Nerdnumbers avatar

The Boxscore Geeks Show: Playoffs - No Place Like Home

Our show is also available in audio only form (Some call it a podcast!)

Playoffs - Away from Home

Patrick is back on the podcast with Arturo to discuss the young playoffs. With only a few games down in each series, there are already tons of stories and questions. Why can't home teams win? Are teams who we thought they were? Can Houston handle Aldridge? Tune in to find out!

The 'bad' call where Freeland got the freebie off of Howard was a makeup call for this Howard / Lopez exchange.

In that exchange Lopez draws the technical and fouls out. Howard is playing at home so he should benefit from hometown refs, however Howard hit Lopez. Best case that Houston had a right to is a double technical and both are out. The league did not 'give' the game to the Blazers.
All the focus on predictions (yes, they drive page views) misses the point. Wins Produced is not a metric formulated to maximize predictive power. It is a metric formulated to allow for unbiased inferences.

There are many other metrics out there may (or may not) do a better job of predicting outcomes, but they cannot tell you *why* certain outcomes are more likely - they have not been constructed in a way that allows for it. Wins Produced has given away explanatory power in order to make stronger 'why' statements about what it does predict.

It's the only metric that does it well. I'd like to see you guys play that up more; it might not play as well with the masses, but the handful of us wonks around certainly do appreciate it.
Let's also be clear - the underlying math of WP does a very good job of predicting. The problem is what. A player's performance in a year is great for predicting their next year's performance. The issue is we can't predict:
1. How many minutes the player will get (this one is killer. Coaches + injuries hate predictions)
2. Where they'll play (Yup, how the coach uses you matters to the win column.)
3. Unexpected stuff (Age, breakout season, Kardashian induced play...)

And it turns out all of that has a huge impact on team W-L. And it's odd that in the mistake column (Us Boxscore Geeks trying to guess that) the part that is most consistent and tested (WP) gets the regular blame. In fact, I get amused when any prediction contest is viewed as Metric vs. Metric.

Sign in to write a comment.