Options Plumbing & Arbitrage

[guru]

Just wondering why/how options market makers do not step on each other's foot, while also being able to decide about pricing of multi-leg combos, for example "locking" the pricing of butterlfies, spreads, boxes, etc - without allowing anyone else to play along. For example if you try to buy or sell a box, you won't be able to make more than the risk-free interest rate.
In trading you'd expect that someone may let go an option a bit cheaper, or pay a bit more than it's worth (slippage that happens to everyone), and even various market makers should not be able to adjust all their prices in exact sync, so during quick price swings you'd expect that one might fall behind with adjusting their quotes, etc.
In the end, at least theoretically, if you'd have a box, butterfly or any spread order with mostly static value already resting on options exchanges, you should be able to "catch" some occasional price misalignment and get "deals", or basically participate in market making. I'd also expect that market makers would step on each other foot and while one is adjusting their quotes, another one could arbitrage them away.
You can still find some seemingly mispriced options, but it is now common knowledge that there is no way to arbitrage them, your orders may only tighten the bid/ask around their true value. Even when occasional arbitrage opportunities still show up (tech holes in the system), they just as quickly disappear, permanently. I still see some opportunities in non-risk free semi-arbitrage, while everyone assumes that all basic risk-free options arbitrage isn't possible.
All this is fine, I would just like to understand the mechanism behind option combo prices being "locked" without room for making an extra cent on a combo like boxes, even with very wide spreads on heavily traded instruments where you'd expect retail buyers and sellers continually losing money on slippage, yet only market makers seem to be able to profit from those.
You never even hear about options traders snatching anything beyond its value, and everyone just places "regular" orders trying to get reasonable price for the purpose of making their own trade, not because they're looking for deals on options themselves.
And the practical question/purpose is whether it's even worth spending time trying to arbitrage some retail orders, and basically doing a bit of market making.

So how are risk-free option combos "guarded" against anyone else besides market makers being able to play along? Or is it still possible, or what happens when multiple retail traders have buy/sell orders on options that make up a spread at less than its value, while you may already have a resting order for such spread?
And if market makers didn't have any advantage, could it be expected to have a few box orders and other spread orders, and sometimes get them at a small discount to their value?

(some of my arguments & examples may be exaggerated and not necessarily always valid, used only for the purpose of describing the question)

 

[gaussian]

I'm not an MM but from what I understand MM's don't always arbitrage. They are busy quoting a fair market based on a model of the market, and the competition is who has the better model.

As for a retail participating in "psuedo market making", I am actually really interested in this. I would assume for a retail the strategy is the same in that you're trying to quote a fair market based on a model, but you've got the problem of execution which is where you're going to get stung.

(The old "make me a market on a phone number" question seems to be enlightening here)

This probably wasn't helpful at all but it got me thinking - thanks.

 

[FSU]

Not sure how other exchanges COB's (the electronic book for option spreads) work, but with the CBOE MM's can't leave resting orders in the COB, they can only interact with one when a customer places an order in. So when a customer puts a spread order in the COB at a "good price" an electronic auction takes place among the MM's and may give the customer a price improvement.

This is really a chance for a customer to have an "edge" as well. If you are watching time and sales and see a spread is trading at a "good" price you can enter the opposite side and leave it in the COB. When it trades again, you will be able to participate. This is where you have an advantage if your broker allows you to direct where your spread is sent. For example if you see these spreads trading at the CBOE, you would want to direct it there. Many brokers wont allow this, they use a routing that will make them the most money, so your order may end up on the C2 and you wont be filled as it trades on the CBOE.

 

[guru]

competition is who has the better model

This also is an interesting point and could be a part of my question. Because I'm sometimes able to derive some parameters or constants from the overall "market model" based on the options pricing themselves. For example my model may determine that I may be able to sell 1 option while buying 3 different but related options elsewhere on the calendar for $0 total cost, and that cost never goes much below $0. And I can validate that this is always true based on price history, which simply tells me that my model is correct and matches what happens in the market. But, I'm concerned that if that "market model" gets changed by someone/something then it will no longer align with my model and then my combo may become worth less than $0. So I still feel like there is some "controlling force" that decides about the "market model", and that can change. I mean it's still Black-Scholes on the surface, but there are various skews, volatility smile, IV itself, and "difficult" instruments like VIX and UVXY where there is room for elasticity. And it almost feels like "something" controls that model and the model itself is quite static, but I'm also not sure whether the pure market supply & demand controls the model and forces it onto MMs, or how specifically such model seems so stable, and under what conditions it may change.

 

[gaussian]

I mean it's still Black-Scholes on the surface, but there are various skews, volatility smile, IV itself, and "difficult" instruments like VIX and UVXY where there is room for elasticity. And it almost feels like "something" controls that model and the model itself is quite static, but I'm also not sure whether the pure market supply & demand controls the model and forces it onto MMs, or how specifically such model seems so stable, and under what conditions it may change.

