thevolleyballanalyst Statisticians’ Manual 2014/15

*** Update 27 January 2021 | You can read the latest thevolleyballanalyst Statistician Manual 2019/20 on this link ***

Previous Manual

Table 1 shows brief operational definitions of a volleyball statistics I collected last season.  I created it for my dissertation where it allowed me to calculate six different types of performance indicators (Table 2).  I tested them on whether they could differentiate between winning and losing teams, i.e. if they are valid performance indicators (see link for what I actually found out).  Figure 4 shows the best performance indicator for each technical skill and the magnitude of its validity as a performance indicator.

I also used this VAS last season for my analysis at Polonia.  There were some strengths in the system, e.g. I found the service and reception ratings to be very robust.  I also found weaknesses in some of the technical skills that were measured, e.g. evaluating setting using the Assist system (see link for more information).

Ly’s Simplified Volleyball System

So based on what my dissertation results showed and some reflection on its practical use at Polonia, I decided to simplify the system for the 2014/15 season.  Below is an image of the current VAS.  The changes included:

* Simplifying attack to kills, attack zeros and errors because I only needed to calculate a hitting pct. * No more block touch action variable. * Setting is based on how many options the attacker has. * Simplifying digs to just digs and missed digs.  This was due to digs in-play percentage being the best valid performance indicator.  I also included cover digs as part of digs.  This is because I found the in-play percentages of digs and cover digs to be very similar. * Simplifying free ball-overpass digs to just free ball-overpass digs and free ball-overpass dig errors.  There was no point evaluating the quality of the free ball-overpass digs because generally, it would perfect or excellent.

Another change is the inclusion of a free ball (no attack attempt).  This was because I felt it should not be calculated as part of a player’s or team’s attack attempts.  It also allowed me to differentiate between free balls and overpasses that resulted in a kill or error.  Previously, those types of kills and errors was statted as a normal attack kill or attack error.  I didn’t like this because it muddled a player’s attack statistics.  Maybe a player may have a higher hitting percentage because they had two lucky free balls that resulted in a kill.  Also, it gets away from a libero having a singular kill on their stats (how can a libero get a kill?) because they had an overpass that resulted in a kill.  In practice, I found that these free ball stats are useful at lower levels of play because free balls occur far more often than actual attack attempts.  So I may select players on how well they attack but by how well they put away free balls.

Other changes include the modifying the operational definitions of some action variables.  For example, previously for every block error, there has to be an oppositional kill.  I didn’t like this because there were situations when the attacker hit the ball out (an error) but was turned into a kill because the blocker touched the net.  So now I have changed the definition so if the attack attempt would have resulted in a kill regardless of the block kill, e.g. it hit the floor, then I awarded the player with a kill.  If it didn’t, then I would have awarded the player with an attack attempt (even if it would have resulted in an error).  

Conclusion

I like this system for several reasons.  The simplified system makes it easier to stat because I don’t need to rate the quality of the attack and digs.  It also allows me to measure the attack in a lot of different ways.  For example, In StatEasy I can calculate the hitting percentage of a player after an excellent set.  This is a purified view of a player’s attacking effectiveness because it does not include any attack attempts after a good set (this can lower a player’s overall hitting pct), any second ball attack attempts (especially applicable to setters – opposites, e.g. Ben Pipes, Alex Minchin) and any attacks from an opposition overpass (this can improve a player’s hitting pct).  The only drawback of this system is I had to create new stats and stat report in StatEasy and also build a new Excel spreadsheet to accommodate the new stats.