• Apr
  • 12
  • 2007
  • 12:56 PM

Questions about trades outside the National Best Bid/Offer

By: Ray Pellecchia
File Under: NYSE, NYSE

We've had a number of comments and questions in this space from people getting trade executions outside the best bid or offer. Four such posts are pasted below, annotated with some comments from my colleagues, which I've tried to summarize.

Before getting to those reader posts, here are some general comments:

-- NYSE has a "ship-and-trade" policy. If an NYSE execution (electronic or manual) would trade through an away-market quote, our system will ship up to the full quantity of that other market's displayed quote, and immediately proceed with trading the balance on NYSE, without waiting for a response from the away market. Our sweep prints may go to the tape before the away market executes the order we sent, making the resulting trades look out of sequence or outside the quote.

-- Often, the away-market quote does not fill our routed order and re-quotes, so there is no corresponding away-market print even though it may have been the National Best Bid or Offer at the time. Other markets have a cumulative fill rate about half that of NYSE.

-- The away-market quote may have reserve behind it, but we ship only to the displayed quantity, consistent with Reg. NMS requirements, as stated in this space previously. The away market might replenish its quote immediately after it fills our routed order, but the balance of the order is already executing on NYSE, at a price that can be inferior to that replenished quote.

-- These are general observations, and individual situations might have different roots. And this is not to make excuses or shift responsibility, but instead is my best effort to try to explain what you may be experiencing. To my read, it sounds like these types of situations are not unique to NYSE, but reflect a highly automated, Reg. NMS environment.

Here's what readers sent:

Please have your colleagues take a look at the stock [symbol deleted].

Perfect example of the NYSE problems. Prints are going off outside the NBBO (or are lagged enough that they show it as such on the T&S), and there is no E condition indicating that the prints were auto-ex'd.

What is the NYSE doing to remedy these problems?
-- J. Cancel

J. -- This might be a result of the situations referenced in the bullets above. The trades not marked with an "e" could be manual trades, or could be a result of an issue that we've had with certain automated sweep trades not being correctly marked with an "e". We're fixing that marking issue.

Recently I've noticed that on some NYSE stocks, there are lots of prints that go off that are maybe 10 cents (or more) away from the bid and ask. These prints seem to go off throughout the entire trading day and they appear to occur more often on the higher volume stocks like TWX and EMC.

At first I thought the problem was with the software I was using, but then I noticed the prints also showing up on eSignal charts.

I didn't notice this until Mon Feb 26, 2007. Are these prints delayed or are they being filled at another market center? (I noticed that these prints were flagged as "T" for the market center).

Thanks!
-- JC

JC -- This might be related to the bullets above. Note, "T" is the indicator for trades done on or reported through Nasdaq.

The other day i send a market DAY order to buy 200 shares of a stock. There was an offer at Arca and i got filled in NYS for 200 shares. I asked people here they said that, NYSE is not complying wih REG NMS for market DAY orders, but things work fine if it is a market IOC (immediate or cancel) order.

MY question is: is there a difference between MARKET DAY ORDERS and MARKET IOC orders? Under what circumstances, these two order types behave differently?

I only send market orders for usually 100-300 shares and want to learn the proper routing method to get the best inside price available including all exchanges for NYSE listed stocks.

A market day order and a market Immediate or Cancel order are essentially the same from an execution and shipping standpoint, but the former will post to the book if the quote is in a slow condition (if not slow, it would simply execute), while the latter will cancel if the quote is slow.

Note: we ship all orders to better away-market quotes at top of book, in compliance with Reg. NMS rules. There are, however, certain order types that will not ship: Reg. NMS Immediate or Cancel orders and Intermarket Sweep Orders. Note, incoming commitments from other markets are considered ISO/IOC as well.

One reason you may have seen your market day order execute on NYSE when there's a better Arca quote may be that NYSE already shipped interest to the Arca quote. We maintain an internal balance for each quote and ship only up to the quantity of that quote.

I thought you should know about the state of things from a different perspective. I run a stat arb hedge fund. Market data is the most important thing to us. I understand what happened with the volume surge, hybrid and the 3rd punch in the eye - Reg NMS. I know you guys have taken strides to fix systems. Sadly, this example is indicative of the real problems: yesterday in [stock symbol deleted] - a very busy stock due to an announced deal - the NYSE was ridiculously behind in its trade reporting. To make sure it wasn't my own systems, I verified the trade report sequences in 3 different venues: Bloomberg, my own feed from my broker (Lime), and an eSignal feed. Obviously all these feeds rely on SIAC. For instance, with e stock quoted 31.2 - 32.23 and prints on Arca, Inet AND NYSE going off at the logical prices of 31.2, 31.22, etc, all of a sudden a burst of prints at 31.15 would occur (all NYSE). Since it happened so fast I wanted to make sure they weren't real so I placed bids at 31.17 and of course a few minutes later another burst at 31.15 and no fill on my order (I didn't expect one) Clearly these were old trades and NYSE was playing catchup. This happened all day. My system relies on accurate sequencing of prints as well as quotes and of course, there were NO condition codes to help me filter. I can't verify how many stocks were affected by this but the real insult was when Lime and I called the NYSE (DOT services) the response was that no problems had been reported in [the stock]. Blissfully unaware! This is the real issue (not the whining "I didn't get price improvement" complaints you usually see). Right now I have no faith in the integrity of NYSE data (that I PAY for) and I am trying to figure out how to adjust my data model to compensate for this serious issue.
-- Mike Pratt

