Blood speckle tracking has shown potential for solving the angle-dependency limitation in color flow imaging. However, as clutter filtering is still Doppler-based, flow velocities at near-perpendicular beam-to-flow angles can be severely attenuated. It is shown that the clutter filter also alters the speckle appearance through a decrease in the lateral imaging bandwidth, leading to poorer lateral resolution and thus tracking performance. Interestingly, at perpendicular beam-to-flow angles lateral band-pass characteristics are inferred, and the resulting lateral amplitude modulation could help improve tracking estimates. Simulations and flow phantom experiments showed that substantially improved results could be achieved by utilizing time-variant clutter filters (e.g., polynomial regression filters) despite the inherent decorrelation inferred by these filters, but only for higher ensemble sizes (N > 36). We found that, compared with color flow imaging, speckle tracking could yield consistent estimates well below the clutter filter cutoff, but with a higher variance attributed to the low signalto- noise ratio inferred by filter attenuation. Overall, provided that a low f-number and high ensemble lengths (N approx. > 36) can be used, speckle tracking can consistently provide angle- independent flow velocity estimates, limited only by a lower bound on the flow velocity itself.