SPY Closing Events of February 29, 2012 and July 29, 2011

Gaming the Last Millisecond of the Day

During the market close in SPY (and dozens of related ETFs) on February 29, 2012, a delay of about 1.25 seconds developed in quotes from Nasdaq relative to quotes from other exchanges. Execution trade prices from Nasdaq appeared synchronous with trade prices from other exchanges, which shouldn't be possible, since Nasdaq's quote was behind. We also noticed an anomaly in the trade condition.

The trade condition Form-T is used to indicate trades executed before or after regular trading hours (9:30-16:00 Eastern). What is puzzling is that the first Nasdaq trades marked Form-T don't appear for almost a full second after the close. Right along side the Nasdaq trades marked ISO or regular are trades with the same prices from other exchanges, but marked Form-T. We are used to seeing Nasdaq trades printing ahead of their quotes because it's so common - yet impossible unless routing is based on direct feeds instead of the SIP (Security Information Processor ~ CQS and UQDF). But we're not sure how the trade condition could get time-warped like that.

How can the Nasdaq trades between 16:00:00 and 16:00:00.841 not be marked Form-T, yet have the same prices as Form-T trades from other exchanges at the same time? It's as if the prices are not delayed, but the conditions are. What is going on here?

Update March 1, 2012

One possible explanation is that the Nasdaq trades between 16:00:00.000 and 16:00:00.841 actually executed in a millisecond or two before 16:00:00.000. That would make the quote delay an additional 841 milliseconds longer than we calculated from the quote price offset. This still does not cleanly explain how trades from other exchanges with similar prices were marked with the Form-T condition (indicating after close).

What if all the data we see between 16:00:00.000 and 16:00:00.841 actually occurred in the very last millisecond of the day? If you have a speed advantage, then wouldn't the last millisecond (or microsecond) become the most valuable time of the trading session? The problem here is that the network carrying SPY has a system capacity of just 1,250 quotes per millisecond for all NYSE, NY-AMEX, and NY-ARCA stocks. One stock could easily plug up the network. But if you are co-located and wired for speed (and there is no competent regulator around) , that problem becomes your advantage.

Another mystery is the source of the closing price of 137.02. We found trades with the condition Official Market Center Close on ARCA at 136.87 and Nasdaq at 136.54, but nothing so far that would indicate a close of 137.02.

In an electronic world, there shouldn't be any mysteries about basic audit trail analysis of trades and quotes from the SIP..

..or where the closing price came from.

Update March 2, 2012

We found the Trade that sets the close for SPY (see chart #10). It was an 800 share trade from the FINRA Trade Reporting Facility where dark pool, internalizer, and other off-exchange trades are reported. This trade was also at least 13 seconds late, and probably executed at 15:59:44 where other FINRA trade reports with a similar price occurred. How does an 800 share trade, that is obviously reported late, negate several million shares that traded at much lower prices?

Update March 6, 2012

We found (by using the chart on this page) another trading day with a very similar event: July 29, 2011. See 2nd chart set below.

Feb 29, 2012 SPY 5, 10 and 100 millisecond interval charts.
Shaded area represents Bid/Ask spread, circles represent trades. Note how far Nasdaq's quote gets behind quotes from other exchanges, yet Nasdaq Trades are not affected (except for the trade condition).

July 29, 2011 SPY 1, 5 and 100 millisecond interval charts.
Shaded area represents Bid/Ask spread, circles represent trades. Note how far Nasdaq's quote gets behind quotes from other exchanges yet Nasdaq Trades are not affected (except for the trade condition).

More Nanex Research