Millions of people play Massively Multiplayer Online Games (MMOGs) and participate in the social networks built around MMOGs daily. These players turn into a collaborative community to exchange game news, advice, and expertise, but in return expect support such as player reports and clan statistics. Thus, the MMOG social networks need to collect and analyze MMOG data, in a process of continuous MMOG analytics. With the appearance of cloud computing, it has become attractive to use on-demand resources to run automated MMOG data analytics tools. In this work we present CAMEO, an architecture for Continuous Analytics for Massively multiplayEr Online games on cloud resources. Our architecture provides various mechanisms for MMOG data collection and continuous analytics of a predetermined accuracy in real settings. We implement and deploy CAMEO to perform continuous analytics on data from RuneScape, a popular MMOG. Using resources from various real clouds, including the commercial cloud of Amazon, CAMEO can analyze the characteristics of a community of over 3,000,000 active players, and follow the progress of 500,000 of these players for over a week. Thus, we show evidence that CAMEO can support the social networks built around MMOGs.