While most of the world focused last week on scandals regarding the NSA’s multi-year phone call and email surveillance program, the high-frequency traders and those who service them were busy fighting back another controversial piece of news.
The SEC has launched an investigation into Thomson Reuters (TRI) , since the company got caught with its hand in the proverbial cookie jar earlier this month. Sharp-eyed data crunchers at Nanex, LLC, a market data services company, noticed a bunch of unusual selling just ahead of the release of highly anticipated manufacturing data from the Institute of Supply Management (ISM). Someone out there was dumping shares of SPY, an ETF that mirrors the S&P 500, as fast as their black box computers could place the sell orders – $28 million worth, as a matter of fact.
The Chain of Events
The trading was all concentrated in the last 15 milliseconds prior to the scheduled 10 AM release of the data to the rest of the world. But it was enough. Someone, apparently, had leaked the data showing some dismal factory numbers for the month of May. According to Nanex, some 30,000 SPY shares hit the market in the last 15 milliseconds. Additionally, 369 stocks took a hit, according to a report from CNBC.
Who is the most likely suspect? Thomson Reuters – the company that ISM stated had an exclusive agreement for a low-latency feed that could have given them a head start. Thomson Reuters confessed that they released the data early, giving their own clients a market advantage over other participants but blamed the early release on a clock synchronization error. On the other hand, ISM insists they never release anything before 10 AM and said that they even build in an extra second of fudge time, meaning that a clock problem on Thomson Reuters’ end would not explain the discrepancy.
ZeroHedge writer with the alias “Tyler Durden,” includes in a post some graphics that illustrate an even more egregious example of market data front-running.
First, there was a huge selling spree in the gold markets 62 milliseconds before a major data release. Then, showing a heat map for the E-mini market, Durden shows that markets briefly panicked – and traders yanked almost every drop of liquidity out of the game as they tried to make sense of the gold move (see below).
Then, markets quickly returned to ‘normal.’ However, if you think about it, every share that got dumped prior to the data release that was actually matched up with a buyer by the Nasdaq or other exchange matching engines, represented a fraud on the market.
The two things any trader should expect from a securities exchange, amount to transparency and fair dealing. Transaction speed is secondary – if market participants cannot expect fair dealing from the bourses, then transaction speed just helps enable their fleecing.
Anyone who had a buy order fulfilled during that time has a right to be annoyed. And the ISM example, particularly, illustrates the way large players rig the game to benefit their favored clients.
But expecting fair dealing from market institutions like Thomson Reuters only goes so far. Assuming no leaks occur to favored institutions prior to a scheduled release, in a world of trade execution measured in single-digit milliseconds, even that is not enough to level the playing field for the small-time trader. Why? Latency.
What Is Latency?
Latency refers to the time that elapses between the time one computer on a network transmits an information packet and the time another computer on the network receives it. There are a lot of variables, but two of the key variables are network throughput capacity (T-1 versus fiber-optic cabling, for example, or satellite transmission vs. 100 percent earth-bound mediums) and distance. The further the distance, the longer it takes for a signal to get to you – all other things being equal.
This wasn’t a huge issue until matching engine caught up with transmission technology. But now latency is a major issue from the point of view of high-frequency algorithmic traders: So much so, that some of them are willing to pay big bucks for the right to store their servers in a colocation facility in or very near to the exchange itself, purely to give their accounts a few milliseconds’ head start against other participants.
One study estimates that a 1 millisecond advantage over other market participants can be worth as much as $100 million to a major brokerage firm – and so they are willing to pay up for the privilege. Another study from February 2013 pegs latency costs for ‘manual’ investors as much as 15 to 25 percent.