: collateral ccy, threshold (different for each party), eligible collateral (cash, securities - % accepted depends on type), etc.I assume this situation isn't unique to my employer, and that there's quite a bit of $ hiding in these differences. Can you update on how it's done in big banks? is there a massive push for standartization of CSAs (pre-SCSA) or are these differences in CSAs handled by banks by incorporating them as much as possible into pricing?Thanks in advance for your inputs.

I'll take a crack at 1. On one hand, I think you are grossly overthinking the issue, on the other you seem much too tied into a Black-Scholes framework. When you buy an option, you are paying for the fact that in some future states of the world you will receive a positive cash flow while you will never experience a negative cash flow. That clearly has some value. For building intuition as to what that value ought to be I prefer to take a single step binomial model as far as it can go (which is quite far), and then look at what happens when you add a state to turn it into an incomplete market (a trinomial) and add a traded option to complete it again.

Thanks bearish.Let me be a little more specific, I'm looking to know what I'm paying theta for, not that options have value in general.If I'll move to a binomial tree, it'll be like moving 1 step forward in a 3rd (nonexistant) state, but I can't see how that helps me...

1. If I might add to what bearish said, you could also say that, among all those greeks, you're buying smth unique and hard to price, i.e. a "liquidity option" on the underlying. So in terms of what bearish described, the ability to avoid a negative cashflow in an extremely adverse future state of the world might be worth a lot.2. No, no implications that I can think of.3. One word (or three): CCP. The writing's on the wall and eventually everyone will converge to standard terms dictated by the clearing houses.My Z$2c...

2. There are two implications really. First, when trading you need to know what will be the actual discounting on the trade, and depending on particular CSA terms, you have slightly different mid prices. Second, even if you account for this little difference, and correctly hedge the trade against some other counterparty(-ies), you have hedged your swap position, but you still have some remaining positions in OIS arising from those different CSA details and therefore different discounting curves. So you have constantly changing OIS risks, and you actually have some cross gamma as well

Thanks.I'm also trying to think (regarding #2) if bootstrapping should be affected by this. OIS discounted swap rates are considered par? suddenly it's not that obvious to me.

well, par is where the current market mid price is, isn't it? Bootsrapping is obviously affected, because in pre-CSA world you only had one market mid price for a 3Y swap. Now when your broker tells you 3Y swap is 35/37 you have to clarify what discounting is assumed in those prices, as for different counterparties you would have different prices. So you observe the price, but you can't for sure tell if this price is what you consider mid market. You may need to apply some corrections to it before you mark your mid price there. Prices in the market can even go choice or inverted, because bid and offer want different CSAs as counterparties. So when you're bootstrapping the curve now, you also need to account for particular CSA in each point you observe in the market, which makes it much more difficult, as strictly speaking you need to have the curve to know how to translate various CSA prices into whatever you assume your internal standard. Luckily, in single-ccy swaps, around mid market, all those differences are not huge. It is becoming much more fun when your trade is already well in the money, or when you have a cross-ccy swap where you have massive notional exchange at the end, and need to accurately discount that huge cash flow