IB Best vs Best ECN Execution?

Discussion in 'Interactive Brokers' started by stock_punter, Sep 6, 2001.

  1. def

    def Sponsor

    jtrader,
    I assure you that you will not receive rude customer service out of the HK office. If you provide me (e-mail) the name of the rep who has not been helpful it would be appreciated.

    As for orders that get in the hands of mktmakers. this scenerio could possibly happen:
    a: order finds its way into the hands of a mm who holds onto it
    b: a better price is found elsewhere and a cancel is sent.
    c: by the time the cancel is confirmed, the other order no longer exists.
    d: the order gets sent back to the best price and ends up back in the hands of a mm and the process repeats itself.

    not sure if you can blame the broker for this as the algorithm is doing what it is supposed to.
     
    #11     Sep 7, 2001
  2. Magna

    Magna Administrator

    def,

    problem is what happens when an order gets into a market makers hands. They hold on to it and will not release it. Thus it is in effect stuck until
    they acknowledge a cancel.


    Now that SuperSOES is implemented for all NMS stocks, when you send a Cyber order using their version of a best routing algorithm it first checks ECN's like most algorithms, and if not filled it hit's the MM closest to the inside. If not filled it self-cancel's within a second (whether or not the MM releases it), and hits the next MM, etc.

    This makes for extremely fast fills as long as the stock is NMS. If it's a microcap or smallcap that is not NMS then it uses the older SOES and SelectNET methods in it's algorithm and things can grind to a painstaking crawl, like the old days...
     
    #12     Sep 7, 2001
  3. def

    def Sponsor

    thanks for the info. what is there is only 1 MM at the best price? Do they get ignored for an inferior price? (This may make sense in any event since they are holding an order)
     
    #13     Sep 7, 2001
  4. Babak

    Babak

    Magna,

    wow! That routing algorithm makes a lot of sense!! Wonder why IB doesn't do the same.

    I also wonder why IB doesn't make public its routing algorithm like their competition?

    :confused: :confused: :confused:

    -------------------
    def,

    any progress on the ARCA/REDI routing question + the IOC question? thanks :)
     
    #14     Sep 7, 2001
  5. aldrums

    aldrums

    Originally posted by jtrader: Def are you saying that the order can be held by a market maker for 2 min or more.

    Jtrader,

    You can be assured that market makers do not "Hold onto orders". The only limitation is orders on Selectnet must stay live for ten seconds before they can be cancelled. Orders in the SOES Queue are not controlled or seen by Market Makers and can be cancelled at any time. Market makers for NNM stocks such as MSFT are subject to automatic SOES execution whenever they are at the inside price, and have no choice in the matter.


    Alex
     
    #15     Sep 8, 2001
  6. aldrums

    aldrums

    Originally posted by def

    As for orders that get in the hands of mktmakers. this scenerio could possibly happen:
    a: order finds its way into the hands of a mm who holds onto it


    Def,

    Please explain your theory of how market makers hold onto orders. From everything I have read and experienced, this is impossible.
    Before Supersoes market makers had the potential to hold the price progress of stocks by only executing a 100 share order every 30 seconds, but they could not literally "hold onto orders". You could always cancel your order after ten seconds if it was not filled.

    Thanks,

    Alex
     
    #16     Sep 8, 2001
  7. def

    def Sponsor

    aldrums, that is what I meant. If you do not get a fill or an execution in 10 seconds, i call that holding onto an order.
     
    #17     Sep 8, 2001
  8. rfoulk

    rfoulk

    I've been using IB to trade equities for over a year. For most of that
    time Best Execution has been junk, at least when the stock being traded
    had any activity at all. Myself, and other's in the room I trade in
    kept testing it and kept complaining.

    A few months ago they started working on it, and after many interations
    it is sooooo much better!

    I used to use Island for most of my trades. For the past few months I've
    been using BE (Best EXecution) instead and mostly getting great fills.
    It's a great trading tool. But it does have some limitations.

    When there's lots of volume and lots of movement it's much safer to use
    Best ECN. After a number of discussions with several programmers at IB
    and enough good and bad fills I think I can give you an idea what the
    problem is.

    BE always attempts to trade at the very best price available. Sometimes
    that's just not possible.

    When an ECN has the best price -- you get filled immediately. When the best
    price rests with a Market Marker and prices are changing things can get iffy.

    Market Makers fill orders through a queue. Each time the operator
    changes his price those orders still in the queue for the previous price
    get dumped. They've missed the boat. When BE recognizes this it looks
    for another `best price' and starts the whole process again. (Oops.)

    An good trader executing manually would see that the only way to get
    filled in this sort of situation is to skip the best price and go for
    one a level or two away.

    This is a good example of why it's so hard to build a smart execution
    system that works for all situations. I've had a few lengthy discussions
    with the IB folks about various ways to work around this problem, from
    making BE a lot smarter, to allowing the user to provide some control.

    It's not easy.

    And no, Cyber's version of this stuff isn't any better, just different.

    BE is a great tool, which I use for the vast majority of my trades
    these days, even in some pretty difficult situations. But every once
    in a while I find I need to switch to Best ECN to get the job done.
    Fortunately, that's pretty simple.


    Richard
     
    #18     Sep 11, 2001