Your center of mass in relation to foot strike position.

For those arm swing/pulsers/ COM and head over foot folks consider some more research below.
Let the CNS drive the show, it is what it is there for . . . The leg motor patterns are dominant, the arms are passive and "shape" and influence the leg swing as a balance and ballast effect. As we discuss in an upcoming podcast, to cross the arms in a pumping motion across the midline of the body means one has to have compromised scapular mechanics (mostly protraction) to afford that much humeral adduction. This means we are forcing thoracic rotation as well. This means we are reversing what we know is more true, that "arm motion is driven passively by rotation of the thorax (Pontzer et al., 2009), an idea which is supported by shoulder muscle EMG data" (and not thoracic rotation by arm swing). Why would we try to create more unnatural axial spin through the spine when we are actually trying to move forward in the sagittal plane? Why would we try to force more rotation through the spine when the function of the thoracopelvic canister (ie. the core) is to stabilize rotational /angluar momentum? Hmmmm, things to ponder.

"Previous modelling studies have clearly shown that motion of the arms effectively counterbalances the angular momentum of the lower extremities during running (Hamner & Delp, 2013; Hamner et al., 2010). It has further been suggested that arm motion is driven passively by rotation of the thorax (Pontzer et al., 2009), an idea which is supported by shoulder muscle EMG data, consistent with the shoulders as spring-like linkages (Ballesteros, Buchthal, & Rosenfalck, 1965). Our data are con- sistent with this idea, showing motion of the thorax to be in the opposite direction to that of the swinging leg. Pontzer et al. (2009) also suggested that motion of the thorax is driven passively by motion of the pelvis. However, our data shows that the thorax reaches its peak angular velocity earlier than the pelvis, indicating that thorax motion is not completely passively driven by pelvic movements."

-S.J. Preece et al. / Human Movement Science 45 (2016) 110–118