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.
On Tuesday, a high frequency trading algorithm apparently spoofed the eMini futures contract. This illegal activity can manipulate a price higher or lower without actually executing a trade, as we saw in Panther Energy’s spoofing in the oil markets. Here in the eMini, the price rose about 1 point, or $50 per contract. Also of note in the chart below is the severe reduction of liquidity around the 10:00 release of Consumer Confidence data, contrary to pro-HFT arguments that liquidity is provided by high frequency participants.
Apple Inc. (NASDAQ:AAPL)
An excellent article from Fortune highlighted some recent studies exposing latency between exchanges which would allow the capture of absolutely risk-less profits through simple latency arbitrage. Professor Hendershott from UC Berkeley conducted research using access to high-speed trading technology granted by Redline Trading Solutions. Due to information delays, he was able to produce $377,000 in theoretical profits trading Apple’s stock on May 9th. One ticker, trading for one day, produced over a quarter million dollars in free money for anyone with fast enough computers.
Redline allowed him to use its “direct market access” — cables that run directly from exchange servers to its own. Redline’s server was co-located with that of BATS Exchange so that the “latency” on information and orders coming from BATS was cut down to barely one thousandth of a second. As a result, some of the quotes on public feeds such as the crucial “national best bid and offer” feed were a few milliseconds behind those Hendershott could see on his direct link with the exchanges. With a half-decent trading algorithm, Hendershott would have had ample time to buy Apple at a stale price with a guarantee that he could sell at a profit. Every couple of seconds. All day. Risk on the trades: zero.
The Procter & Gamble Company (NYSE:PG)
On Friday the stock of The Procter & Gamble Company (NYSE:PG) took a sudden and curious 5% dip in about 100 milliseconds. The curious part of the drop is the amount. It was just within the bands set by exchanges in their Limit Up Limit Down rules to prevent “extraordinary market volatility in U.S. equity markets.” Some may consider 5% in 100 milliseconds extraordinary volatility. Previously we’ve displayed how the NYSE’s LRP program is far superior to LULD rules used by other exchanges.
NASDAQ OMX Group, Inc. (NASDAQ:NDAQ)’s Weak Excuse
This past week NASDAQ OMX Group, Inc. (NASDAQ:NDAQ) released an explanation, based on preliminary investigating, for the 3 hour exchange darkness the week before. Essentially continuing in its immediate response that the fault lies with NYSE-Arca, the exchange found that large message traffic of over “26,000 quote updates per-port, per second as it attempted to reconnect” caused the SIP to malfunction. A glaring question resulting from this reasoning surfaces as we realize that this level of traffic is not too uncommon and has never before resulted in an exchange blackout.
since 2003, there have been 411 instances where a single stock (one port) had 20,000 or more quotes in one second and in 62 of those instances there were 26,000 or more quotes in a second. In almost every case, the exchange sending this high quote traffic was Nasdaq.
In fact, Nanex has cataloged every event and even produced this video showing one example of 28,000 quotes per second being sent by Nasdaq. So while it appears the SIP breakdown was caused by a glitch in Nasdaq’s software, causing an hour’s worth of old data to be resent as new data in only 3 seconds and therefore self-inflicted, their reasoning holds little water and only raises more questions.