All is Quiet

I was going to start a series updating the ability of the different NBA productivity metrics to predict future outcomes today, but then I remembered that if certain bills were passed by Congress, I could very easily have my website shut down in the future if I said anything even interpreted as negative about one of them.  Did I say something mean about APM again?  Maybe basketballvalue wants to exercise a copyright.  Is PER kind of useless?  Maybe ESPN has had enough.  Am I completely ignoring NBA Efficiency but I have it in my publicly-available database?  The NBA may decide that I don’t have written consent.  I think many of us take the internet and its resources for granted, but it doesn’t have to be that way.

On a more positive note, enjoy these graphs about correlation and causation.

About these ads
This entry was posted in Uncategorized. Bookmark the permalink.

2 Responses to All is Quiet

  1. EntityAbyss says:

    Hey Alex, a couple of things. I’ve checked much of your work on all of the models, and I think you’re doing a great job. Which model or models do you think is the most useful and why?

    Also, when you do your future forecasting with the models, are you going to consider the average increases of production by year for each model? Have you started working on it?

    • Alex says:

      Thanks. I’m going to try to get my prediction update(s) done today, so I’ll try to answer your question there. But the quick answer is that I don’t think a single model is “the” way to go, and it depends on what exactly you’re trying to do.

      For your second question, I don’t know that I’ve seen an increase in average production so far but I also haven’t specifically been looking for it. Most (if not all) of the metrics should automatically adjust for anything like that by subtracting off average production; what you’d see instead maybe is more or less variance in player quality from year to year. Using WP as an example, say a player is league average one year; he’d have a .1 WP48 and produce about 4 wins in 2000 minutes of playing time. The next year if he produced the same stats but the league average went up, he would now come out at below .1 WP48 and produce fewer wins in the same amount of playing time. Or if he got better but everyone else got the same amount better, he would come out at average again and produce 4 wins again. I haven’t looked at the formula for Win Shares in detail, but I imagine they adjust similarly; I know PER adjusts for league average. The only situation I think might be weird is if you were using multiyear APM or RAPM or something like that. If there were a dramatic shift in production from one year to the next and you didn’t include year as a predictor, your baseline would spit out something more like the average production of those years. Of course, you don’t usually have player-year in that case (like LeBron 2011 and LeBron 2010), you just have players (LeBron for 2011 and 2010 combined). So I don’t know if that’s an issue per se in terms of making predictions, but I’d have to think about it more.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s