It is conventional wisdom in volleyball, and indeed in most sports, that the team that makes the fewest errors should win. Many maintain that attack efficiency, and therefore implicitly attack errors, are the key determinant to success. Conversely many say that a minimum number of service errors is required in order to develop enough pressure to win.
However to the best of my knowledge noone has ever looked at other kinds of errors or how the total number of errors might influence the outcome. It sounds like something we should know about.
The first problem is how to measure errors. Total errors might be one way to go if we are looking at individual sets. But in general totals, or even per set averages, are not very good because they don’t take into account number of opportunities. Therefore the obvious measurement is a percentage of total contacts. But that ends up being a very small number if you include every single contact. And how do you include block attempts which don’t necessarily touch the ball but can still lead to a rally ending error?
The number I have started working is errors per 100 rallies. The errors that are included are the unforced errors, those is service, attack, setting, free balls, blocking (i.e. net touches). Reception errors are considered ‘forced’ errors. The numbers ends up being quite nice. In my league the average number of errors per 100 rallies is between 12 and 16. Conversely the number of points ‘won’ is between 30 and 36. I haven’t done any serious analysis but eyeballing it, it looks like there might be something there.
My question is: Is this a reasonable way to measure errors? Can you think of a better way?
All comments welcome.
Tagged Volleyball Analytics, Volleyball Statistics, Science Untangled, Conventional Wisdom, Errors
The collection of Coaching Tips can be found here.
For more great coaching tips, check out the Vyacheslav Platonov coaching book here.
A further complicating factor is the situation of the error. i.e. a player attacking a ball out of the back row on an out of system situation ( on coaches instructions) compared with simply free balling it over the net which has a high likelihood of a point loss but is not stated as an error.
A player called for lifting when diving to retrieve a ball off the block would result in an error, yet the point was in all reality lost. Filtering these errors may be required to give a clearer picture of errors with/without “worthy errors”.
The ball handling error on a difficult defensive situation, I wouldn’t count as an error in this context.
It would be interesting to know if the errors go up when points go up, or if they go down. More aggressive play might include more errors; it may also lead to more points. Somewhere there might be a sweet spot, where a certain number of errors may lead to a certain number of points.
For serving, at least, it’s possible to show that there is an optimal error rate that maximises point scoring: see this article.
In terms of ball-handling errors with reference to those called on plays made on very difficult balls, I’d venture to say those are infrequent enough not to really need to be filtered out.
The bigger question for me is whether you focus on unforced errors (those that show up directly on the stat sheet – hitting, serving, blocking, ball-handling) or on other types of errors. In the latter category I included errors in decision-making and “forced” errors which are technical mistakes when attempting to dig, receive serve, and the like. Both require more detailed analysis, of course.
It all makes sense.
The biggest problem though, is considering all passing errors to be forced. I understand why you do this, but I think it is a big flaw in the analysis.
Because what I imagine would be quite a large proportion of errors are being assumed as forced because there is no real way to distinguish between forced and unforced. There would be a proportion of hitting errors that are actually forced, same with setting errors.
Maybe the problem is with terminology.
You are right about the difficulties of separating forces and unforced. In working with large datasets, that you are not personally controlling, you have to make some assumptions.
But… My argument to players is and always has been that when setting and spiking you dont have to make an error. Ultimately you have the choice to just keep it in play. What you feel as pressure is not the same as being forced.
In practice, I do diffentiate, for example on some ball handling calls.
The ideal team makes no unforced errors and kills as much hits/serves/blocks as possible. Only focusing on the error-part of this assumption in the analysis is looking at half of the picture. I would combine the analysis of errors and direct kills in this kind of “game winning”-assumptions. Unlike sports such as tennis, a volleyball-team at a good level needs to kill the majority of the rally’s to win the game, if not it will face a certain defeat (even if it plays without unforced errors).
Exactly. That is why I focus on points won. And when I do studies, points won always correlates more to wins than errors.