Alright all: I did get some software to try and help with sound normalizing. Sadly there was a slight echo that no amount of post-processing could get rid of. I’ll work on getting a separate recorder for next week. Thanks for putting up with an amateur attempting to podcast.
If the player above doesn’t work for you, then you can:
Dave and I discuss some of my experiences at Sloan and some of Dave’s takes on how research is done and should be done.
Dave’s gotten around 40 papers published in Sports Economics and his work actually shows up in Basketball on Paper.
The current version of Wins Produced is probably version 6.
People like Henry Abbott have argued stats people need to communicate their ideas better. Dave points out consultants are often trying to effectively sell managers confidence and affirmation as opposed to the truth.
Dave points out in Academia that people with experience in the area are who evaluate papers. We don’t know who does this for Sloan.
Dave argued that Economics conferences are less of a “Dog and Pony” show and more about presenting ideas without embarrassing yourself.
Dave points out the Wins Produced metric is used to explain evaluation in the NBA. Additionally, he does respond to criticism (see his FAQ) and that good criticism needs to be framed in a cohesive manner.
Dave says regressing a model on another model doesn’t really test anything. Regression is using a model to explain a dependent variable (e.g. Wins explained by stats. Yes I did mess up and say independent variable when I meant dependent variable) When the dependent variable is a model, nothing can really be concluded.
A misspecified model can be dangerous. For example simply comparing offensive rebounds to wins will show more offensive rebounds decrease winning! When you control for missed shots though, we see offensive rebounds contributed to wins.
Hoopdata does have some more granular stats (e.g. charges taken)
Dave has a bold claim: All Sports are the Same!
Many sports experts though, are single sport experts. That’s why baseball experts may miss the parallels with basketball.
In hockey shot creation is actually a skill as opposed to basketball.
In the Sloan talk Boxscore Rebooted Dean Oliver pointed out +/- is not a very useful stats. Dave described this problem with an Upton Sinclair quote “It is difficult to get a man to understand something, when his salary depends upon his not understanding it!” This can also apply to stats like PER.
I want to make some thoughts on Sloan clear. It was an amazing experience and I enjoyed it immensely. It is great to have lots of people interested in stats come together to discuss them. I also want to make sure I give a shout out to Deconstructing the Rebound with Optical Tracking Data, which I feel did approach a problem with a very good scientific mindset. My “problems” can actually be described with a Kurt Vonnegut quote:
Another flaw in the human character is that everyone wants to build and no one wants to do maintenance.
Sloan is a conference where people show off what they’ve built. I described Sloan as a great stats party. When we look to the “future of analytics” though, it is not simply through enthusiasts having fun building models. Bill James showed in the 80s — without the internet even — that you can get fans together to have fun with stats. The next step is hard. You need insane, crazy people to comb over the work and verify it. This means ideas need to be deconstructed, tested and even, possibly thrown away if they aren’t any good.
When Dave brings up the Western Economics Association, he makes the point that peoples’ work is evaluated by others that are experts in their field. It is very easy to be proud when something gets popular on the internet. However internet popularity is not validity. I will personally attest that the reason many of the writers on the Wages of Wins Network were drawn here is because Dave Berri is an expert in this field. When we ask him to evaluate our thoughts, he is candid and has credentials to back it up.