Today's blog post is dedicated to my wonderful, faithful friend named statistics, or "stats" or "bae" for short. As is the case with any of these writing exercises disguised as blog posts, this post will focus on whatever subject happens to be on my mind on Thursday evening or Friday morning.
I'm currently working on a project with the BYU volleyball team. In different interviews and conversations with friends and enemies, I've been asked what exactly my project consists of. I want to practice giving a concise response that explains the nature and extent of the problem in terms a normal person would understand, so here is an attempt:
An important metric to look at while analyzing team performance is serving performance and efficiency. A common, yet ineffective way to accomplish this is by looking at the number of service errors and aces. While a good step, it's hard to really see how effective a server is by the number of serves that immediately end the point since the majority of the serves stay in play. Each serve and serve receive is graded. We're focusing on serve receive grade, meaning the player on the opposing team who makes contact with the ball after the serve is graded on how well he passes a ball.
Using data from about 48000 games from 2016-2021, my job is to sort all the serves in to buckets on if they were perfect passes, positive passes, average passes, incomplete passes (ball goes all the way back over the net), poor passes, failed passes (service aces), or if there was no serve receive grade (service error). What we're looking for is called the first ball side out percentage, which is the percentage of times the receiving teams wins the point on that very first possession. We want to find the overall first ball side out percentage for men's volleyball, as well as the expected first ball side out percentage for each of those serve receive buckets.
When the pass is perfect, the setter is in perfect position to run the offense. He has the option to set the pin hitters, set the middle, set a back row attacker, or dump it himself. However, with a poor pass the setter is off the net and his options are basically reduced to setting the pin hitters. With fewer options, the serving team has a better idea of where to set the block, how many blockers to send, and allows the remaining defenders to properly position themselves to dig the ball and continue the point.
Having the coefficients for each category will allow for in-game analysis of serving performance. Using only aces and errors, it is hard to tell how effective a player's serving is with, for example, high numbers in both aces and errors. Are the 3 aces worth 4 free points from service errors? Hard to tell. However, since we expect the FBSO to be lower on poor passes than on perfect passes, we can get a good read on how truly effective these servers would be. Despite not having an ace, a player could have a very low expected FBSO. This tool provides coaches with better service indicators to enable them to be able to properly instruct players to either continue with the same serving strategy (speed, location, serve type, etc) or make adjustments to be more effective.
As part of this project I'm also pulling out the expected FBSO for individual players. This will allow coaches to see how their players stack up nationally on the service efficiency spectrum and to be able to provide their players with individual benchmark goals to work towards.
It's been fun watching the volleyball games this season in a new light. Understanding the impact of a good or a bad serve receive has really opened my eyes to why these numbers would be so useful to have as a coach. I'm grateful for STAT 495 R, Zach Knowlton, and BYU Men's Volleyball, especially Devin Young and Giuseppe Vinci, for the opportunity to do this project.
No comments:
Post a Comment