In the spirit of a new seasonal ladder starting soon, I did some digging into the data on seasonal ladders, and compiled it into a spreadsheet, which is linked at the end of this post.
The spreadsheet has the number of players who joined each season and the number who got ranked* each season. I calculated the retention rate of players per season (what % of the players who joined stayed for all 20 games) and graphed it over time. I'm not a member, so I can't post the picture of that graph on the forum, but it trends downwards slightly. The number of players ranked, however, trends upwards overtime, but peaked around 22-25, and has trended down substantially since. That graph and data is also on the sheet.
The sheet also has data on the average rating and top rating per season, and graphs that show both of those overtime. Both trend upwards, but it's interesting to note that the slope of the trend line for top rating is higher than that of the average rating, 23.8x to 15.8x. Both rose quickly around seasons 8-11 (with the exception of season 10, which was the FFA) and then have risen very slowly since.
Finally, the sheet has a ranking of players by seasonal wins, mod and Deadman are tied at 3, as well as the rankings and template from each season. (Since I wasn't around for the early seasons, some of the template names may be incorrect. If you notice that, let me know.)
I found this interesting so figured I'd share it. If you have any comments or suggestions for new data let me know.
*Each seasonal says at the bottom of the front page "There are currently x players in the ladder." This number was never the amount of players who had actually been ranked with all 20 games finished. I'm not sure where that number comes from, but I used the amount of people ranked with 20 games instead.