How fast are options market makers?

[zdreg]

Quote from FrankSlaughtery:

to answer another poster's question, i did not enter an order b/c the spread was too wide and it was impossible to calc fair value using b/s or any other model since the iv was off the charts. i still made a profit on the trade though.

in those situations you ignore b/s. you enter an order between the bid and ask and hope that u get filled.

 

[quatron]

Quote from nitro:

it would take days to finish the overnight computations for even small number of underlying move ranges. There are probably tricks that may have reduced the dimension, I just tried brute force.

You don't need to fill all the table overnight, all you need to do is to make sure the regions of your tables are filled by the time underlying gets there. You will also want to choose nodes carefully and store only them in the table and then interpolate in realtime to reduce dimension.

 

[nitro]

Quote from Rationalize:

Sure .. but first .. are you fast enough on a slow stock .. can you quote tight enough to get hit on a slow stock .. & can you do this without getting arb'd when the tick rate increases suddenly. Not easy imo.

On a slow stock with few strikes? No problem.

One trick is to do interpolation. Quantlib is quite good at this, e.g.

http://quantlib.org/slides/dima-ql-intro-2.pdf

 

[Rationalize]

Quote from nitro:

On a slow stock with few strikes? No problem.

One trick is to do interpolation. Quantlib is quite good at this, e.g.

http://quantlib.org/slides/dima-ql-intro-2.pdf

I would be surprised if you can cancel before being filled on a negative margin by a *hardware* based competitor. All you need is some order blast through the top price level on the underlying .. another mm arbs you in the option(s), and you're screwed .. now hedging 1 tick off .. which is esp an issue on a low priced stock, where a 1 tick is everything.

 

[nitro]

Quote from Rationalize:

I would be surprised if you can cancel before being filled on a negative margin by a *hardware* based competitor. All you need is some order blast through the top price level on the underlying .. another mm arbs you in the option(s), and you're screwed .. now hedging 1 tick off .. which is esp an issue on a low priced stock, where a 1 tick is everything.

I generally agree. If you look through my posts, I understood as early as 2003 or 2004 what needed to be done. I know exactly how to do it today in hardware, if I had the money.

For as long as I can remember, I have known that speed is of the essence in options MMing. The problem is, that all the fastest hardware and computers and table lookups are meaningless unless you have the fastest data coming into these systems to begin with. In order to do it correctly, you have to spend thousands of dollars a month to have access to direct feeds, usually collocated on a cross connect. Without this, no amount of optimization that you do after the data arrives on the NIC will do you any good. So why even bother.

The whole thing has become quite expensive. I have resigned myself to the fact that mm'ing is for the big houses that can afford it. Market taking is the only viable trading venue for even well capitalized retail traders.

 

[eudaemon]

Quote from Rationalize:

I would be surprised if you can cancel before being filled on a negative margin by a *hardware* based competitor. All you need is some order blast through the top price level on the underlying .. another mm arbs you in the option(s), and you're screwed .. now hedging 1 tick off .. which is esp an issue on a low priced stock, where a 1 tick is everything.

Especially when all the op arbing is done BEFORE the planned blast through in the underlying by the same guy doing the arbing. First the illiquid leg, THEN the liquid leg. Always, always, always. Impossible to counterattack.:eek:

Hence the need for slow folks like me to always be convex, if at all.