I don't think this is an issue. Rebates don't outweigh costs right? So the taker would pay 1 and the liquidity provider would get 0.80. Therefore, there's no gain from 0-sum transactions... since there's a cost. If this wouldn't be the case, and MM's would collude in the way you suggest... I would think the exchange would lower or end the rebate scheme.
That's my point. Rebates encourage market making which benefits the traders who would otherwise have to find another venue. The exchange could rely on order size (I.e. You don't get fills in the 10k shares, only the 500 you present to market), but there's no reason to assume these orders don't get hit by retail as the bid moves (my experiences notwithstanding). Which is to say it's in the exchange's interest not to match orders against two market makers and pay both.
Sophisticated market making today is all volatility based. The systems adjust size and skew to reflect the net positions. The books at the big firms are all volatility books. This creates two market issues - most of the quote traffic isn't price changes it's size changes. As you provide liquidity some things become more attractive to you and some things become less attractive. This is frequently why markets don't move and a resting order that has been sitting around all of the sudden gets filled for no apparent reason. A MM can't be primary on more than one exchange(although there are exceptions). Of course a change in the cash will drive a price change, but frequently and a lot of the quote traffic is driven by skew changes. It can be very expensive to trade stock delta if you are not a specialist in the cash. The SEC didn't use to allow this - today it is common. So in essence you gain an advantage if you're large and have good technology if you trade a book rather than a series or a position. It's also a big advantage if you have a routing deal you have a sense of how your book changes. On up days you are commonly losing net delta and vice versa on down days. With equity, etf and index options it is very common that the book actually get adjusted by trading something other than what you would expect. This is also why firms like Citadel and SIG(and others)can just run over you if your technology and trading ability is not as good as their's. Think of all things that correlate to the big indexes and active etfs. Think about this - almost all of the microsecond to microsecond quote traffic is size changes.
You've said it... semantics However you look at it... you come to a certain volatility, by looking at the past/historical volatility and current volatility of the underlying, together with expected volatility (since we're pricing future options)...
>> ajacobson: [...] Optiver’s story [...] Well, I'm in EU and I applied to Optiver Amsterdam a while ago. I think the response I got was ... crickets. Not even an "after careful consideration, we regret to inform you that you have not been selected for interview". And anyways these developer interviews always assume that no matter how many years of experience and systems built (my name's Bernard Montgomery and I've successfully devised a strategy to push Germans out of Africa), you're a drooling idiot who's got to be tested on the most elementary (and preferably obscure) stuff (a 22 years old drill sergeant will test you on pushups and 10 miles run in full combat gear: ahh, you caved in after 9 miles, you're clearly a fraud). This is in the fortunate case when you actually get to speak with another developer, oftentimes you only get to speak to the HR, who decides you're not a good fit for company values. And if by chance you're still hired you're getting the soul-sucking and career destroying work of wiping some old man's ass, i.e. bugfixing. With bugfixing, the nice thing is no matter how many crap you clean, there's always much more coming at you. And the 'performance' metric always being not how many bugs you fixed but how many still are and keep forming. In a single good day, a developer can produce enough to keep one cleaning lady/man occupied for a year B-) And the icing on the cake, you have to sign a non-compete agreement and intellectual property transfer, to make sure that if you happen to have the wits and the drive to still do valuable development at home, outside work where you are not allowed to do it anyways, then all your work are belong to us. My experience is that employed finance is a shithole and I'm not looking to get back into that.
So you applied to a developer position? Not a trading position? No trader will do any 'bug fixing', even if they come from a programmatic background. You're a trader and that's it. Firms like Optiver generally want people that have no former experience, so they can train them in their own way. The non-compete for traders is usually a bit of a BS thing... since you would only know about current positions, which for MM's don't last longer than 2-3 months. They're fairly easy to break. IP, what would you expect? You think you can build programs, systems etc on your boss's time and money and walk away with it? They make it difficult, because they get 1000's of applicants... and everyone thinks they have a system or idea, and they all think they can make it... I've interviewed my share of newbies and fired my share of trainees/junior... less than 0.1% at the start knows what goes on in options market making... so the response you got maybe seems rude, but isn't really unheard of. Also, some companies (trading or any firm really) like to not send out any letter until the applicant actively calls them to check up... that's called being pro-active....
