IB vs Thinkorswim RE: price improvement

[jimrockford]

IBSoft,

since we are on the subject of Automated Trading Desk, I thought it would be a good idea to mention that ATD's order routing algorithm does scan for hidden liquidity, at least for stocks. I'm not sure if it also does for options, and of course, hidden liquidity scanning for options would be a bit more complex because of order cancellation fees. My point is that since ATD scans for hidden liquidity, IB should take this fact as one more reason why it should consider improving SMART to include such functionality for hidden liquidity scanning. It appears to me that IB-SMART's lack of hidden liquidity scanning may be behind the recent spate of public complaints that SMART is giving inferior performance.

 

[IBsoft]

Quote from jimrockford:

IBSoft,

since we are on the subject of Automated Trading Desk, I thought it would be a good idea to mention that ATD's order routing algorithm does scan for hidden liquidity, at least for stocks. I'm not sure if it also does for options, and of course, hidden liquidity scanning for options would be a bit more complex because of order cancellation fees. My point is that since ATD scans for hidden liquidity, IB should take this fact as one more reason why it should consider improving SMART to include such functionality for hidden liquidity scanning. It appears to me that IB-SMART's lack of hidden liquidity scanning may be behind the recent spate of public complaints that SMART is giving inferior performance.

Thank you for your post.

To my knowledge there is no such thing as hidden orders in options.

Our stock smart does scan for hidden liquidity in some of the pools we have access to. I am looking into what we could improve.

 

[Option Trader]

If I may add that TOS displays the BRLC quotes in tenth & hundredth penny increments. With IB, I don't have this, and I'm wondering if this service is available with IB.

 

[jimrockford]

Quote from IBsoft:

Thank you for your post.

To my knowledge there is no such thing as hidden orders in options.

Our stock smart does scan for hidden liquidity in some of the pools we have access to. I am looking into what we could improve.

IBsoft,

maybe this will help.

SMART could and should scan for hidden liquidity, for example, by momentarily flashing each stock order at progressively more aggressive prices on INET, ARCA, BATS, etc., before allowing a non-marketable order to rest at a default destination, or before sending a marketable order to interact with the best available displayed quote.

Hidden orders do exist in options, but not the kind you are thinking about. Hidden orders need not reside in order books at exchanges. Any off-exchange black box, for example, can monitor the displayed quote, and send an order to interact with that quote when it reaches an appropriate level. This would not be a hidden order native to the exchange, but it would be hidden liquidity that could be found by the use of appropriate scanning logic in the order routing algorithm.

Another example of hidden liquidity would be a floor trader, specialist, or market-maker who is not publicly displaying the price at which he or his machine is willing to trade, and who has not lodged a native hidden order in the public order book, and who is instead waiting for display of an order at his price, so that he can immediately interact with that order as soon as it is displayed. This applies to options as well as stocks.

Options exchanges also have other order types which would not be termed "hidden orders", but which do amount to hidden liquidity. The price improvement auctions on BOX, for example, provide hidden liquidity. You have to flash your order on BOX in order to benefit from that hidden liquidity. Price improvement, in general, is one form of hidden liquidity.

The thing to remember about hidden liquidity is that hidden orders native to exchanges are only one of the many forms hidden liquidity can take. Options may not exactly have what are called "hidden orders" native to exchange order books, but they do have other similar forms of hidden liquidity, and so it is necessary to scan for hidden liquidity in order to have an optimal algorithm for order routing of options as well as stocks.

 

[IBsoft]

Quote from jimrockford:

IBsoft,

maybe this will help.

SMART could and should scan for hidden liquidity, for example, by momentarily flashing each stock order at progressively more aggressive prices on INET, ARCA, BATS, etc., before allowing a non-marketable order to rest at a default destination, or before sending a marketable order to interact with the best available displayed quote.

Hidden orders do exist in options, but not the kind you are thinking about. Hidden orders need not reside in order books at exchanges. Any off-exchange black box, for example, can monitor the displayed quote, and send an order to interact with that quote when it reaches an appropriate level. This would not be a hidden order native to the exchange, but it would be hidden liquidity that could be found by the use of appropriate scanning logic in the order routing algorithm.

Another example of hidden liquidity would be a floor trader, specialist, or market-maker who is not publicly displaying the price at which he or his machine is willing to trade, and who has not lodged a native hidden order in the public order book, and who is instead waiting for display of an order at his price, so that he can immediately interact with that order as soon as it is displayed. This applies to options as well as stocks.

Options exchanges also have other order types which would not be termed "hidden orders", but which do amount to hidden liquidity. The price improvement auctions on BOX, for example, provide hidden liquidity. You have to flash your order on BOX in order to benefit from that hidden liquidity. Price improvement, in general, is one form of hidden liquidity.

The thing to remember about hidden liquidity is that hidden orders native to exchanges are only one of the many forms hidden liquidity can take. Options may not exactly have what are called "hidden orders" native to exchange order books, but they do have other similar forms of hidden liquidity, and so it is necessary to scan for hidden liquidity in order to have an optimal algorithm for order routing of options as well as stocks.

Jim,

I thank you for your advice and will take it under consideration.

(In the above you make some statements that are not correct, e.g. your paragraph about BOX is incorrect).

IBsoft

 

[jimrockford]

Quote from IBsoft:

Jim,

I thank you for your advice and will take it under consideration.

(In the above you make some statements that are not correct, e.g. your paragraph about BOX is incorrect).

IBsoft

Oh, yes, I see I was in error. The BOX price improvement auction process, it appears from my quick look at the BOX website, is not triggered by non-marketable orders. Sorry, I haven't traded options in years, so I am liable to get some new details incorrect.

