CLASSIC SCIENCES ARE TAUGHT AS THOUGH THE ANSWERS ARE ABSOLUTE. Thanks to Isaac Newton, we can predict the eclipse times and locations with great accuracy. Centures of chemists taught us what happens when burning hydrogen and oxygen together - it makes water - and we even know how much.
But most sciences have a degree of uncertainty, especially sciences involving trying to predict people. It doesn't mean that we can't predict people or, for this purpose, basketball players. It just means that we can't predict them perfectly. There are error bars on those predictions.
With sports, there is additional uncertainty beyond making predictions. There is actually disagreement on how productive a player is right now, not just in the future. Given the same numbers in the same sequence of events, people won't agree. Whereas the position of the moon can be agreed upon, "performance" isn't completely agreed upon even when you're looking at it.
One thing I do to illustrate this is to show people one basketball play and have them tell me who should get credit for the two points scored. Or, as a harder example when no points are scored, what defenders should get credit for stopping a basket. People don't agree. Or people say nothing because they don't really have any idea. People are scientifically a pain-in-the-ass to study and understand.
But this is basketball, a game with rules, with strategies and solid ideas for what players should and shouldn't do. It has definite outcomes with detailed information on how we got there. There are some rules that sports analytics professionals ("sports geeks", who are sometimes not considered people) have developed to actually say who's to blame and who's to credit. Those rules turn into little formulas that scare the livin' bejeesus out of "people". I am not the only one who has scared people with numbers - there are lots of player value metrics out there. They tend to agree on a lot of general concepts, especially offensively. But how they get implemented and how they deal with defense - those vary and lead to uncertainty.
The object of this article is not to rank those player value metrics or to document what they do. It's to just show you their worst cases. It's kind of a Down With People article, showing everyone how bad things are - in the worst case. This article identifies the players from the 2017 season with the most inconsistency in their value across metrics. There isn't a perfect way to even come up with that list, so pipe down. I chose a method so I could talk about a few players. Live with it.
A few notes on this list:
You may wonder why I didn't label the graph with the actual names of the methods. Actually, you probably want to throw things at me for doing that. And that's why we live in a virtual world where sticks and stones break no bones (because no one goes outside), but social media will get you in trouble.
I didn't label the methods because people (yes, people) have these amazing biases built up. They like the President or they hate the President, so any action that he or she takes is rationalized as good or rationalized as bad based on that, not on whether the action itself fits their principles. There are a lot more people who love or hate the President than love or hate these metrics, but I don't want to support those lovers/haters who don't think. The message here is about the uncertainty, not the names or pros/cons of the methods.
I chose metrics that could be accepted based on general judgment. Metrics that generally add up to the right totals - a team's win totals, for example - were candidates. Among those, each of these have supporters. Some supporters of one hate supporters of another and, yes, "hate" is something that does show up in metrics debates. I certainly have my opinions, but "hate" is not something I'd use to describe my feelings about these numbers.
Stop throwing things at me.
You want to know what metric is what? OK. Now you can feel justified to form opinions based on your previous biases.