Automated Options Market Making

[Jurassic]

@ronin thanks

[nikol]

> "You can't manage the risks on the level of a single contract."



On the level of single strike, yes, you can. almost always.

What you trade here is obligation and premium. If you are long, the max you loose is premium. If you are short option, then you can revert it to synthetic long for small cost. Maturity match is prerequisite. But even if there are no correspondent futures, you can always "box" it with an interest rate exposed, which we know how to hedge. etc etc.

It is a bit more complicated and involves more instruments, but all those additional steps are known and can be automated such that you execute the cheapest transaction.



it is like:

trade A or

trade B-C or

trade D-E-F -- whichever looks cheaper

[riskPremium]

@nikol,



sounds you are still hedging out risk on same strike/nearby options/underlying futures which might not give you a competitive quote.

[ronin]

> On the level of single strike, yes, you can. almost always.



Meaning what - you can trade one strike but multiple maturities? As @riskpremium pointed out, you still have vol model risk between maturities.



But it gets even better. Say you are long one month, and you can go short ten years, same strike. What would happen to your gamma and vega? Does it even matter it is the same strike?



> If you are long, the max you loose is premium.



Sure. And if you are short puts, the max you can lose is the strike price.



Or even better - no matter what you are trading and how, the max you can ever lose is your net assets. Still limited. So no problem I guess?



Why do we even bother with risk management - I mean there is only a limited amount of money in the world, so all losses are contained, surely. No?

[nikol]

"sounds you are still hedging out risk on same strike/nearby options/underlying futures which might not give you a competitive quote"



Yes. I treat strike as a single "market".



If you talk about "market wide" price alignment one can always build a price(strike) function without involvement of the concept of volatility. Trade around that price and account for various distortions and secondary effects too, like e.g. pinning, but volatility is not generally needed.



@ronin



I exclude cross-term trades from this discussion.

But you have to take into account all non-arb conditions and they are not volatility related. Once you speak about volatility, it is always about specific model.



Although I do agree that even in our discussion I have a difficulty to avoid volatility concept as it is 'burned' in the brain.



PS. and yes I have to limit myself to European options although I had some success to produce model free prices for Bermudans. For American can not say with certainty.

[ronin]

Sorry @nikol, I am a bit slow today. But I have literally no idea what you are talking about. None whatsoever.



You are right that if you are dealing with a single option contract, you have no need for a concept of volatility. Volatility exists as a way of interpreting a multitude of option contracts and their relationship with the underlying. If you are looking at a single option contract in isolation, sure thing.



But with that framework, you can not hedge a single risk of your option contract. Not delta. Not rho. Not gamma. Not vega. Not anything. All you have is some random walk with a weird returns distribution.

[nikol]

Ok. sure.



Use model free put-call parity to trade at single strike and

Use terminal probability distribution across strikes for single maturity for "competitive" pricing at single strike.

Agree with this concept in mind I would avoid creating risks across maturities and strikes. But still one can hunt for arbitrages across strikes, like this:

K1 "" Put(K2) , sell Put(K1), buy Put(K2), keep till maturity.



Let's come back to this later.

[Strange]

"Use terminal probability distribution across strikes for single maturity for "competitive" pricing at single strike."



That is a proxy for volatility, you have a terminal distribution that has some standard deviation etc. In fact, there were some people using that type of model before Black Scholes came around.

[Lebowski]

My questions pertain to being on the wrong side of the electronic eye trade:

1. [Main question]: how does an OMM manage the risk of getting more inventory than they were bargaining for all at once all in one direction because they were quoting many highly correlated instruments and experienced a “sharp delta move?”



Here’s my attempt to wrap my head around this without handholding:

1. Found this helpful post from @radikal:

> Well, if you're working a LOT of orders to hold queue position, and something happens, you potentially are filled on a LOT of deltas. This is especially a problem if you do this overnight when there's not much exit liquidity; so you keep some teeny puts on the book always or some vix call 1x2s etc.

2. Just a bit of uninitiated intuition. Higher delta = higher risk of getting hit when delta is stale, but in practice this is probably largely offset by how wide the bid ask is at different strikes. e.g. near the money you have only half the delta but because the spread is thinner the market doesn’t need to move against you as far for you to become stale.



2. [bonus points]: let’s say one of these Chicago OMMs experiences a knightmare. Because they run relatively lean in terms of collateral and they quote so many different products the net result once the dust settles is they’re filled on all sorts of stuff and they fall below margin requirements. Would they be forced to liquidate like anyone else? Obviously the Chicago props are sophisticated enough to have some level of risk management built into their autoquoters but what other safeguards exist to prevent this sort of scenario?



Thanks. Getting a lot more out of this phorum than I put back in but hopefully this will change in time as I become less green.

[ronin]

> [Main question]: how does an OMM manage the risk of getting more inventory than they were bargaining for all at once all in one direction because they were quoting many highly correlated instruments and experienced a “sharp delta move?”





I think Blackadder has it covered.



George: If we should step on a mine sir, what should we do?

Blackadder: Well, lieutenant, the normal procedure is to leap 200 feet into the air, and scatter yourself over a large area.