stb and tran analysis discrepancy

I'm having trouble understanding the results of a stb simulation I'm running over process corners and temperature for a basic switched capacitor active integrator.  There is a low frequency pole due to switch leakage/gmin setting depending on the process/temperature which shows up in both the loop gain and phase plots.  What I don't understand is why the phase plot starts at -180 or -90, the phase shift of an (ideal) integrator at DC should be +90 so I'm not sure why the stb simulation shows -180 in the midband.  I'm using ideal switches to short the input and output of the amplifier for biasing at DC and the switches open for the stb portion of the simulation (see the screenshot below).  The voltage source in the DC feedback path is used to precisely set the bias point of the amplifier output.  I looked at transient simulations for some of the corners and the results look great: no ringing, high gain (confirmed by the ratio of the output/input voltage step between phases) and the settling time of in the hundreds of time constants.  It looks like the system should be unstable from the stb results but the transient results don't confirm this, what am I missing?