Volleyball Analytics

First, I’m wondering if anyone has any references for volleyball analytics work. I’m having a hard time finding much of anything substantial.

Second, I have a rough idea where I’d like some input. One of the main stats for individuals in volleyball is hitting percentage, which is (kills-errors)/total attacks. It’s a good stat but I never know how to compare a 40% hitter with 400 kills and and 30% hitter with 700 kills.

So my idea is to run a really simple linear regression on total attacks and kills using all players over the last 4 years with >0 kills and >35 sets played in each season. The next step would be to use that to generate a simple expected kills calculation based purely on volume of attack. From there, actual kills-expected kills would be a sort of kills added stat. I’ve gone through the process already and the output seems reasonable for a descriptive stat.

Does this sound passable or am I way off in my thinking here?

Here is the plot of total attacks and kills at the player level

vball plot

And here is the top 20 in kills added for 16-17 using this approach in big five conferences, which includes eight All-Americans and lots of other really good players.

I have a bunch more ideas but wanted to try to at least start a conversation somewhere and see if I’m even anywhere near a reasonable line of thinking on something as simple as this.

I thought this series of posts was presented very well: http://varianceexplained.org/r/simulation-bayes-baseball/ and could be applied to your question of finding a hitter’s “true” percentage.

I ended up buying his book (accepts donations) and worked my way through it, while also applying it to a player’s 3PT % in basketball.

It took me some time as I had never really worked with Bayesian methods but might be worth your time. All his code examples are in R.

Thanks for the suggestion. I’ll definitely take a look. I think there are some obvious parallels between shooting percentage and hitting percentage so I’ll dive into this for sure.