Unreliable ARCA executions have given me a new appreciation for the NYSE

Discussion in 'Order Execution' started by fbell50, Sep 23, 2008.

  1. fbell50


    On 9/19/08 I had an order working to sell IYF, an ETF listed on ARCA, on the ARCA open. I received a fill of 94. This was 34.9% higher than the previous close 69.66. An extraordinary move, but 9/19 was an extraordinary day marked with numerous dramatic opens. Many stocks including ETFs had equally large moves on the open. MS, a component of IYF, opened up 42%, many other components of IYF such as BAC and GS opened up 25% or more. 94, up 34.9%, even if a little high, seemed reasonable.

    At about 7:50 pm ARCA busted the opening trades on IYF and several other ARCA stocks. Prior to this I received no notification that my trade might be busted. My broker claimed they did not learn of any problem until ARCA busted the trades.

    When I called the ARCA trade department I was told that it was a volatile day and there were many execution problems that morning. The ARCA rep said that ARCA was not the only exchange with execution problems and that they had emailed brokers warning that some trades were being questioned.

    The ARCA rep would not tell me exactly why the IYF open had been busted but I believe I can guess. In order to execute the eligible sell orders, the ARCA opening cross algorithm had to move up the ask list until it reached a price that would execute all the sells. Since the ARCA open is completely automated and liquidity thin a significantly higher price was reached than would be if the cross had been carried out on the NYSE. On the NYSE as the price rose, additional new liquidity would have been sought from the outside keeping the price reasonable.

    This raises several issues:

    1) ARCA executions are not reliable and it is not always obvious when a trade will be busted. EWS, also an ARCA ETF, opened up 35.5% and was not busted.

    2) ARCA does not address potential busts in a timely manner. Busting trades after market close is not timely. Sending emails to brokers that trades might be busted is usually of little value.

    First, your broker is unlikely to have a mechanism to broadcast this to customers. Second, it will not always be clear whether the trade will be busted. Third, even if you know that a trade might be busted and want to take action assuming it will be busted, it might not be possible to take action. In my case it would have required borrowing enough IYF to close my position, which might not be possible.

    3) The ARCA rep claimed all exchanges had problems on 9/19. This is true, but in proportion to their listings ARCA had substantially more. By my count of the approximately 4000 most liquid issues I follow, of the opening crosses 1/249 AMEX, 4/1687 Nasdaq, 17/1878 NYSE, and 17/188 ARCA were busted. The opening crosses in a full 9% of the ARCA listings were busted.

    4) How can you trade on an exchange where the executions cannot be trusted, especially when the busted execution was not due to a systems error such as a software bug? My quote service provides every quote and trade. Extreme ARCA executions are much more common than for any other exchange, even the Nasdaq’s fully automated system.

    5) A busted trade is an opportunity lost. I wanted to sell IYF on the open and would happily have accepted a price exactly in line with the underlying stocks. That opportunity was denied me. My exit was at a price substantially lower than what that price would have been.

    Until this happened to me I just sort of ignored ARCA’s flaky executions. Now I more fully appreciate the value of trading at a reliable exchange.
  2. The ISE, EDGX, NSDQ, NYSE, and BATS also busted a lot of trades from the open later that day...

    So, it's ok that you're pissed at the market for this BS...
    but if you're pissed at all the ECN's and you're not sending them bussiness where are you gonna trade? TRACK or DATA?
  3. If there is a real problem, it should be solved by adjusting prices, not by busting trades. Busting trades creates totally unpredictable losses extending far beyond those resulting from the trade that was busted.
  4. The Difference Between NYSE and Nasdaq: Who Is Master, Man or Machine?

    By: Ray Pellecchia
    File Under: NYSE, NYSE Arca

    Here's a story that has not been reported to date: Last Friday's unprecedented trading activity revealed more about the difference among markets than any day since, perhaps, 19 Oct. 1987.

    Let me explain.

    Recall the events of last Friday. It was crunch time for U.S. markets, right from the get-go. It was the quarterly expiration of stock-index options and futures. Stock prices exploded upward at the opening, driven by the announcement that Washington was working on a huge financial bailout. And the SEC had just announced a new ban on short selling in approximately 800 issues, which initially resulted in confusion among traders, many of whom pulled their bids and offers until they could sort out whether they could short, whether they were exempt, which stocks were shortable and which weren't.

    All of that helped produce record trading volume on the NYSE in the first half hour, the first hour, and for the full day. In our opening auction alone, NYSE executed 744 million shares, or 98.8 percent of the consolidated opening volume.

    The electronic markets, too, were doing a lot of business, but there was a difference. Nasdaq had to cancel more than 1,000 trades in NYSE-listed issues in the first 10 minutes of trading. For the full day, Nasdaq would go on to cancel 11,000 trades in stocks listed on itself, NYSE and the Amex.

    The number of NYSE cancels for the day: none. Zero, zip, zilch. Nada.

    Here's why. On the NYSE, there are people with machines (algorithmic trading tools), not just machines. So when there are all buyers and virtually no sellers, or vice versa, there are people who risk capital on the contra side, people who bring in customers on the contra side, and a process that injects some time and transparency so that the market can find equilibrium and discover a real price.

