On Monday, ISM non-manufacturing data was released at 10:00 and it appears as though another early data release may have occurred. As the chart below shows, there was a burst of trading in gold future contracts approximately 90 milliseconds before the official data was released. Unsurprisingly the high frequency trading machines were again correct in their bets that gold prices would drop.
Every Thursday the EIA releases storage numbers for natural gas. Every Thursday, the futures market for natural gas moves sporadically upon the 10:30 data release. Most of the pro-HFT argument centers around providing liquidity to markets. As the charts below show, liquidity evaporates before 10:30. This shrinking of size in the order books makes jarring price movements all the likelier.
Also on Thursday morning, the stock of Drew Industries (ticker: DW) seemed to be the target of some sort of algo experiment. In the pre-market hours from 8:38 until 8:47 a strange pattern occurred in the NBBO. As you can see in the chart, the Best Offer moved steadily from around $44 up to $440 while the Best Bid moved inversely from $44 to down near $0. The pattern takes a minute to run its course, then resets and repeats. This sequence is very odd, and may point to some software testing similar to what sparked the meltdown at Knight Capital last year.
On Tuesday, the BATS exchange shut down around 1:00. The notion of one of the larger U.S. exchanges shutting down in the middle of trading sounds chaotic and dangerous. Thankfully though, the folks at BATS handled it perfectly. In fact, according to Nanex, they couldn’t have done anything better.
We hope other exchanges take notice of how this is done. When BATS went offline, they first zeroed all existing quotes in all stocks so that the NBBO doesn’t end up crossed or locked from a stale quote.
The curious finding from the BATS shutdown was the reduction in quote spam across all NMS stocks during their outage. The thick red line in the chart below represents quotes per $10,000 worth of trades for the day BATS went offline. As you can see, the line drops quickly around 1:00 then begins to recover a little after 2:00 when BATS came back online.
Every week Floating Path looks to highlight some of the crazier examples of high frequency trading running amok in the markets. We typically do so with the help of data and charts from the good folks at Nanex. Eric and his team monitor and analyze market data like no one else we’ve seen and visualize it in a manner that allows most to understand a usually complex subject.