Every week Floating Path looks to highlight some of the crazier examples of HFT running amok in the markets. We typically do so with the help of Nanex which monitors, analyzes, and visualizes high-frequency trading market data.
TVIZ
This ETF from VelocityShares is one of those obscure products that doesn’t see a lot of trading volume. It’s a prime candidate for the NYSE’s new incentive program tailored to high frequency traders. The fund “seeks to replicate, net of expenses, the returns of twice (2x) the daily performance of the S&P 500 VIX Mid-Term Futures index.”
On Tuesday, the quoting in VelocityShares Daily 2x VIX Medium Term (ETN NYSEARCA: TVIZ) was subjected to an odd HFT algorithm that produced huge bid/ask spreads in a steady, repeating pattern for about ten minutes. During the time, no trades were executed as the best ask gyrated from about $56 to $90 every second. The three charts below capture the algo wildly quoting TVIZ beginning with the entire event, and then zooming in to a 30 second and 1 second level.
Jobs Data Reaction
On Friday the all important monthly employment report was released at 8:30. Upon this release it wasn’t apparent that anyone had prior access to the data as initial trading was not in the correct direction. Although, as Eric Hunsader from Nanex mentions, maybe that was actually intentional.
Either a large early bet went horribly wrong, or a cleverly designed fake-out strategy was unleashed on the many high frequency trading algos that infest our markets. Most algos designed for speed will shoot first, and ask questions later: that is, they will immediately follow a trend, because being fast allows them the flexibility of dumping the position on slower algos and humans.
The chart below shows trading activity in all futures contracts, note the huge spike just 700 milliseconds before 8:30. That would be a lot of risk to take less than one second before the most market sensitive data point is released, especially considering how risk-averse HFT actually are.
As you can see in this chart, the price of gold initially swings $10 in the wrong direction, before rocketing $25 off it’s lows in only a few seconds. Is this some type of algo war we just witnessed?
SIP Down Again
With the sting of a 3 hour system-wide shutdown still fresh at the NASDAQ OMX Group, Inc. (NASDAQ:NDAQ), this past Wednesday the same piece of market infrastructure failed again. The Securities Information Processor broke down for a series of ticker symbols for almost 10 minutes, however just like before, trading continued on unaffected for those HFT firms using co-located computers and direct data feeds. Nasdaq issued the following statement, completely devoid of their infractions of Reg NMS allowing information asymmetry for those 10 minutes.
A preliminary review of the incident found that the SIP experienced a hardware memory failure in a back-end server at 11:35 a.m. ET, affecting price quotes for a limited range of Tape C securities (PC through SPZ). The system successfully failed over and operated normally to disseminate price quotes.
The chart below of message traffic clearly shows the outage on the effected portion of data (SIP Tape C, Line 5).
The following chart for SNDK is an example of the information asymmetry the Nasdaq officials claim to be so worried about. While any participants relying upon the SIP were in the dark as the NBBO froze, those directly connected to exchanges were able to trade on, and assumingly take advantage of firms with no information. Nanex has compiled a list of those tickers most effected including how many trades were executed during the outage.
