The local velocity distribution of dark matter plays an integral role in interpreting the results from direct detection experiments .
We previously showed that metal-poor halo stars serve as excellent tracers of the virialized dark matter velocity distribution using a high-resolution hydrodynamic simulation of a Milky Way–like halo .
In this paper , we take advantage of the first Gaia data release , coupled with spectroscopic measurements from the RAdial Velocity Experiment ( RAVE ) , to study the kinematics of stars belonging to the metal-poor halo within an average distance of \sim 5 kpc of the Sun .
We study stars with iron abundances \text { [ Fe / H ] } < -1.5 and -1.8 that are located more than 1.5 kpc from the Galactic plane .
Using a Gaussian mixture model analysis , we identify the stars that belong to the halo population , as well as some kinematic outliers .
We find that both metallicity samples have similar velocity distributions for the halo component , within uncertainties .
Assuming that the stellar halo velocities adequately trace the virialized dark matter , we study the implications for direct detection experiments .
The Standard Halo Model , which is typically assumed for dark matter , is discrepant with the empirical distribution by \sim 6 \sigma and predicts fewer high-speed particles .
As a result , the Standard Halo Model overpredicts the nuclear scattering rate for dark matter masses below \sim 10 GeV .
The kinematic outliers that we identify may potentially be correlated with dark matter substructure , though further study is needed to establish this correspondence .