In Response to Oberon: A Thought About Maybe Using Statistics

And that's all it is, a thought, and it isn't particularly directed at Oberon's nice post below. But it kind of is. I read that and thought about going through each team's statistics over the last 25 years using some sort of brilliant algorithm to use statistics even recoverable from the 1980s to model the "best" team in the league that year, and see if they won the championship. The same model will be applicable to the other sports, and we can make likey time on this blog.

So this model. I have a trust for point differential, so start there, but the problem with straight-up point differential is blowouts. This is assuming that "clutch" doesn't exist, because theoretically a "clutcher" team would win more games with close point differentials, and become more valuable than their point differential suggests, but to pacify Oberon I will go right into a consideration of point differential.

To accurately calculate the aggregate win probability for the NBA teams each year, you'd want to find the team who was winning the most games by the largest margin. All blowouts. Because standard deviation is a negative point-related quality when the total points are already taken into account, perhaps (point differential) - (std div [point differential]).

But fuck, that's far too much work. Especially for a blog that nobody reads ever and on an issue that's really just a response to Oberon's mind dump. This story is probably the best illustration of why I never get anything done.