Now I see! The NASDAQ Tape has a challenge at hand. The realtime stream I am using shows out of order reporting. The investment FAQ sheds light on the problem:(http://invest-faq.com/articles/trade-trust-tape.html)
Subject: Trading - Can You Trust The Tape?
Last-Revised: 10 July 1999
Contributed-By: John Schott (jschott at voicenet.com), Chris Lott (contact me)
Considering that there is big money involved in every trade, it is no wonder that a great deal of effort is made to insure the accuracy and completeness of each day's trading records. Yet despite this effort, there are cases where the trading tape you see on your computer, intraday charts, and in end-of-day data is not really telling a totally accurate story.
To settle each day's trading obligations (shares and/or money), each brokerage maintains a large "back office" function to ensure that each trade is accurately recorded and reported. In fact, months after, Standard and Poors publishes large reference volumes that list the official day's prices (Open,-High,-Low,-Close) and volume for each security traded on the NYSE, AMEX, and NASDAQ. Yet, the contemporaneous data you get from your Internet or other data provider may not reflect just what happened on any given day.
What can go wrong with the data? The answer is that a variety of factors, some of them mistakes, can put bad or misleading data into the stream. Consider the following cases.
After-hours trading
Transactions after hours (trades marked .T and "as of" trades) are generally not included in the price and volume information that is published daily. On the NASDAQ, volume data for after-hours trading is integrated into the statistical record next day with a 24 hour cut-off. Price data for after-hours trading is not integrated into the statistical record. Volume data reported outside of 24 hours and price data are recorded for surveillance purposes only.
Out of order reporting
On the NYSE and AMEX, there is only one specialist to report orders. On the NASDAQ, the floor is spread electronically over the world. So time stamped execution reports don't necessarily flow into the reporting systems in order. Sometimes there is an advantage for participants in delaying a report beyond the exchange-mandated minimums - for example, when someone is urgently trying to move a big block quietly. But most of the problems are simply due to the chaos that is the exchange day.
Stocks trade everywhere - on multiple worldwide exchanges, on electronic exchanges, at brokerage houses, and if two of us want, behind the local hamburger joints just after the 2am close. Many years ago, when this diffusing trend started, the NYSE made it a rule that any trading by any member firm had to be reported on the exchange even if the trade was executed elsewhere. And that rule applies today. So the Merrill Lynch office in Tokyo, Rome or London can handle a trade on one if the local markets in IBM, while the traders in New York are still sound asleep - and report that in hours (days) later.
Eventually those trades, and others crossed in local offices of exchange members, filter onto the NYSE tape at some time during the trading day. This would also be true of trades crossed by the Merrill Lynch office in Dallas during NYSE hours. Those trades make the tape sometime - but not always in order of trading or nearly in real time. And these trades may appear potentially outside the boundaries of the exchange-mandated maximum delay.
Trades in Nasdaq listed securities by foreign broker/dealers that are not NASD members are outside NASD/Nasdaq jurisdiction and would not be reported except if they involved some organization that had a trade reporting requirement under U.S. securities regulations. Some firms exist specifically to provide the large trader with discrete private placements which largely go unreported.
If you are confused, consider the poor specialist who arrives early only to find a variety of trade reports from Tokyo to London that don't match yesterdays prices nor the orders on his book - where do you open the stock? (See the article "Trading - Opening Price" elsewhere in this FAQ for more discussion of that issue.)
Errors do happen
If you every get a chance to see a live exchange ticker you will get to see the errors, too. Sometimes it is merely a misplaced trade reported way out of order. Perhaps it is an incorrect price or volume reported later as a correction. And then there are trades that just didn't happen for one reason or another - cancellations, repudiations, double fills, etc. They show up on the ticker, but some information gathering systems have no way to back them nicely out of the days activities. Some are not discovered until days later in the back offices.
Simple data entry errors still happen. Looking at an interday chart, one sometimes sees a single transaction far off the run of contemporary trades. Quite often the offset is $3 or $30, which is a clear signal that someone hit the wrong row of keys on a numeric key pad. Those errors show up in the interday charts all the time and often make the end-of-day quotes.
Even the floor traders get involved. When four or five people are competing for a specialist's attention, it is not hard for several people to hear the specialists "Done 500" as a fill of their order. So two orders become one or one becomes two executions. Naturally they all get corrected eventually - but does the tape ever show it?
Is volume really volume?
On the NYSE and AMEX when the specialist crosses an order and reports 1000 shares traded, we all assume that this means 1000 sold and 1000 bought (even if one party to the trade is the specialist himself). But there are complaints that NASDAQ reported volume may be far higher than the actual public trading. It is likely that this is true given the multiple competing market makers, most of whom actively trade for their own accounts. Sensing a trend, such a market maker may sell stock not owned or scarf up offered stock with the intention of laying off the stock on his competitors later - something the NYSE/AMEX specialist really can't do. If you watch intraday volume, you'll occasionally see such trading pairs pass across the tape with a few minutes separation - some may represent real trading, some merely various forms of market maker transfers.
Teasing the market
Technical analysts look for breakouts and other signals in their data. And the wolves on Wall Street know that. Occasionally they have a chance to push a few trades through to tip an indicator one way or the other. Often this happens near the end of a quiet day. Considering the spread, merely whether the last trade of the day is on the buy or sell side is often enough to bias the day's technical indicators. Recently I tried a $12 experiment on a NYSE stock that had held one price for almost six hours of NYSE trading. I wanted to see if the prevailing executions were on the buy or sell side. My 100 share order 1/8th point off that price brought a quick day-ending burst of trades - at successively different prices. Someone with real malevolence could do even more to trigger a technical move.
A dramatic example of off-exchange trading occurred on 26 Feb 97. After a 17-month battle, noted investor Carl Ichan sold off his entire 19.9-million share holding of RJ Nabisco Holdings (RN). He did this in an after-hours deal with Goldman Sachs at $36.75, a $1 price concession from that day's close. It is unknown if Goldman Sachs held the block for eventual distribution or acted for another firm. Trading was 2.4M shares on 26 Feb and 4.6M and 3.3M on the following two days, respectively, likely due to other arbitragers moving out of the stock. Interestingly, the stock price held, closing only 1/8th below the deal price. So this block never showed up on the tape nor in your TA program's data base. Although this transaction became public knowledge via a timely SEC filing and extensive press coverage, other large block trades may be effectively masked from public view.
Perhaps there is only one real lesson to be gained from understanding these and other forms of data inaccuracies that can creep onto the tape. It is that technical analysts should not regard all reports on the tape as gospel. |
