Seven Years of Order Information Leakage

 

“The Exchange believes that this interaction is inefficient and detrimental to investors, to members, and to the market.” – Nasdaq, 9/22/16

Back in February 2009, Nasdaq received SEC approval for the Post-Only Order type which was described as:

“A Post-Only Order is an order that does not remove liquidity from the System upon entry if it would lock an order on Nasdaq’s system for trading cash equities (the “System”). If, at the time of entry, a Post-Only Order would lock an order on the System it will be re-priced and displayed by the System to one minimum price increment (i.e., $0.01 or $0.0001) below the current low offer (for bids) or above the current best bid (for 2 offers).”

In their 2009 filing, Nasdaq claimed that “the Post-Only Order is designed to encourage displayed liquidity and to offer Nasdaq users greater discretion and flexibility to post liquidity on Nasdaq”.  They also justified this order-type by saying that similar order types already existed on NYSE Arca (ALO Orders) and BATS (post only orders).

What Nasdaq failed to highlight in this original filing was that in addition to displayed orders, post-only orders would also not interact with non-displayed liquidity.  In other words, if a post-only order was entered at a price which would lock or cross an existing non-displayed order, then that post only order would not trade and instead the price would be slid either one tick below for buys or one tick above for sells.  This is an enormous detail that has been taken advantage of for years by some market participants to give them an advantage over other market participants.

We have highlighted this order type many times since the SEC approved it in February 2009 and cited how we thought that post-only orders were unfair since they allowed the initiator to gain valuable knowledge about existing hidden order flow since these orders would have their price slide from their original limit price.  Our claims fell on deaf ears and Nasdaq (and other exchanges) has been allowed to use these orders for more than seven years.  Think about how much information was being leaked during those seven years?  How many algorithms placed non-displayed orders thinking that they were hiding their intentions but in fact were consistently being pinged by post-only orders?

On September 22, 2016, Nasdaq filed a proposal with the SEC to change the way that post-only orders interact with non-displayed liquidity.  They will no longer be price sliding post only orders when these orders interact with non-displayed liquidity.  Here is how they describe the changes in the rule filing:

“A Post-Only Order that is entered with a price equal to a resting Non-Display Order will be posted at its limit price (or its adjusted price if applicable), rather than being re-priced as it is today. This allows the Post Only Order to lock the resting Non-Display Order. Both the displayed Post Only order and the resting Non-Display order will remain available for execution at the locking price. In this way, neither order is disadvantaged; the Exchange Bid/Offer spread is tightened; and no signal is sent to the member that entered the Post Only Order. In this scenario, efficacy is maintained or enhanced for both the Post Only Order user and the Non- Display Order user compared to today.”

Read that highlighted sentence again.  Nasdaq states that under the amended post-only order protocol they will not be sending a signal about hidden order flow.  They have just admitted that they have been leaking information about hidden client order flow for the past seven years.  They detail the original post-only order behavior further by stating:

“In addition, the member entering the Post Only Order learns through the repricing action both that there is a Non-Display Order resting on the book and also the price at which the Non-Display Order is resting. The Exchange believes that this interaction is inefficient and detrimental to investors, to members, and to the market.”

This is beyond astounding.  Nasdaq has just admitted that one of their order types has been detrimental to investors, to members and to the market.  We have a lot of questions about this admission including:

Why did they just admit this now?  Why did it take seven years for them to realize this?  Where has the SEC been all this time?  Did the SEC direct Nasdaq to change this behavior? How much information about client order flow was leaked?  How much money was made off of that proprietary information?  What damages did institutional and retail clients suffer during the last seven years?  How many other Nasdaq order types are detrimental to investors and the market?  What about the other exchanges order types? Will we see similar filings soon from these exchanges?

We also wonder if the SEC will just approve Nasdaq’s amended post-only order type and hope that nobody notices? The problem for the SEC is that they approved Nasdaq’s post-only order type in 2009 and appear to have never realized that it was actually leaking client information.  How can the SEC punish Nasdaq for an order type that they approved?

We think we have proved today exactly what is wrong with equity market structure.  For-profit exchanges have lost sight of their original purpose and instead have continued to invent new ways to attract their highest volume customers.  Their goal is not to help traditional investors but to figure out ways to make their own market data products more valuable.  Proprietary data feeds are valuable not only because of their speed but also because of their content (revealing information about hidden orders is very valuable).  In addition, this lack of knowledge about post-only order behavior demonstrates that regulators are not properly vetting exchange proposals. Maybe this is because of the lack of comments by the industry or maybe this is because of the regulatory revolving door?  Either way, how can we have confidence in our market structure if exchanges are continuing to disadvantage traditional investors and regulators have no clue that this is happening?