Mike -- I checked around and could not find any apparent data or other problem in that stock that day. I would suggest that your experience might be related to NYSE routing orders to top of book away, and the balance going off against the NYSE book, while the National Best Bid/Offer gets replenished from reserves from away. But that wouldn't explain why your "test" bids at 31.17 didn't get hit. Question: did you post those bids on another market? If you did, the order placed to NYSE would have hit the top of book in that other market, but not your bids deeper in the book.

To really know definitively what happened in this situation (as well as the others in this post) would require review by our Market Surveillance folks, a formal and confidential process that traders can initiate with an online request.

I want to thank everyone for writing and sharing their experiences and questions, and I also want to reiterate that we're continually working on ways to improve the trading experience. Will let you know as soon as I have word of new developments.

Tags: , , , , , ,

Comments

I concur with Mike et al. I have seen the exact same behavior in many stocks over the past few months. Under certain conditions, NYSE prints coming in on the CTS feed are out of sequence. Mike's description of the behavior is exactly right, and the explanation that I offer Mike is to look at the prices of those odd prints. You will see a pattern -- the price is where the stock was trading N minutes before (not always constant). You will see that these prints always reflect where the stock was trading at some previous time (3-5 minutes is my experience), which suggests that these are simply late prints. When this happens, not *all* NYSE prints are late -- just a spurt here and there. This suggests that some portion of the NYSE's new systems is to blame, rather than the system as a whole.

Incidentally, I went through the same process of verifying that the problem was streetwide, reporting it to the NYSE, and receiving the same response.

by Ian Cognito on April 12, 2007 1:25 PM

Ray,
Excellent explanation on REG NMS and routing procedures above. One question though...What is the reason for deleting the symbols that posters are giving you as examples? I think we could all learn from these experiences.

by Jimmy on April 12, 2007 1:50 PM

i know exactly what you are talking about. the prints will go off about N seconds above a stock that begins to trade lower and n seconds below a stck that begins to trade higher. seems like late prints. thanks hybrid

by bri on April 13, 2007 3:30 AM

Jimmy --

I set that policy back in November 2006 in this post. An excerpt:

To answer questions about a specific situation, I would need to involve our NYSE Regulation folks, who would need to examine the situation from their perspective. That involves gathering and analyzing data and could possibly lead to further review, interviews with market participants, and other investigative steps, even formal regulatory actions. There's no way that could happen quickly enough for us to carry on a blog conversation. So I'm going to excise some detail as needed from the trading scenario, in the name of keeping the blog timely.

I totally understand your point about getting the maximimum amount of educational value out of the trading examples, but I need to keep the discussion on the blog separate from the formal regulatory process here.

by Ray Pellecchia on April 15, 2007 5:06 PM

Ray - There's a difference between formal regulatory process and what we're trying to accomplish here. Clearly the folks that post here have knowledge and experience in the markets and, referencing the thread about late nyse/hybrid trade posting - there are enough examples and detailed explanations to warrant some inspection by you guys into your system (our goal) You state in your response to my email that you "checked around" and didn't find any problem in the stock that day. Interesting. What, exactly, did you check? Even today (the data does not appear to have been scrubbed/modified) a check of any data service (Bberg, et al) shows a massive number of prints by the NYSE clearly out of order. To clarify - I posted test bids on every single market that trades the security. It still printed en masse through. It is unquestionably a case of late "burst" reporting - I'll venture from a server that was playing catchup or perhaps a stuck queue. I don't need to file a complaint. I was merely hoping you'd notice what we all did (this stock was a perfect example) in order to improve the system. In terms of the explanation of away prints under the "T" flag. It happens quite a bit but I would characterize it as likely a market maker (I used to be one and did this all the time) printing straight to the tape because the quotes do not change during this time and it's definitely not a sweep or trade-thru. Traders just need to remember that odd away prints can still happen even under Reg NMS.

by Mike on April 18, 2007 4:28 PM

Mike -- In looking into your original note I checked into whether there were any system problems on the data or trading side that day in that stock, and could find none. But thanks for this note back; I understand your point better now, though unfortunately I don't have a good explanation yet. We're continuing to look at this issue and others, and will keep you posted.

by Ray Pellecchia on April 26, 2007 11:03 AM

I'm just commenting here b/c of the point made about correct sequencing of trades. I am tired of the "NAQS" prints going off all the time nowadays.

by Jon on June 15, 2007 10:54 AM

Comment on this entry

Forward this entry to a friend