Not surprising, in a recent posting with QSO data from Storvreta, Sweden (Grid: JO89ux) (https://thewakesileave.wordpress.com/2015/11/30/europe-and-asia-qrpp-from-norse-lands-with-a-magnetic-loop-indoors/), the average bearing of 20m contacts was found to correlate with the azimuth of the Sun. However, although at noon they both seem to coincide, the contacts moved west slower than the Sun.
A similar graph and correlation were attempted with 10m contacts made from Piriápolis in Uruguay (Grid: GF25id):
In this case the correlation was even stronger and the coincidence at noon was almost perfect. Naturally, in Sweden the azimuth of the Sun, as well as all the contacts were south of the QTH, while those in Piriápolis both were north of the QTH. Besides different propagation conditions, different HF bands and obvious geographical and demographical differences, the QSOs made from Storvreta were made during winter months in the Northern Hemisphere, i.e., with the Sun reaching closer to the Tropic of Capricorn and the furthest from the QTH; while those made from Piriápolis were made during the summer in the Southern Hemisphere, with the Sun also reaching closer to the Tropic of Capricorn and hence, the closest to the QTH.
The actual contacts can be seen in http://www.hamlog.eu/index.php?pg=11&m=list&usc=8720.