My colleague Eamon Javers has a fascinating story that I recommend market watchers read
Eamon noted that when the Federal Reserve released its statement on interest rates on September 18, some traders in Chicago appear to have received the information a bit before anyone else. Specifically, they began trading in Chicago-based assets (including eMini futures) before anyone could have been aware of the Fed decision.
We are talking about a difference of several milliseconds. But information takes seven milliseconds to travel from Washington to Chicago, Eamon noted, and there was buying in the eMini futures market in Chicago just three milliseconds after the 2:00 p.m. ET release, according to data provided to Eamon by Nanex. That's faster than information could have traveled between the two cities.
Eamon reported that Fed officials have contacted news organizations to discuss procedures for the release of sensitive information. In other words, they're trying to make sure everyone followed the rules.
Did someone know something early? In conversations with several traders, there possible explanations put forward:
- Someone cheated. A trader or traders--somebody outside the Fed press room--received the information beforehand and programmed their clock-synchronized servers in New York and Chicago to trade just after the news hit. Possible, but unlikely.
- A clock synchronization issue. There can be drift between the clocks of one firm, like the Fed, and others, like traders. If, for example, you send something out at exactly 12:00:00:00 noon on your clock, it is possible your time is off by several milliseconds or more. So timestamping when something was sent may not be the same timestamp that someone else has. This is possible, but also unlikely. Number one, I am told by traders who know a lot about clock synchronization that a lag time of seven milliseconds is unusually large for a clock synchronization issue. Second, if the Fed had released information late or early, there would not have been a reaction at the same time in Chicago and New York, as appears to have happened. The reaction would still have been about seven milliseconds later in Chicago than in New York.
- Someone in the Fed newsroom transmitted information before it was embargoed, but only to internal servers that allowed it to be released from multiple points at the same time the embargo was lifted. This seems like the most likely explanation, because it would explain why it looks like information was trading at roughly the same time in New York and Chicago, for instance.
What to do? Assuming that the final explanation is the correct one, the simple answer is to end the practice of releasing the information to a select group beforehand on an embargoed basis and just release the data to everyone at the same time.
But, as my colleague Bill Griffeth noted, this opens up another can of worms, since the purpose of an embargo is to make sure reporters have time to report the story correctly!