IB should allow user to set Smart's default exchange - Vote Now

Discussion in 'Order Execution' started by fbell50, Jan 29, 2007.

  1. fbell50


    IB should allow user to set Smart's default exchange - Vote Now.

    If you agree then please vote in the Interactive Brokers Features Poll on item 2000, currently on page 2 of the other category at http://www.interactivebrokers.com/en/general/poll/poll.php?ib_entity=llc#cat_oth.

    IB’s Smart routing should allow specification of the default Exchange/ECN while continuing to support the pre-open re-route feature. With the current turmoil in order execution and pricing it is no longer obvious that there is one clearly superior unchanging default Exchange/ECN. Some of the reasons are:

    1) Now that NYSE’s Hybrid system is fully implemented it executes the lion’s share of the volume. Formerly a large marketable order would cause the specialist to find liquidity at one price and then fill all intervening quotes at that price. Now Hybrid walks the book filling quotes without price improvement. While price improvement is still possible if your quote is past the LRP, the opportunities are greatly reduced.

    2) Reg NMS will soon be implemented. It protects the top of book on all fast exchanges, unlike the current trade through rule which only protects the NBBO. This means the NYSE will be forced to route a greater portion of their orders to 3rd parties.

    3) NYSE charges liquidity providers $.025/100. ARCA, Nasdaq, and others pay $.20/100.

    Add it all up and it is no longer clear that providing liquidity on the NYSE is the best option. Currently IB routes unmarketable orders to the exchange they are listed on. IB should allow specification of a default exchange. Preferably they would allow specification on an order by order basis and on a primary exchange basis (for example, all NYSE routed to ARCA unless overridden on a specific order.)

    Since the opening cross is a significant liquidity event, they should honor the pre-open re-route feature. So, assuming an RTH order, the order should first be routed to the primary exchange for the opening cross, and then routed to the default exchange.

    There is another similar proposal in the Interactive Brokers Features Poll, but it does not specify maintaining support for the pre-open re-route feature.
  2. I don't disagree with anything that you have written...
    But SMART is optional...
    You can route anywhere you like and accept the consequences.

    It's also good to keep in mind...
    That ** IB's profits ** are directly related to the quality of the SMART routing algorithm.

    So it's safe to assume...
    That 2-3 top software engineers work full time on the SMART Routing Algorithm...
    That they are getting continuous sophisticated feedback from many sources...
    And tinkering with the Routing Algorithm constantly.

    I work with the API...
    And was gonna code my own routing algorithm...
    But scrapped the idea...
    Because it presumes that I know more than the IB engineers.

    I'm sure many changes will be made in the next few months.
  3. fbell50


    Not if you want to preserve unbundled pricing using the api, a minimum difference of .0055/share (.008 vs .0035), a difference which grows with volume. That's a big difference to me.

    I'm not as impressed with SMART and I do not believe it is safe to assume they have 2-3 software engineers working full time on the routing algorithm. I suspect they actually spend little time on the algorithm and more on the mechanics of handling the communication with the exchanges. For example, IB recently recoded their interface to support version 3 of OUCH, NASDAQ’s updated communication protocol. And then there is ongoing maintenance and troubleshooting.

    I'm relatively new to IB, but I've executed 670,000 shares primarily as a liquidity provider and I'd say their performance is roughly equal to Ameritrade’s with whom I've executed over 87 million shares. I haven't executed enough with SMART to draw a firm conclusion on their relative performance. I suspect SMART is slightly better, but I have experienced at least one significant missed NASDAQ fill that would not have happened with Ameritrade (I was told it was routed to ARCA, but I could not find the order which would have been the NBBO for 30 seconds in either of two different T&S logs. How can my order have disappeared for 30 seconds? There’s obviously a downside to switching orders from one place to another.) For my trading I suspect I will find that SMART is at best a minor plus.

    But this is really beside the point. I do not see how the change I’m proposing would significantly impact the part of SMART’s algorithm I assume you are referring to, namely the handling of a limit order that becomes marketable. What is the impact if the order is switched from a user specified default exchange vs SMART’s default? SMART needs to cancel/replace the existing order on one exchange and dispatch an order to the marketable exchange. Is it faster to cancel/replace on the NYSE, AMEX, and NASDAQ than on ARCA?

    I do not see any downside to my proposal, and potentially a significant upside. Nothing would change for customers who accept the current defaults. Other customers could experiment to find what works best for them. It would also add one more pressure point on the NYSE to start paying liquidity providers.
  4. No, I don't think it is safe to make this assumption. IB has never allocated the human resources needed to optimize its SMART-routing algorithm.

    I personally found many bugs in SMART, and many other opportunities for optimization, which had been overlooked by the IB programmers. I had to expend considerable effort to get these problems addressed. Some of these problems were treated with such low priority that it took months or years to get them fixed, and others were never addressed at all. Market rules and structures are, furthermore, constantly changing, so that optimization would require frequent changes to the algorithm. The need for frequent change has been highly accelerated in past months, as compared to the past few years.

    I think IB has the best SMART-router available, but that doesn't mean it is perfect. A customer who knows what he is doing should be given the necessary options so that he can customize SMART to fit with his own particular trading strategies and his own particular knowledge on how best to route orders. IB should not fall into the trap of a "one-size-fits-all" or a "broker knows best" design philosophy for its SMART-router.

    I would just like to emphasize that I think IB has the best SMART-router, and that my criticism is made only for the purpose of helping make it even better.