Whoops! Bonds get jobs wrong at the last minute
The surprisingly positive October jobs data that was released at 8:30 a.m. EST hurt the bond market, but it must have been especially painful for those who bought bond futures seconds ahead of the report. At 8:29 a.m. EST, five-year note futures soared to a nearly five-month high—before losing all those gains, and then some, to hit a three-week low. Ten-year note futures similarly moved much higher before dropping.
The move was so powerful that it led the CME, the exchange on which Treasury futures trade, to automatically pause trading. CME Group told CNBC.com that a "velocity logic event" was triggered, meaning that the market was automatically paused because of an extreme move. Not only was the five-year Treasury note paused, but the 30-year Treasury bond and the 30-year Ultra Treasury bond were paused as well. The 30-year was paused first, starting at 8:29:57 a.m. EST and ending at 8:30:02. The five-year note pause began at 8:30:01, and ended at 8:30:06.
The 204,000 jobs created in October, which was well above expectations, fostered perceptions that the Federal Reserve will reduce its $85 billion monthly bond-buying program earlier than it otherwise would have. As a result, bond prices dropped, and Treasury yields rose.
But at literally the last minute before the report was released, five-year note futures advanced to the highest intraday level since June 19.
Interestingly, the move was not the result of one massive buy order.
"It was lots and lots of trades," noted Eric Hunsader of Nanex.
Jim Iuorio of TJM Institutional Services admitted that if the last-minute move took the market in the right direction, he would be suspicious.
"My first thought is that someone had the number, but that was clearly not the case in that instance," Iuorio said. "It was probably someone covering a position right before the print."
But Brian Stutland of Stutland Volatility Group still says the move was most likely an instant reaction to the unemployment rate. Pointing out that the rate rose to 7.3 percent from 7.2 percent even as a large number of jobs were created, he says the move smacks of an algorithm-based trade gone wrong.
"All code is written based on if/then statements," Stutland said. "This one probably said, 'If unemployment is greater than 7.2 percent, then buy bonds. And if NFP comes in less than estimates, then buy bonds. Otherwise, sell bonds.' Clearly, somebody got the number early, and they probably got unemployment first."
If one was just looking at unemployment, then one would have assumed that job growth was weak, which would have be good for bonds.
"I'm assuming when the print came out at 7.3, their algo model just triggered a 'buy bonds' order," Stutland said.
After all, Stutland reasons, "It's crazy that there were a flurry of trades as the number is coming out. The only way that can happen is if people are trading before the number."