(03-19-2015 09:35 PM)Monarchblue Wrote: Why do we need an empirical method. Can't we just have experts evaluate teams by watching them. I watch a lot of basketball, and among the teams I watch, I can tell who is better.
Because experts and fans a like have personal bias, and will see what they want to see. Thus analytics helps us remove our bias. Theres a reason why physical evidence is of a higher value than eye witness accounts. If its just about experts being able to evaluate teams by watching them, then why can't you accept the committee, who watched way more basketball than you did, was able to tell who was better?
Watch all the basketball you want, and use analytics to help process what you see as some of the leading coaches and teams in this country do:
http://bleacherreport.com/articles/15568...basketball
I also found it funny how people used Baylors and ISU loses as vindication against kenpom.
Heres kenpoms predictions via simulation:
http://bleacherreport.com/articles/15568...basketball
Looking at ISU chances they had a 87% chance of advancing to the round of 32.
The person who apparently didnt pay attention in stat class will look at that and say "Hah! See they had an 87% of winning and they lost, shows how much kenpom knows."
The person who paid attention in stat class will look at that as if there is a bag with a 100 balls with 87 being red, and 13 being green. They stick there hand and pull out a ball and its not red. What happened? Well what happened is despite having an 87 percent chance of pulling out a red ball, the chances of pulling out any one ball is the same as the other 99 balls, and it just happened the specific ball he pulled out was one of the 13 green ones. And thats how you can view the infinite amount of outcomes to a basketball games, they all have an equal chance of happening, it just happens that 87 percent of those outcomes had ISU winning.
Its even lower for Baylor: 77 balls
Texas even lower: 52 balls