In another case of purely coincidental serendipity, three days ago Zero Hedge informed readers that the "NYSE Boerse [sic] has just announced its purchase of Kingsbury International Ltd., which surveys managers for the Chicago Business Barometer, also known as the company that hosts the Chicago PMI data, in order to bring PMI data direct to feed subscribers. Net result: expect even more market volatility at each PMI release, now that the market is not two but three-tiered, and consisting of regular HFTs, HFTs with access to the Deutsche Boerse feed, and everyone else." We concluded: "It is unclear if the ultra-speed, HFT friendly feed would be activated before its next release on June 30. That said, we will certainly coordinate with our friends at Nanex for any trading abnormalities, primarily in the critical ES futures, this Thursday at 9:42am, keeping a close eye on the tape, and indicating precisely when the tiered data release hits." Well, as promised here is the Nanex data. As expected, it's a stunner.
The shocker, however, resides not in the stock arena, but in what is now becoming the go to place for bulk frontrunning high frequency trading algorithms to chase what little volatility is left in the equity market: options, which, as previously noted, we now are confident will be the cause for the next big market wipe out.
Approximately 1/2 second before the 9:42 release of the Chicago PMI report, the option market exploded setting new records in quote rates, saturation, and delays. We have not yet determined why the equity market did not see a record explosion of quote traffic; rather it experienced the normal saturation/delay that happens all too frequently every trading day.
The electronic S&P 500 futures experienced a withdrawal of liquidity beginning about a minute before the release of the PMI number. At approximately 9:41:59.550, 1275 contracts cleared through 4 levels of the offer side of the order book. This coincided with the explosion in OPRA quote traffic.
The first image shows quote message rates for each of the 12 CQS data lines that carry data for NYSE, AMEX, and ARCA equities and ETFs in 2ms intervals. Notice how quickly activity drops after the peak compared to the OPRA images below it. Normally, options activity follows equity activity very closely.
The images that follow, show each of the 48 lines individually along with the total, so that you can clearly see the saturation events and estimate the duration and extent of the delay for each line. The flat-top areas you see on the charts are caused by something gating, or queuing the data. Since OPRA, like CQS, timestamps after data exits the queue -- right before it's transmitted to subscribers -- it is impossible to know the exact duration of these hidden delays, but 500ms to well over 1 full second is a conservative estimate. We believe OPRA capacity would have to increase at least 3 times, to 12 million/second, in order to avoid these significant delays. However, at those message rates, a significant number of quotes would have already expired before they even left the exchange networks. (for all source images, please go to the Nanex site)
Several individual OPRA data lines show gaps which we believe are exhaustion events. These quiet periods of no quotes are common, can last 20 milliseconds, and almost always follow a spike in activity. We have verified that there were no drops in the data and that the charts accurately show the quote traffic rates. Several lines, noteably #37, show a period of fluttering between a high rate and zero which seem to appear during times of severe saturation.
While it is impossible to determine if this is indeed a case of broad embargo breach, it is imperative that Kingsbury International and the Deutsche Boerse immediately announce are precisely what moment they releasaed the PMI data to i) subscribers of Alpha Stream, and ii) to subscribers of the PMI service, who up until now thought they were getting a bargin by frontrunning the general population by three minutes courtesy of a public embargo, and now seems are themselves being frontrun by almost 500 milliseconds.
Furthermore, if no advance data release is confirmed, can OPRA please explain what the reason for this bizarro frontrunning activity is as traditionally a massive burst of trading action in advance of news dissemination indicates something is terminally broken with the checks and balances in the system. While we know that is the case, with the aid of Nanex, we will continue exposing each and every act of public data frontrunning going forward until every last retail investor is permanently out of the market and central banks and primary dealers can throw the hot ponzi grenade amongst themselves.