With a little extra time on my hands I have taken started to look at what might be the differences and similarities between leagues, and with it whether there are any structural ‘rules’ in volleyball. Thanks to Michael Mattes and Manlio Puxeddu, who collected the files and Ben Raymond, who wrote the apps that let me crunch the numbers for the whole leagues. The leagues I will focus on are France, Germany, Italy, Poland, and Russia. For these leagues, I have almost full data for the 2016-17 season and I also think these are probably the top 5 leagues right now.
As befitting someone who is incurably curious about volleyball, has done a single statistics course nearly 30 years ago and has a slightly above ‘basic’ level of expertise in Excel, I wanted to try to see if any of these statistics that I have been writing were actually important. To achieve this high goal, I took the set win percentage of each team in each league and correlated that with various statistics I have presented in this series. I used set win percentage instead of match win percentage on the assumption that it was a slightly more sensitive indicator of differences between teams. I am completely aware that using a correlation in this way is only marginally better than just having a list of numbers, but better is better.
One quick note for those interested in / concerned with sample sizes, each league data set has at least 150 matches and around 30,000 attacks.
ATTACK PERCENTAGE v ATTACK EFFICIENCY
We can see that attack percentage and efficiency are both highly correlated to win rate in 4 of the 5 leagues. In 3 of the leagues the differences were small and in Russia the difference was a little bigger but both measures still correlated highly. In France spiking does not seem to be highly correlated with winning, whichever way you want to measure it. This is a strange result that I will postulate on in a future post when I look at the internal strength of each league.
Interestingly, the rally win rate measure that we introduced yesterday is more highly correlated with winning than the other measures, even in France. It seems that the previously unmeasured balls could well have an impact on the result, which should not be a surprising discovery. There may be better, more accurate ways of measuring those ‘missing’ balls but the rally win rate is particularly easy to measure and doesn’t require any subjective rating of contacts or extra coding of events (similar to the attacks per defensive opportunity statistic approach for measuring team defence).
ERRORS AND BLOCKS
As we know the attack efficiency takes account both direct errors and blocks conceded. The question which of those, or indeed whether one of them, is more important than the other is not one I had spent a lot of time on. Now I have and it seems pretty clear that we can’t make any decisive statement either way. In France it is irrelevant if you get blocked out. In Poland it barely matters if you hit it out. And everything in between. The sum of the two is fairly stable from league to league, but nowhere near as important as attack percentage alone, efficiency or rally win rate.
Please note that as errors are a negative correlation, the lower the number the higher the correlation.
So there you have it. Whatever it is.
The collection of Coaching Tips can be found here.
For more great coaching tips, check out the Vyacheslav Platonov coaching book here.