You will find that MMs are using models like SABR rather than pure BSM. By using one of these models (aside from information they have that you don't) you may be able to get closer to your goal of modeling a market.

 

[guru]

You will find that MMs are using models like SABR rather than pure BSM. By using one of these models (aside from information they have that you don't) you may be able to get closer to your goal of modeling a market.

I may look into SABR and other models in the future, though usually the intelligence bias (following specific standards) kills my ability to find flaws in it and hack it. And I usually look at very low and very high strikes, where all models seem to have issues, so I may get a bit of an edge by deriving implied model and information from pricing (if pricing is derived from information then it should go both ways anyway).
But still, if multiple models (and information) are used by different market makers, then how & why they don't step on each other's foot and arbitrage each other's away? For example Citadel's model could buy up all options they consider too cheap, while Virtue may consider them too expensive and be selling them based on their own model. Unless the supply from Virtue would meet the demand from Citadel and the price would settle somewhere in the middle, which I think is the general definition of how prices and IV get settled - through market demand.
And that's where I also see room for arbitrage - when multiple MMs use different models then there should be some room for price inaccuracies and swings that wouldn't keep even box prices so stable. So it feels like there is another level of control that prevents price anomalies. Though it was also a good point from FSU that individual exchanges may work differently, while placing orders directly at exchanges may have some benefits.
(though I'm pretty sure you still won't be able to get discounted boxes directly at exchanges, as I've tried that)

 

[traider]

the big ask spreads for many options is huge, the pricing diff in the various models by different MM are tiny in comparison.

 

[guru]

the big ask spreads for many options is huge, the pricing diff in the various models by different MM are tiny in comparison.

In the past, even earlier this year, I was able to do some options arbitrage in the mornings after earnings, or during days when some news on specific stock/underlying were announced - during such times even the underlying stock price may zigzag up and down for a few seconds, and thousands of option prices may get adjusted every millisecond or so. Many resting retail orders also get swallowed instantly during news events (and again, now MMs seem to be the only ones being able to swallow them). Certain arbitrage strategies worked when you already had some resting orders on exchanges, especially on more complex option combos spread across calendar with value that could be difficult to calculate instantly.
But now nothing works, which means “someone” fixed some bugs in the system. So the question is who and what did they fix

On another hand, I’ve read in some exchange specs that exchanges themselves prevent things like buying basic spreads or calendars at credit. That’s fine, but do they also calculate the exact box value and prevent anyone from making profit beyond the interest rate? At least that’s how it feels, though the mechanism for this may not need to be controlled by the exchanges.
While it’s just difficult to believe that all market makers adjust their quotes the same exact way in an instant.
Oh, possibly exchanges also need time to scan resting orders, so cannot fill them in an instant. While they could even purposely delay fills to allow MMs to adjust & settle their prices. Just a thought, but not sure if that’s what happens.

 

[traider]

" more complex option combos spread across calendar with value that could be difficult to calculate instantly"
this isn't difficult to calculate, mm have systems that are optimised heavily, in fact don't even have to do full model calculations for certain scenarios. mm is so competitive even Timber Hill died.
you might be lifting other retail orders?

 

[guru]

" more complex option combos spread across calendar with value that could be difficult to calculate instantly"
this isn't difficult to calculate, mm have systems that are optimised heavily, in fact don't even have to do full model calculations for certain scenarios. mm is so competitive even Timber Hill died.
you might be lifting other retail orders?

Some of my friends argued with me that I was lifting retail orders, but to me this was very unlikely because I was trading 100s combos at a time on far OTM options that usually wouldn’t even have much or any volume.
Though possibly that’s it: even Wikipedia states that models like SABR break down on options far from the current underlying price, so the models themselves could be broken - and I could see this in options pricing. And such breakage could actually be favorable to MMs at the time as they were posting very wide bid/asks on options worth $0.05, but they might’ve fixed any issues once they noticed leakage when being arbitraged away.
So ok, MMs may now utilize their models across full options spectrum, etc.
But this still doesn’t explain two things: (1) the point you’ve just made that retail traders should at least be able to arbitrage orders of other retail traders, unrelated to MMs unless MMs get in the way and flush all the resting retail orders during news events while somehow being able to bypass other retail orders - and no one ever claims they’ve got a box spread at a discount: and (2) I feel like it’s just a single MM controlling options pricimg, or I don’t understand how 10 different models (or even same ones) from 10 different MMs wouldn’t step on each other’s foot when prices go crazy and may go from $0.05 to $1 during news events. Though some people make money during such events, just not through risk free arbitrage.