I don't think that any errors as to details undermine my basic argument, which is that hidden liquidity exists for stock and options alike, and that an optimal order routing algorithm will allow the customer to set an option to first try to benefit from hidden liquidity, before giving up on it and trying to hit displayed liquidity. This functionality needs to be added to SMART. Thank you for considering my suggestion.

 

[IBsoft]

Quote from jimrockford:

Oh, yes, I see I was in error. The BOX price improvement auction process, it appears from my quick look at the BOX website, is not triggered by non-marketable orders. Sorry, I haven't traded options in years, so I am liable to get some new details incorrect.

I don't think that any errors as to details undermine my basic argument, which is that hidden liquidity exists for stock and options alike, and that an optimal order routing algorithm will allow the customer to set an option to first try to benefit from hidden liquidity, before giving up on it and trying to hit displayed liquidity. This functionality needs to be added to SMART. Thank you for considering my suggestion.

The point of BOX is not that the auction is not triggered by non-marketable orders. Rather, the broker has to first (off line) find a party that will provide at least 1 penny worth of improvement. Without that the auction will not start.

I don't agree with scanning for liquidity in options for several reasons, the most important one of them being the cancellation fees.

I agree w/ scanning for liquidity in stock. IB already does that in several liquidity pools. We will expand this functionality and likely add ATD as a venue. (BTW, I looked at IB audit-trail yesterday for BRCL and I could see executions that received an improvement).

I am leaving the thread now.

 

[jimrockford]

Quote from IBsoft:

The point of BOX is not that the auction is not triggered by non-marketable orders. Rather, the broker has to first (off line) find a party that will provide at least 1 penny worth of improvement. Without that the auction will not start.

I don't agree with scanning for liquidity in options for several reasons, the most important one of them being the cancellation fees.

I agree w/ scanning for liquidity in stock. IB already does that in several liquidity pools. We will expand this functionality and likely add ATD as a venue. (BTW, I looked at IB audit-trail yesterday for BRCL and I could see executions that received an improvement).

I am leaving the thread now.

Allowing customer orders to interact with dark pools is only one way to scan for hidden liquidity. I don't think it is sufficient, all by itself, to achieve optimal order routing.

I am more concerned with hidden liquidity which does not execute in dark pools, but which instead executes in places like INET, ARCA, BATS, etc.

A hidden liquidity scan would go something like this, assuming that you are trying to buy stock at the market, with an NBBO displayed of 30.32 X 30. 36, where there are N different possible order destinations:

SET fullyexecutedflag = FALSE

SET orderprice = NBBO of 30.32

SET orderdestination = 1


WHILE NOT fullyexecutedflag DO

_____flash order at orderdestination with limit price equal to orderprice

_____IF fully executed THEN SET fullyexecutedflag = TRUE

_____increment orderdestination

_____if orderdestination > N THEN
__________SET orderdestination = 1
__________increment orderprice by one tick
_____ENDIF

END WHILE

 

[Option Trader]

Quote from jimrockford:

Allowing customer orders to interact with dark pools is only one way to scan for hidden liquidity. I don't think it is sufficient, all by itself, to achieve optimal order routing.

I am more concerned with hidden liquidity which does not execute in dark pools, but which instead executes in places like INET, ARCA, BATS, etc.

A hidden liquidity scan would go something like this, assuming that you are trying to buy stock at the market, with an NBBO displayed of 30.32 X 30. 36, where there are N different possible order destinations:

SET fullyexecutedflag = FALSE

SET orderprice = NBBO of 30.32

SET orderdestination = 1


WHILE NOT fullyexecutedflag DO

_____flash order at orderdestination with limit price equal to orderprice

_____IF fully executed THEN SET fullyexecutedflag = TRUE

_____increment orderdestination

_____if orderdestination > N THEN
__________SET orderdestination = 1
__________increment orderprice by one tick
_____ENDIF

END WHILE

(For Jim, or IB, or anyone) Does the smart route issue that you mention also explain the reason for lackings in getting price improvement? Or does IB actually get the price improvement & not pass it on to the customer?

 

[jimrockford]

Quote from Option Trader:

(For Jim, or IB, or anyone) Does the smart route issue that you mention also explain the reason for lackings in getting price improvement? Or does IB actually get the price improvement & not pass it on to the customer?

I would say yes, to your first question. I believe that the recent spate of public reports, that IB failed to give price improvement in a situation where a competitor did give price improvement, probably was a direct result of IB not doing the particular type of hidden liquidity scan I just described. IB interacts with dark pools, but this is not sufficient to eliminate the need for the type of scanning I am proposing.

My belief, in response to your second question, is that IB does not get price improvement, withold it from the customer, and instead keep that price improvement for itself. I believe IB's assurances that its proprietary traders use the same SMART order routing algorithm made available to IB customers. I believe that IB is a basically honest company, that honesty is part of its business model, and that it acts in good faith with respect to order routing.

I would also say this. If IB does take the other side of a customer's buy (sell) trade, at the NBBO ask (bid) price, when in fact, the true hidden ask (bid) is a more advantageous price which would have been achieved by proper scanning for hidden liquidity, then IB does take a benefit for itself at the customer's expense, and I think this is a breach of the broker's duty to his customer. But I also believe that IB sincerely hasn't yet figured this out, and that IB is unaware it is doing this, and that there is no intentional wrongdoing, and that it is only a matter of time before IB does figure it out and make the necessary changes to SMART so that the situation will be corrected. I don't believe that any of IB's retail competitors are doing a better job of order routing than IB, even if there are some specific cases in which a particular competitor does better on a particular order.