[cline6]

Plotting the first half of the season on the X with the second half on the Y seems very strange to me. If you were just trying to separate the win percentages of players, wouldn't an X axis for the win percentage with a count on the Y provide you with a proper distribution? Plotting win percentage in different time periods on the two axis serves to essentially double count hits within the distribution, provided people's win percentage did not change significantly across the season.

I also don't really understand how a large difference in win percentages means that something is significantly skill based. What's certainly true is that you cannot say with any confidence that a tighter distribution means something is less based on skill, which was the extremely bold claim in this article.