That chart suffers from the same confusing issue as the one here:
https://steamcharts.com/app/359320#All :
Once you go past a few months from present (cant remember if it was 3 or so) the charts dont track the same numbers. They dont track actual average concurrent but maximum peak concurrent. Dont ask me why. You can see that transition for what is actually plotted around August in the chart, highlighted with a red line:
View attachment 275832
For example if you take a look at the historical minimum concurrent average players that happened in April 2016 (which was 2,841) you will also see to the right the max peak concurrency in that month to be 5,337 which is around the same level you see plotted in your chart for those days in April, i.e. not the 2,841.
View attachment 275828
On the other hand the last few months in your chart do not plot maximum peaks but actual average concurrent which are always to show much lower than peaks. Both maximum peaks and current averages are mixed in the same chart, which makes it thoroughly confusing to use and lead often to the wrong conclusions.
In case of doubt it is much better to use the actual average concurrent monthly tables here, below the chart:
https://steamcharts.com/app/359320#All