Historical data includes some fluctuations that are not based on reality, but were rather caused by bugs in PlayTracker and changes to our formula and database over time. Newer data is more reliable.
The concurrent players graph shows how many players were playing a game at a given time, sampled from Steam. If only a date is given, it was a point in time on that date, not a peak or a sum of players. No internal PlayTracker data is used.
Popularity over time
The popularity over time graph shows PlayTracker's own "Popularity" metric for a game over time. "Popularity" is calculated from values like total owners, active owners, concurrent players, and more. An active owner is valued much higher than an inactive owner in "Popularity" calculation. It scales linearly - there are no diminishing returns.
The recent playtime graph shows total estimated playtime (in hours) clocked for a game in the past two weeks. So if a player plays an hour on June 10th, that hour will display on the graph until June 24th. Uses Insight data (n=~120000).
The active players graph shows the number of users that launched a game in the past two weeks at a given point in time. So if a player plays on June 10th, that player will count as an active player on the graph until June 24th. Uses Insight data (n=~120000).