A New Statistic For Attack?

Posted by

block gustavo

In the never ending search for understanding volleyball just a tiny bit better than I did yesterday, I have been looking at different ways to measure attack.  Obviously, I am not satisfied with the current way (kill percentage / attack efficiency).  Kill percentage is pretty good because you have to win points to win a match, so there is something important there.  Many people maintain that attack efficiency is more important.  However, attack efficiency says that 50% kills and 20% errors is the same as 40% kills and 10% errors.  It is obviously not.  In addition, using only these numbers ignores the 30-40% of attacks that don’t finish the rally.  Those attacks must be worth something.

So I started thinking and looking at something that was easy to measure and came up with ‘new’ statistic: rally win rate.  The rally win rate is the percentage of rallies won in which the team made any attack.  Each rally is counted only once, so if there were three attacks in that rally, AND the team won, the rally win rate would be 1 or 100%, even if none of those attacks made a direct point.  The logic being that the quality of the attack that is not a point influences the outcome of the rally, either positively or negatively.

If you don’t follow, here is an example.  A spiker spikes a ball into the block, his team covers, he hits again and this time scores.  His kill percentage is 50% (1/2), his efficiency is 50% ((1-0)/2) and his rally win rate is 100% (1 rally won / 1 rally with at least 1 spike attempt).  If in the next rally he spikes into the net, his kill percentage is 33% (1/3), his efficiency is 0% ((1-1)/3) and his rally win rate is 50% (1 rally won / 2 rallies with at least 1 spike attempt).

In another example, a spiker spikes, the opponent defends but can only spike a high ball which is blocked by his team.  Kill percentage is 0%, efficiency 0%, rally win rate 100%.  And so on and so on.

In the next couple of days, I will test this out and see how it works.


The collection of Coaching Tips can be found here.


For more great coaching tips, check out the Vyacheslav Platonov coaching book here.

Cover v2

Advertisements

14 comments

  1. We also know that not all kills are equal in terms of difficulty. A kill against no block is clearly easier than when the hitter than against a triple in an out of system situation. This measure would allow for the contribution of a hitter that didn’t actually achieve the kill in the rally to also be included. Contributions of passes/digs could also be attributed, however like most statistical measures it would have more reliability for players who are involved in a lot of points in a match/season.

    In an assistant coaching role, I had difficulty convincing the head coach and setter of the value of a particular pass-hitter. This player would routinely be set in difficult situations and rarely got kills, however more often than not his attack placed the defense under pressure, resulting in a weak attack or free ball. The #1 hitter was generally then set in a more positive situation usually resulting in a kill. Confirming in the mind of the setter and coach of the importance of the #1 hitter and the impotence of #2. The irony being that in those difficult out of system situations when the #1 hitter was set, he generally made an error, hence he rarely was. I noticed that despite low attack percentages for #2, the team won more points, sets and matches when #2 had more court time and so felt his contribution was being undervalued.

    This statistic would give a measure to the value of hitter #2, who would retain his low kill percentages but achieve high rally win rates.

    Like

    1. Different tempos of attack are completely different in their expected results, but we still lump them all together. Which, as in your example, leads to spikers being judged less accurately. I don’t know that this particular statistic addresses that issue, but it might be closer.

      Like

      1. As I read it, your article provides a great way to adjust player stats up and down, depending on difficulty of attack situation and strength of opponent. Exactly what I needed in my situation. I guess like all stats it requires large data sets and probably is most accurate post season when all of the data is available. The holy grail of accurate, relevant and timely stats continues to elude. We are yet to see a volleyball Hari Seldon with tiny sample sizes and accurate predictions.

        Like

      2. Like with anything, the larger the data set and the stronger the data crunching tools to analyse them the better.
        If you have some historical data you can still make some adjustments. For example the first time I started down this track I only made adjustments for tempo and not for phase or position. I think this still gives some better information than the raw number.

        Like

  2. mmmmm……. it’ll be interesting to see how the numbers work out. In theory you could have 6 players earning a ‘rally win’ in the same rally.

    Like

    1. You won’t have to wait long…
      Correct. You can have a rally with a team attack win rate of 100%, and all six players also having an attack win rate of 100%.

      Like

  3. Don’t think I like this one. I get the idea that you want to try to value non-terminal attacks in some way. It begs the immediate question, though, of whether you should consider rally wins equally if they include differing number of attacks in them. After all, if you don’t get the kill, but force a free ball that you then terminate (2 swings), it implies a stronger initial attack than if it results in an series of out-of-system swings by both teams (3+ swings).

    Don’t know if the data is readily available, but my first choice would be to look at the immediate outcome of the non-terminal attack.

    If we have to stick to more general data, what about looking at things in the same way you’ve talked about judging your own defensive performance in terms of generating successful attacks? Meaning, how does the opposition perform in the transition phase?

    Like

    1. The object is to win one point. So given that all rallies are potentially one point, I think it is fair to judge them equally.
      Don’t forget the object is not to be perfect, just to be more accurate than the current measures.

      Like

  4. It seems to me that while seeking to find measures “that are more accurate than current ones” is at first glance logical and worthwhile, the reductionist intention of this whole process will always mask more than it reveals. With so many dependant and independent variables in a rally, suggesting the winner of the rally measures something specific is surely a stretch. It might measure the effectiveness of a system or the team as a whole… but the scoreboard will do that in most cases anyway.

    Like

    1. Great point.
      Understanding the game better, even if the numbers don’t specifically help me, helps me make better judgements and decisions.
      All variables in volleyball are dependant. Even serving, while a closed skill, is dependant on the quality of the receiver. Everything else is more dependant.
      Winning a rally measures the effectiveness of a system, but not all rallies are equal or have the characteristics. Over a longer period you can group together rallies that have similar characteristics and perhaps piece together the contribution of different parts of that system.

      Like

  5. Really am intrigued with this approach. It would take into account a level of effectiveness of each attacker. For example, it looks like if the player (assuming) wisely did a push attack for a recycle on a poor set to give his team a chance to cover and set up another attack, it is a wash. The next attacker, s/he or someone else now gets the opportunity to put away the point, will now “inherit” the rally with an opportunity to end it with a pt won, lost, or another recycle.

    Like

    1. I like this idea. I tried taking this attack rally win percentage data for the team as a whole during a match using a clipboard. I also took the data for a similar win rally percentage for digs and block touches at the same time. It gave a nice snapshot of how effective we were in all phases of the game.

      Liked by 1 person

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

w

Connecting to %s