.png)
On Creating a New Volleyball Analysis System
I have created a new volleyball analysis system (VAS) for my dissertation which is based on rating every technical skill on a five point scale (definitely different to my previous one). This may sound simple to do but the process of creating a new VAS is actually a lot more complex in practice. (I like the quote by Einstein, “In theory, theory and practice are the same. In practice, they are not.”) Before I go through some of the difficult issues I encountered,I will write about the VAS basics.
Volleyball Analysis System Basics
#1. Any VAS should cover all aspects of the volleyball performance to gain an understanding of the overall performance. This will mean covering the six technical skills of volleyball: serve, reception, setting, attack, blocking and digging.
#2. It cover point ending actions (point won or lost) and continuity actions (play continues). The continuity actions can be subdivided into the quality of these actions. It is normally based on attack options and the rating system for both continuity actions and point scoring actions are described below.
#3. Each action should be fully defined. Precise definitions of the operational terminology is important because it provides objectivity. That is, two different statisticians using the same VAS should be able to stat the same match and come up with the same data. This ensures the data is accurate and then those responsible can make informed decisions to enhance performance.
Sounds simple to create your own one right? Well, below are some of the issues I encountered…
Creating Two New Technical Skills
VAS Basic #1 above says any VAS should cover all aspects of the volleyball performance. The six technical skills doesn’t cover all ball touches, as there was number of ball touches that were not recorded and were subsequently missed. They were digs from free balls, overpasses and a blocked balls. It does technically not go under the definition of a dig, which is defined as is the first ball touch (not including a block touch) from an opposition attack. Free balls, overpasses and blocked balls are not opposition attacks. The definition of a dig can be changed to include these digs but I felt these were separate skills. Digging a ball from a free ball is not the same as covering the attack. There are differences in the incoming ball speed, ball trajectory and time to react. So by putting them together, you lose information about a player’s digging skill. They may have got good digging stats because they taken a lot of free balls. For this reason, I created two new technical skills of cover digs and free ball-overpass digs. By differentiating this, I have the advantage of calculating the hitting percentages from reception, digs, cover digs and from free ball-overpasses. There was also a missing ball touch of free balls itself. It isn’t an attack attempt but I included it as part of an attack rating.
How to Rate the Art of Setting?
The way to rate reception, digs, serve etc is actually really good. However, I found there isn’t a really good way to rate the setting skill (this may be the reason why all the research studies into valid performance indicators have not tested setting). I found three options on how setting can be rated:
1. Based on location. A good setter should be able to set the ball not too short or too long and not too close or far from the net.
2. Based on assists like the NCAA and Olympics. An assist is a set attempt that results in a kill. A good setter should be able to set the ball that increases the chance of a kill.
3. Based on number of opposition blockers. A good setter should be able to disguise their sets as to create more 1v1 or 1v0 attacker vs blocker situations.Option 1 is terrible mainly in due to its subjectivity. When is it too short or too long? When it is too close of far from the net? If we have distances involved, e.g. 1 metre from the net, then there is the subjectivity of judging distances. Then there are preferences (an attacker may like it short and tight on the net), coaching philosophy (never close to the net) or tactics (set tight over a short blocker). I am not a fan of option 3 either because it’s fine when the setter has all attack options. In any other situation, it is generally obvious where the ball is going and the middle blocker can read it. So through no fault of their own, a setter can only create 2v2 or 2v3 situations. This may encourage the setter to make the non-obvious choice. This has issues itself because the non-obvious choice is usually difficult to execute and increases the chance of a free ball or an assist error. Also, I’ve statted matches when an outside hitter had to play middle and she was terrible at moving to the left side or right side for the block. This meant the setter had many 1v1 situations and thus great setting stats but it wasn’t due to their setting skill but to a poor movement and reading skills of opposition middle blocker. Also, it depends on the opposition’s blocking system. I’ve seen a system where each blocker is responsible for their own hitter, i.e. 1v1 situations. I’ve also seen coaches telling blockers not to block back court attacks, i.e. 1v0 situations. On both occasions, a setter will have great setting stats but it was not due to their setting skill but to the opposition’s blocking system.For this reason, I chose the lesser of two evils in option 2. It is not ideal because it depends on the preceding pass and it also depends on the attacker. However, I feel a good setter should be able to make bad passes look good, set the ball where the attacker likes it, set at a higher tempo, set to their go-to hitters, who is running hot, run frequent combination plays etc. which all accumulates into increasing the probability of a kill.
Making Every Skill Rated on a Five Point Scale
VAS Basic #2 was you can subdivide the continuity actions based on attack options. Thus, point scoring skills (attack, serve, block) were rated on a five point scale and continuity skills (dig, reception) were rated on a four point scale. The missing point on the scale doesn’t mean much but when I tried to do a radar chart showing the average rating of the six technical skills like the one below, it doesn’t match up. A player with the maximum average reception rating of 3.0 would not be obvious on a radar chart because it goes to 4.0. So I decided rate all technical skills on a five point scale. It also had the added advantage you can be an average rating of ALL the technical skills, something I termed “ball control”. The extra action I gave to the continuity skills was an overpass that resulted in an immediate opposition kill. It wasn’t exactly a reception error because the opposition still had to get the kill but it worse than a poor reception with no attack options because we still we created an easy situation for the opposition to get the kill.
