Accuracy of a tree is best assessed by seeing if it contains the same groups as the true tree. The trees that we are inferring are unrooted, so what could be compared are the internal branches of the true tree and of the estimate. Each branch divides the species into two sets -- those connected to one end of the branch and those connected to the other. For a tree of 10 species there are 8 (not 9) internal branches of the unrooted version of the tree. How many of them divide the species into the same sets in the two trees?