xP & POE – Receive (2017)

We treat serve receive in the same manner we handle serving. We identify the expectations and then quantify the results appropriately. In this case, we break down expectations in FBSO efficiency by conference. We look at how well teams in each conference do when receiving serve (with service errors removed, as these are not really due to the receiving team). Below are the results:Screen Shot 2018-10-26 at 10.37.35 AM.png

Keep in mind that the Big West refers to the newly created men’s conference which split from the MPSF. The MIVA, EIVA, and CC are also all men’s conferences. You’ll see that FBSO eff is greatest in the Big West and MPSF which makes sense, but that the middle of pack isn’t far off.

Now that we have the expectations for every time a team in one of these conferences receives serve, we must now quantify the results. This is also pretty simple and is done in the same manner as serving. We use the R (receive) code as provided by VolleyMetrics, as well as the gender of the teams, to create a linear regression model. This model looks like this: receive_out.png

Again we see that using these “output contacts” in combination with gender yields a very strong model with an adjusted R-squared of 0.9823. Meaning that 98% of the variance in can be explained with this model. Pretty excellent. The model reads just as the serving one did, where you add or subtract the estimate for the coefficient if the specific conditions are met. Oh, it’s an R! contact in a men’s match? Well that’s just the intercept (0.109), plus 0.144 for a predicted result of 0.253. Simple enough. R then goes through the data and predicts all these values for us based on which conditions were met by the specific touch. The 2017 Big Ten and Pac 12 teams are visualized below. Hi Rutgers…

Screen Shot 2018-10-26 at 10.32.35 AM.png

So what we see here is ONLY the Points Over Expectation (POE) for what the passers on each team create. This has absolutely nothing with Stanford having Kathryn Plummer and the ability to crush out of system passes. This is just looking at the opportunities created, on average, by the passers themselves relative to the league. If you took Stanford’s top performing passing lineup and gave them to Rutgers, Rutgers would serve receive at a much better rate. Would Rutgers be able to score? Irrelevant. But their passing would be predicted to be about 0.015 better than the league average. They would be 0.015 POE. Why Stanford and Nebraska are so good is likely due to their low receive error and bad pass percentages, but we can get into that another time…