A glitch in QQQQ leads to enormous mispricing on opening cross.

    Most of us don't give a second thought to what goes on behind the curtains when we're buying and selling mutual funds and ETFs. All we know is that we are buying a piece of an entire portfolio, whether it's the 200-plus stocks in an actively managed fund like Fidelity Magellan or the 500 stocks in the Vanguard 500 fund and the SPDRs (SPY) ETF.
    ETFs in particular are interesting. Because these funds (all index funds thus far) trade on the secondary market like a regular stock, they bring additional players into the game. Specialists, market makers and an exchange are all involved when you buy or sell shares, as opposed to the purchase of mutual fund shares, which is between you and the fund management company.
    Of course this structure brings with it myriad benefits, of which we're all well aware. You can trade ETFs in real time, with extremely low fees in a structure that can be more tax efficient than a traditional mutual fund. These benefits have not been lost on investors, because ETF assets have been the fastest-growing segment of the index market, which is itself the most rapidly growing segment of the financial services industry.
    All of these additional players, however, and the growing strength of indexing have led to some interesting quirks. One, a trading quirk in how the opening price is set for the Nasdaq 100 ETF (QQQ), recently rebranded as the QQQQ when it moved from AMEX to Nasdaq, cost a few shareholders about $20 million, as recently reported online ("Free Money?" www.indexuniverse.com, January 12, 2005).
    For reasons explained below, the opening price of Nasdaq-traded securities is determined by what's called a cross-when investors and arbitrageurs are able to bid on the price of shares in quantity prior to the open. Typically, these prices converge towards a fair market value.
    But on December 17, due to a heavy imbalance of trade and insufficient arbitrage activity, the QQQQ opening cross diverged significantly from the underlying value of the exchange-traded fund (ETF), which happens to be one of the most liquid and heavily traded equities in the world. The official opening price of QQQQ on that day was $37.21, a full $2.40 (or 6%) below the volume-weighted average price for shares trading immediately after the open, which at $39.61 was very close to the underlying NAV of the QQQQ at the time.
    The net result was that two very large investors, and many smaller investors, lost a lot of money on the cross and in its immediate wake, while more than 800 buy orders-a few of them for very large buyers-were able to reap enormous profits in a short amount of time. Of course it is unclear who lost and profited, but we can tell from publicly available trade numbers that the five biggest buy orders were 311,700, 311,500 (twice) and 250,000 (twice).
    The profit made on each of those buys was $784,080, $747,600 (twice) and $600,000 (twice). The buyer(s), if interested, could have simply sold the stock once regular trading began and pocketed the $2.40 per share difference, less any trading costs.
Not bad for two minute's work.

What Actually Happened
    The chart on the preceding page illustrates exactly how the December 17th opening cross for QQQQ unfolded. For more detail on the logistics of how the Nasdaq opening and closing crosses work, see the www.indexuniverse.com articles covering the Nasdaq opening and closing crosses.
    Basically, as you can see, on December 17 there was a huge imbalance between those who wanted to sell (the blue bar) and those who wanted to buy (the pink). The pink and green lines going across the chart indicate the near and far price points where all of the buy and sells would meet at each moment leading up to the open. These lines should and do converge, as market participants get frequent updates on the balance of trades-every 5 seconds during the last part of this time frame. In theory, those green and pink lines should also converge with the black line on top, which is the open market inside bid.
    The final crossing price, which is also the official opening price and is the point at which the green and blue lines end up on the right side of the chart, was in this case at $37.21. This was about $2.40 below the inside ask and bid prices, which were on the NAV and were where the market traded after the opening cross. Bear in mind that you are looking at just two minutes of activity, with market participants getting an update on the imbalance as often as every five seconds in the last minute before the opening cross (which is also the official opening price) is set.
    The peculiar issue dramatically illustrated by this chart is that in theory, the colored lines in the middle of the chart should continue their upward path until they converge with the black lines at the top of the chart (or should very nearly do so). One other peculiar and unexpected development we can see in this chart is that in the very last five-second interval of trading, the amount of shares that would actually be paired at the market value actually drops-indicating that some of the potential buyers of these discounted shares may have been spooked by the peculiar market, or perhaps were altering their bids to push the spread further. The net result was that to meet the 2.7 million shares with no pair, the price had to meet at a level that was $2.40 below the value of a share, which was $39.61, according to where the inside bid and ask were as represented by the black lines at the top of the chart.
    According to the Nasdaq Stock Market (which was extremely cooperative with us in our investigation of the event, and supplied us with all of the specific numbers in this article), in the 15 trading days since December 17, the difference between the official opening cross price and the first unlocked, uncrossed consolidated trade was 1.7 CENTS ... a long way from a $2.40 departure from the NAV.  Even in the days before December 17th, just after QQQ had moved to the Nasdaq, the spread was only 3 cents.
    The Nasdaq Stock Market was also quick to raise a number of factors that contributed to ensuring that this sort of event is likely to be an incredibly rare one: 1) The open of December 17 was the time of index derivative expirations. The QQQQs are not included in any index derivative products and therefore were not expected to trade heavily, so many index and arbitrage traders following the opening that day were not looking at the QQQQ. 2) The QQQQs do not typically trade heavily on the open, so ETF traders were not focusing on the opening cross. 3) December 17 was quad-witch Friday. 4) Market participants were engaged in the year-end harvesting of capital losses, which led to an increase in trading activity (particularly selling) 5) December 17 was the date on which the annual rebalancing of the Nasdaq-100 index took effect.
    All that said, clearly something like this should never happen, and I am reassured that the Nasdaq has taken steps to correct the issue, just as it worked closely with index investors to launch an opening and closing cross to promote more reliable and obtainable opening and closing prices in the first place.

A Solution?
    Indeed this may well have been a once in a lifetime anomaly. We're still concerned that something like this can even occur if there is no one watching over the trading in a particular corner of the market. Although now, clearly more participants are watching, we think that it might be possible to learn from this event and possibly implement some kind of control at the cross. In the same way that stocks are capped from moving more than 10% now on a cross on the Nasdaq, maybe it makes sense to have a limit by which the cross (and official opening price) can vary from the inside bid or NAV in the case of an ETF.
    An ETF is different than a regular stock, because it actually has a stable underlying value in its constituent stocks. And the price the ETF should never be able to trade at this large of a variation from the NAV. Perhaps a control could be put in on ETF trading that ensures that a cap (50 basis points?) of the variation from the NAV or inside bid. When this cap is reached, trading is cut off, the cross that can be made is made at that level and the unbalanced shares are sent back unexecuted at the opening (or closing) cross in the same way that stock shares are when the 10% price movement level (formerly 20%, through the Russell rebalance last June) is reached.
    In any event-though the odds of any of these issues ever affecting you when you purchase an ETF are slim to none-it is often interesting to get under the hood and see what is going on with these products. Even now, 12 years after the launch of the very first ETF, these funds are still being refined to be made even better.

Jim Wiandt is president of Index Publications LLC, editor of the Journal of Indexes and publisher of IndexUniverse.com and Exchange-Traded Funds Report.