>> So you applied to a developer position? Not a trading position? Yes, a development position because I've been doing development most of my employed life (about 15 years) of which about 10 in finance. Started as generic developer with an options market-making desk in Asia then naturally drifted towards a quant dev role. Unfortunately they ran out of business before I started getting good at it, i.e. figuring out what matters and what can be done to get profitable again. Brute force (lower latency) was doing more wrong than good, the strategy and models themselves were flawed so more trades just meant more losses. Then I got moved to another division, pricing, volatility & risk for a portfolio manager and lost contact with people who knew and gave a damn. >> No trader will do any 'bug fixing', even if they come from a programmatic background. You're a trader and that's it. There was no career ladder towards trader where I worked, for funk's sake they even ridiculed and marginalized my efforts to understand quantitative finance. Take Black-Scholes formula from Wikipedia and type it in, any monkey can do that so quit wasting your time at work when we've got all these fires to fight (overcomplicated systems, constant outages, absurd figures, angry clients etc). >> Firms like Optiver generally want people that have no former experience, so they can train them in their own way. I have a pretty good understanding on how options work, how their price is formed and the significance of volatility. Even published my own model as a demonstration of my capabilities, as part of my PhD (interrupted for now). But of course the best (i.e. working) stuff is unpublished and it's based on hard won experience and countless experiments to get a grasp of how things work. >> You think you can build programs, systems etc on your boss's time and money and walk away with it? I asked my managers to do that during work time and was denied (at best they were amused at my proposal). So it's not on my employer's time. It is on my employer's money since I'm using them to pay for living expenses, but outside working hours the time and what I do with it is mine. Especially I repeat, since I'm denied using it properly while on payroll. >> They make it difficult, because they get 1000's of applicants... and everyone thinks they have a system or idea, and they all think they can make it... I've interviewed my share of newbies and fired my share of trainees/junior... less than 0.1% at the start knows what goes on in options market making... so the response you got maybe seems rude, but isn't really unheard of. I'm aware this is a number's game, just like trading. I never get into a trade where the probability of making a profit is even 10%, let alone lower. Even if there might be a ton of money at the end of it, I don't have the capital in trading and I don't have the time to waste on failed interviews in real life. It's a better use of my time to work a regular, non-finance developer job (hence get rid of the non-compete and IP clauses from guys who are anyways at best doing nothing of help to me and most of the times are actually hostile). Then in the spare time work on my own trading system (now thinking of upgrading from arbitrage to market making) and eventually employ other developers to help with the tedious stuff.
The discussion has deviated so I'm gonna bring it to my original point of interest. To open source or to not open source? a) If I open source there's a very slim chance I might get some help on the non-proprietary part so things would go faster. But my experience matches what one guy once observed, namely that there are higher chances of meeting a talking potato riding an unicorn on the way to a leprechaun pool party than finding a developer willing to work for free (because programmers are among the most cowardest, risk-adverse people you're gonna find in this world and everyone of them knows it's better to take a 100% chance of delivering a useless product that gets payed now than a 10% chance of something that'll make them rich in the future but draws resources from them now). So I'm providing my competition for free with a stub they can develop upon for the hypothetical chance of some hypothetical help. b) If I develop close source, then I don't get any help apart from what I can pay for, which is my own spare time after the day job and perhaps some small punctual tasks on Upwork. Or better not, if I provide my sources to some random guy in India, it'd be the same as #a and I'm also paying for it. What would you choose? Could be a #c I haven't thought about.