It's a full year, but since the summer months (northern hemisphere) ends up towards the end of the graph it will tilt the trend curve upwards. Natural variations, September to April have a relatively coherent number of player variations, May through August has a [bleep]e due to summer breaks on the northern hemisphere. In the example posted the summer [bleep]e is put at the end of the graph, which naturally tilts the trend curve upwards. Take a snapshot from May to April and it will tilt the result downwards. What I meant in the first post was that placing the graph from January to December would put it (the summer [bleep]e) in the middle and have less effect on the trend outcome. To get a more reliable result you would have to compare two years however (my second post) since the data from within a single year varies too much (particularly the summer [bleep]e) to make any solid conclusions from it. You could also compare it month by month (ie jan 2008 to Jan 2009) or even week by week (w1 2008 to w1 2009, though weeks would probably be less reliable). The graph only shows that more players are playing the game due to summer breaks, but it does not show you the overall trend or whether less/more players are playing due to updates. Compare it to weather, sure you can see that the weather is getting warmer as winter ends. But you can't see if the climate as a whole is changing by just looking at a randomly positioned one year data set.