The Dangers Of Machine Readable News

The above chart is not the May 6th flash clash. It is an intraday chart of Constellation Energy from yesterday. If you ever wanted an example of how machine readable news and trading bots can wreck a stock in a matter of minutes, then look no further than the above chart. Let’s reconstruct the events as they happened:

11:58 – Headline crosses that “US sues to block Excelon acquisition of Constellation Energy “. CEG is trading at $38.93 at the time of headline.
12:03 – CEG is halted due to a circuit breaker popping when the stock drops 10%. CEG last trade before the halt was $35.03
12:08 – CEG reopens and shoots straight back up to $38 on heavy volume.
12:10- Headline crosses saying “US settles with Excelon”

In the time span of 12 minutes, two distinct headlines ripped a stock up and down 10%. Some may say that the circuit breaker did what it was supposed to do and “allowed cooler heads to prevail”. And they would be partially correct. But the question remains, how does a stock drop that quickly and spring back up that quickly? The news was barely disseminated and the stock had already dropped 10%. Bids disappeared almost instantaneously and no doubt stop-loss orders kicked in. Volume weighted participation algos also no doubt chased the spike down as volume soared in the 5 minutes of trading before the halt.

The way this news was released is also very questionable. To announce a lawsuit and then a settlement 12 minutes later is not fair to investors. The companies involved should have requested a trading halt prior to the news announcement. But the fact remains that the news was disseminated immediately to the trading bots since it is now available in machine readable format. There was no time for a human to intervene to try and make sense of these headlines. The trading bots that subscribe to machine readable news services interpreted the news as bad and pulled out of the market. It is kind of ironic that earlier in the week Nasdaq announced that they bought a machine readable news company named RapiData. The WSJ said that “such machine-readable news is used by sophisticated trading firms that pull in signals from market prices and other sources to inform rapid-fire buying and selling of securities and derivatives contracts.”  They described Rapidata as:

RapiData packages and sends out figures from the U.S. Bureau of Labor Statistics and Treasury Department, according to the company. Staff enter data into RapiData’s systems, which check figures for accuracy and disperse them electronically when the government sends out the figures on its own. The firm uses high-speed fiber optic corridors and strategically placed servers that communicate the figures to trading systems in sub-second speeds. Such machine-readable news is used by sophisticated trading firms that pull in signals from market prices and other sources to inform rapid-fire buying and selling of securities and derivatives contracts.”

Not only do we now have a market that can whip stocks up and down 10% at the flicker of a headline, but now we have an exchange that is getting into the business of machine readable news. What ever happened to the real function of an exchange? Aren’t exchanges supposed to treat all investors fairly? Are exchanges that desperate for revenue that they have to get into the shady business of machine readable news so that they make sure that their best customers continue to have an advantage over everybody else? Maybe the demutualization of exchanges was not such a good idea after all. Their constant quest for profit places them in a conflicted position and makes their ability to treat all investors fairly very difficult.