    In contrast, on solely electronic markets, when there are all buyers and virtually no sellers, there are only computers, doing what they are designed to do: matching bids and offers quickly. When the market is all on one side, you're going to get a market-clearing price, and it's very likely you're going to be extremely happy or extremely unhappy.

    Last Friday, there were thousands of such trades taking place in electronic markets, and those markets got together to agree to cancel opening trades that took place 20 percent or more away from the previous day's close. It wasn't just Nasdaq but all electronic markets, including our own NYSE Arca, as well as Bats.

    That said, if you're an issuer, where do you want to be listed: with the exchange group that gives investors the choice between trading on markets that are fully automated (NYSE Arca) or high-tech/high-touch (NYSE); or the exchange group that gives your investors no choice? It sure came in handy last Friday to have the two models. And Friday was only a dramatic example of what happens all the time, albeit on a smaller scale.

    The cancels were only part of it. Look at the trades that weren't cancelled. Some 327 of Nasdaq's 1,090 opens were outside the full-day NYSE trading range. Of these, 26 were in the top 100 NYSE-listed stocks by share volume. Nasdaq’s average open was more than 14% away from the NYSE crossing price.

    For example:

    • Nasdaq opened AMR at $12.45, less than 30 seconds before the NYSE open. The consolidated high between the Nasdaq opening cross and the NYSE open was $12.39. NYSE opened AMR at $12.15, and in the next 3 minutes, it traded only as high as $12.28.

    • GE opened at $29.12 on Nasdaq, less than 24 seconds before NYSE opened it at $28.80. Nasdaq traded more than 205K shares in its opening cross. NYSE auctioned 16.1 million shares. In the three minutes after NYSE opened GE, the high was $28.85.

    • Deutsche Bank opened at $87.00 on the NYSE for 42,100 shares versus $90.00 on Nasdaq for 525 shares. The full-day NYSE high was $88.13.

    • Morgan Stanley opened on 4.1 million shares on the NYSE at $32.04. Nasdaq crossed 39,929 at $33.06. The full-day NYSE high was $32.20.

    • Build-A-Bear Workshop opened on 133,300 shares at NYSE. On Nasdaq, BBW crossed 100 shares at $7.26. The NYSE low for the day was $7.90.

    • Nasdaq crossed 1,000 shares of AGM at $14.78. The NYSE’s open of 90,400 shares was at $17.55, and the low of the day on NYSE was $16.67.

    So if you want it fast, or you want it right (and still pretty fast), NYSE Euronext offers you the choice of both. The other guys, not so much.

    Looking across the market, you can also compare how the stocks traded in aggregate, immediately after the opening. We examined the NYSE opening auction price and volume-weighted average price (VWAP) during the ensuing 10 minutes versus Nasdaq's open in NYSE issues and the VWAP for the 10 minutes after it crossed NYSE stocks. In the two minutes after the NYSE opens, the median VWAP differential to the NYSE open price was 15 basis points (bps), vs. 180 bps on Nasdaq.

    In other words, the price variation following NYSE's opening was 1/12th that of Nasdaq's opening in NYSE stocks.

    A hat tip to my colleague Steve Poser for the analysis in this post.

    I don't intend this to diss electronic markets. Again, they did exactly what they were designed to do, and that's exactly what some people want markets to do. But I do intend to underscore something that is often misunderstood and misstated: The real struggle in markets is not man versus machine; it is machine versus man-with-machine. Last Friday demonstrated the value of the latter, here at NYSE.

    There is a clear difference between NYSE Euronext and Nasdaq OMX, as illustrated last Friday, and that difference is choice for issuers and their investors and traders, and that difference is people. Even in this day of automation, choice and people really can and do make a positive difference.

    Sorry for the long-winded post. Have a good weekend.

  5. fbell50


    If their response was really to cancel opens 20 percent from the previous close then I really feel screwed. Many major components of the IYF opened 30-40+% over the previous close. The underlying index might well have opened up 25%, not that far from where IYF actually opened. To use an arbitrary rule like this is very poor form.

    There was an article in todays WSJ on this also.
  6. fbell50


    ARCA Targets 'Erroneous' Trades Problem


    "The unit of NYSE Euronext has started to incorporate its competitors' depth-of-book data into its own depth-of-book feed. That will allow it to route orders to more price levels than is required by Regulation NMS.

    That Securities and Exchange Commission rule requires market centers to route unfilled orders to their competitors. But they only need to route to the markets' best prices.

    Under the Arca initiative, the exchange will also route to the second-best, third-best, etc., prices. That will give the competing markets a chance to refresh their top-of-book prices, thereby preventing customers from executing at wildly different top-of-book prices."

    This is a good step. I'm not sure if it will apply to the opening cross, but it should certainly reduce and perhaps eliminate Arca's crazy intraday fills.

    I've always thought Reg NMS should apply to depth-of-book, not just top.
  7. I wonder how this order will differ from the more costly proactive Arca order.