Unclear Definitions
VAS Basic #3 is the biggest challenge you will get. The short descriptions of the ratings above are not exactly well defined. For example, when do we have maximum attack options on reception. Further reading says it is when the middle player is available for a quick attack. So when we see a reception with nice height, lands precisely between positions 2 and 3, setter jumps, middle is flying in the air, then that is a 3 rated reception. When we see it, it’s obvious. But they are not all that simple. Those who has ever rated reception will know there are many tricky situations. For example:
What happens if the ball lands 2 metres off the net?
What happens if it goes to the right spot but with a low trajectory that the middle player cannot there get quick enough?
What happens if it was a nice trajectory but the middle player is too slow?
What happens if it lands between positions 3 and 4 and the middle player adjusts for the reverse quick attack?
What happens if the ball lands 3+ metres off the net but the setter manages to shoot the ball to the middle player for a kill?
What happens if it lands between positions 2 and 3, middle is flying, the setter jumps but he or she sets with one hand – would maximum attack options be available then, i.e. able to set quick middle, left side, right side and pipe with one hand?
What happens if the team does not play quick attacks but metre balls through the middle?
What happens if there is no middle attack at all? Lynx’s Antonio Codrington will hit slides as a middle and Sheffield’s middle player Sonny Chaney will hit through 2 when in position 2.
What happens if the middle player doesn’t commit to the quick attack?
These are the grey areas that needs to be addressed so when these situations arise, we record the same thing and it doesn’t depend on how we feel on the day. To address these situations, what I do is I try to understand the philosophy of what you are trying to measure. The point of the quick middle attack is to commit the middle blocker. This means there is a better chance of creating 1v1 situations or a badly formed block if the ball goes to the wings. So for this reason, the middle player has to be jumping or in the air as the setter sets to commit or at least occupy the middle blocker. If this does not happen for whatever reason then it defeats the purpose of a perfect reception. The setter also needs to set within the 3 metre line zone. This gets away with the ambiguity if the ball is off the net or towards position 4 (but the middle players still needs to commit and run the quick middle). Now, it may not be the picture perfect reception attempts like in Seain Cook’s highlight reel but I didn’t say it was easy.
The above is an example of rating one action variable only. I haven’t even talked about the grey areas between what is “limited” attack options and what is “no” attack options. I also haven’t talked about the other technical skills. For example, a block kill is awarded to the player who blocks the ball that directly results in a point won for the blocking team. Easy? But what happens in a two person block, it hits the first blocker, comes off the second blocker and then lands in the opponent’s court directly for a point won – which player gets the block kill? The second blocker got the block kill in the end but it would not have happened if the first blocker blocked it first.
Another example is a missed digs, which is defined as a dig attempt that results in a point lost, i.e. an opposition kill. We understand that if a ball lands near the defending player then that player will be charged with a missed dig. But what happens if it lands between two players? It should be charged to whoever was closest to the ball but sometimes that can be hard to judge. What happens if the attack comes off the block and lands short inside the court. The defending player who is closest to the ball does not move but a defending player who is further away, dives for the ball but to no success. Should the missed dig get charged to the diving defender or to non-moving defender, who was closer to the ball? If it is charged to the diving player, then he or she has just lowered their digging stats for a ball that they had almost no chance of retrieving and was ultimately not their responsibility. A coach will encourage effort to dive and chase down balls but equally may use those same stats to decided who starts a match. If it charged to the non-moving player then we enter the subjective world of what should have happened instead of what actually happened. What “should” have happened will depend on coaching philosophy, which may be different for different statisticians.As I said, this is the biggest headache you will encountered when deciding your own VAS.
Creating a Logical Sequence of Events and Better Data Accuracy
Events list example
I wanted to create a VAS where the events list was easy to follow, e.g. see left. To do this, every ball touch had to be counted (done – see above) and you needed rules to put in place. For example, there cannot be an assist attempt if there is no reception, dig, cover dig or free ball-overpass dig preceding it. Also, for every kill has to be followed by either an opposition block error, opposition block out, opposition missed dig or opposition free ball-overpass dig error. I also had to change the block ratings to what actually happened rather than based on attack options; block error (0 rating), block out (1), block touch (2), blocked ball (3), block kill (4). The blocked ball action variable was particularly important because now I know if there was a blocked block, it should follow with an opposition cover dig attempt. On the example on the left, if Seain Cook’s block touch was actually recorded as a blocked ball then a dig excellent by his team mate John Chapman would not make sense. If blocks was based on attack options but there might be ambiguity in the sequence of events.By having a logical sequence of events, this also enabled me to create more data accuracy checks. We have ones before like the number of service aces has to be equal to the opposition reception errors. Now, we can have simple ones like the total number of block kills has to be equal to the number of opposition missed cover digs. There can be a few more complex ones as show in the data accuracy matrix.
Ly’s Volleyball Analysis System – Definitions of the Operational Terminology
This blog post is long enough to that I won’t post up my final VAS. I will probably write another blog post briefly describing the technical skills and how to rate it. I say briefly because the full version currently stands at 6,018 words and counting! Feel free to comment on this blog post, e.g. whether you agree or disagree. It’ll be interesting to here what people think.
Comments