So, this season at Harker, I convinced my AD to splurge a little and get us Volleymetrics for the spring. I’m stoked to finally get some coded data on the high school level that can hopefully inform my approach to coaching similarly skilled teams in the future – while also validating or disproving some of the methods we’re already trying. To that extent, because nobody actually reads this blog, I don’t mind sharing my high school team’s data with the world and give you some insight into how we quantify our performance from an execution standpoint.
So far, I have player-specific serving, receiving, and attacking stuff built – with the goal of adding some blocking and defensive pages later – as well as overall team numbers. But for now, here’s what I have (using 2018 Stanford Women’s data until I have my own):
The goal is certainly that these charts read pretty intuitively so that even smart private school kids can figure them out without asking the same questions 100 times. On the left is the expected FBSO Eff of the opponent when each of the Stanford players serves. An ace being valued at -1.000 for the opponent and a service error being +1.000. The other outcomes of perfect, good, medium, poor, and no attack all have specific values based on what opponents will typically hit against us given those types of passes. Therefore Sidney Wilson is Stanford’s best server. When she serves, we would expect the opponent to FBSO at the lowest level of any Stanford server. The beauty of using expected FBSO numbers is that it eliminates the idea that Sidney can be successful just by serving in a great rotation for Stanford – because we don’t look at whether the opponent actually sides out, we look at how we would expect them to do given the resulting pass quality (which is an aggregation from all 6 rotations).
On the right is a breakdown of each player’s results. If you look at Holly Campbell at the bottom, she’s technically Stanford’s worst server. We can see visually that her high error percentage (29%) is likely the major drag on her overall efficiency. Just provides a quick glance as to why a player may or may not be experiencing success from the service line.
Here’s the same type of chart, but from the reception side. Holly Campbell is actually Stanford’s best passer, but likely because she only handles easy short serves or simple balls that hit the tape and fall in her lap. We know she doesn’t pass many balls because her bar is light blue. The darker the blue, the more balls overall the player has passed. This means that Morgan Hentz, unsurprisingly, is the top Stanford receiver who is considered a primary passer. When Hentz passes, we expect Stanford to hit 0.289. This is obviously different from a passer rating, where you’re on a 3 point scale and if you get aced, it’s a 0. In our build, you get the value of the pass as it relates to your team’s FBSO Eff. Therefore if you get aced, that’s -1.000. A perfect pass here only buys you about 0.398 in Stanford’s case, so you’d actually need 2.5 perfect passes for every time you get aced just to be at a 0.00 FBSO Eff. Kinda wild when you look at it mathematically.
The same breakdown of passer results is on the right and you’ll see that the primary passers all are about 60%+ perfect or good and very very low reception error. Huge deal for Stanford and their ability to side out quickly.
Now we get to attacking. Again, the darker the blue, the more swings the player has taken – and we have a similar breakdown of results on the right. To be clear, this is not a traditional attack efficiency in that it’s (Kill-Errors)/Total Swings. In this case, we add or subtract value based on what you do on non-terminal attacks. So obviously a kill or error results in +1 or -1 respectively. But we need to value recycling off the block, putting the opponent out of system, and putting the opponent in system all very differently.
If you look at the breakdown on the right, you’ll see the codes we look at and value individually. Working from left to right in terms of the value of the outcome – we see kill is the best, then bad digs and balls recycled off the block, then medium digs and good block touches, followed by perfect digs, and lastly errors are the worse outcome for the attacking team. This again allows us dive deeper into why an attacker might be succeeding or struggling overall – and often players who handle non-terminal balls well can boost their performance greatly.
Finally we look at both In-System and Out of System attacks individually using the same charts:
Again, these are broken down the same way as all the other charts.
So there you have it. That’s where I’m starting with our data for the Harker season. Hopefully this won’t be too much for the players to handle – and will allow a clearer look into where we can improve our efficiencies the greatest – the lowest hanging fruit so to speak. First match is tonight so we’ll see how things go!
Rethinking Volley 