Title: Orbital phase-driven biases in Galactic mass constraints from stellar streams

Abstract: One of the most promising tracers of the Galactic potential in the halo region are stellar streams. However, individual stream fits can be limited by systematic biases. To study these individual stream systematics, we fit streams in Milky Way-like galaxies from FIRE cosmological galaxy formation simulations with an analytic gravitational potential by maximizing the clustering of stream stars in action space. We show that for coherent streams the quality of the constraints depends on the orbital phase of the observed stream stars, despite the fact that the phase information is discarded in action-clustering methods. Streams on intermediate phases give the most accurate results, whereas pericentre streams can be highly biased. This behaviour is tied to the amount of correlation present between positions and momenta in each stream's data: weak correlation in pericentre streams prohibits efficient differentiation between potentials, while strong correlation in intermediate streams promotes it. Although simultaneous fitting of multiple streams is generally prescribed as the remedy to combat individual stream biases, we find that combining multiple pericentric streams is not enough to yield a bias-free result. We finally show that adopting the two-component St\"ackel model does not fundamentally induce a biased mass estimate. With our full data set of two multi-wrap streams, we recovered the true rotation curve of the simulated galaxy within $12\%$ over the entire range of radii covered by our set of stars (10 - 176 kpc) and within $6.5\%$ between the 5 and 95-percentile distance range (23 - 109 kpc).