Online manipulation of information has become more prevalent in recent years\nas state-sponsored disinformation campaigns seek to influence and polarize\npolitical topics through massive coordinated efforts. In the process, these\nefforts leave behind artifacts, which researchers have leveraged to analyze the\ntactics employed by disinformation campaigns after they are taken down.\nCoordination network analysis has proven helpful for learning about how\ndisinformation campaigns operate; however, the usefulness of these forensic\ntools as a detection mechanism is still an open question. In this paper, we\nexplore the use of coordination network analysis to generate features for\ndistinguishing the activity of a disinformation campaign from legitimate\nTwitter activity. Doing so would provide more evidence to human analysts as\nthey consider takedowns. We create a time series of daily coordination networks\nfor both Twitter disinformation campaigns and legitimate Twitter communities,\nand train a binary classifier based on statistical features extracted from\nthese networks. Our results show that the classifier can predict future\ncoordinated activity of known disinformation campaigns with high accuracy (F1 =\n0.98). On the more challenging task of out-of-distribution activity\nclassification, the performance drops yet is still promising (F1 = 0.71),\nmainly due to an increase in the false positive rate. By doing this analysis,\nwe show that while coordination patterns could be useful for providing evidence\nof disinformation activity, further investigation is needed to improve upon\nthis method before deployment at scale.\n