The corrective weakness in bond markets ended at 10am after the JOLTS data (Job Openings and Labor Turnover Survey), which showed 5.249 million job openings (seems like a lot, right?) versus the previous month at 5.357.  The survey also includes hiring rates, and 'separations' rates (which is a fancy way of saying "people who stopped working for any reason, including but not limited to being fired, quitting, or ending a contract").  Separations rose even as hiring fell. 

Bottom line, not a good JOLTS report.  Even though it rarely moves the market (largely because it's very old data.  Today's is for June.), this one looks like it did.  When JOLTS does move markets, it's not uncommon to see it happen gradually like this as the data speaks to the slower-moving strategic trading types (as opposed to the accounts that trade the headline data as quickly as possible).

MBS were hovering at their weakest levels of the morning before the data.  Some of the earliest-pricing lenders were already looking at 3-4 ticks of losses day-over-day.  That's right on the risk threshold where negative reprices first become a twinkle in lenders' eyes.

Obviously, that risk is nil now that we've recovered all of the morning's losses, but if we happen to lose ground later today, we now have a good line in the sand to watch for reprice risk.  Falling below 103-25 in Fannie 3.5s would be where things get questionable.  100-22 is the corresponding level in Fannie 3.0s. 

Fannie 3.5s are currently back up to 103-31 and 3.0s are back up to 100-27.  10yr yields are at domestic session lows of 2.091.