Space coronagraphs are projected to detect exoplantes that are at least 10 10 times dimmer than their host stars. Yet, the actual detection threshold depends on the instrument's wavefront stability and varies by an order of magnitude with the choice of observation strategy and post-processing method. In this paper the authors consider the performance of the previously introduced observation strategy (dark hole maintenance) and post-processing algorithm (electric field order reduction) in the presence of various realistic effects. In particular, it will be shown that under some common assumptions, the telescope's averaged pointing jitter translates into an additional light source incoherent with the residual light from the star (speckles), and that jitter "modes" can be identified in post-processing and distinguished from a planet signal. We also show that the decrease in contrast due to drift of voltages in deformable mirror actuators can be mitigated by recursive estimation of the electric field in the high-contrast region of the image (dark hole) using Electric Field Conjugation (EFC). Moreover, this can be done even when the measured intensity is broadband, as long as it is well approximated by an incoherent sum of monochromatic intensities. Finally, we assess the performance of closed-loop vs. open-loop observation scenarios through a numerical simulation of the Wide-Field Infra-Red Survey Telescope (WFIRST). In particular, we compare the post-processing factors of Angular Differential Imaging (ADI) with and without Electric Field Order Reduction (EFOR), which we extended to account for possible telescope rolls and the presence of pointing jitter. For all observation parameters considered in this paper, close-loop dark hole maintenance resulted in significantly higher post-processing accuracy.