Scroll Top

Hitting Percentage In Volleyball – Is This The Best Metric?


Hitting Percentage In Volleyball – The Metrics We Use And What They (Don’t) Tell Us

Definitions of the statistics mentioned in this article:

  • Hitting % (also referred to as Attack Efficiency):
  • (Kills – Errors) / Total Attempts
  • Kill %:
  • Kills / Total Attempts
  • Error %:
  • Errors / Total Attempts

Using data to gain an objective understanding of how a team or individual performs over the course of a match or season has been around in our sport for a long time. Many of the metrics we use today are ones that have been around for just as long. But with advances in technology and more readily available understanding of statistical concepts, it may be time to reexamine our use of these metrics, and to make sure we understand what they are and aren’t telling us.

For this post, let’s take a look at attack efficiency. Attack efficiency is one of the most commonly used metrics in volleyball and as such, one that is worth understanding on a deeper level. Attack efficiency is calculated by taking Kills (k), subtracting Errors (e), and dividing by the total number of Attack Attempts (att). This metric in and of itself gives a good, general gauge of how well a team/individual performed offensively, but what is it actually measuring? Let’s use some attacking data from the Men’s Olympic Tournament from Tokyo as an example, focusing on the top 15 Outside Hitters ranked by overall attack efficiency.

POLANDWilfredo Leon8514149.4771
ITALYOsmany Juantorena9714179.4642
CANADAGordan Perrin6613121.4383
USATaylor Sander6516118.4154
ITALYAlessandro Michieletto7413154.3965
IRANMeisam Salehi7120130.3926
FRANCETrevor Clevenot10635201.3537
BRAZILRicardo Lucarelli8826180.3448
JAPANYuki Ishikawa9025191.3409
USATJ Defalco6420133.33110
FRANCEEarvin Ngapeth12443246.32911
RUSSIAEgor Kliuka8527183.31713
ARGENTINAFacundo Conte11236250.30414
ARGENTINACristian Poglajen261053.30215


While the ability to compare attack efficiencies has its merits, efficiency alone provides insight on a limited scope in two major ways:

1) By definition, attack efficiency is equivalent to taking kill percentage (k/att) and subtracting error

percentage (e/att). Combining these two pieces in this way essentially yields a plus-minus calculation

for attacking and as such, tells us more about how kills and errors are related to each other, and less

about the rate at which kills or errors are being made overall. Using a simple example to illustrate

further, two athletes with 100 attack attempts and .300 efficiency can be vastly different – one athlete may have 40 kills and 10 errors while the other has 60 kills and 30 errors. These are two very different attacking performances, but looking at efficiency alone does not show the difference between them.

2) Attack efficiency (and thereby, kill percentage and error percentage) takes only the attack into account without any consideration of how difficult the attack attempts were (i.e. first contact quality). For example, let’s take the same simple example with two athletes (Athlete 1 and Athlete 2) with .300 efficiency. Now let’s say all of Athlete 1’s attack attempts are out of system and all of Athlete 2’s attack attempts are in system. This added context sheds a very different light on Athlete 1’s .300 attack efficiency to Athlete 2’s .300 attack efficiency.

To hit on Point 1, let’s go back to our sample data from the Olympics and see how each of the top 15 Outside Hitters by efficiency rank in both kill percentage and error percentage.

POLANDWilfredo Leon.477157.0%19.4%3
ITALYOsmany Juantorena.464254.2%57.8%1
CANADAGordan Perrin.438354.5%410.7%4
USATaylor Sander.415455.1%213.6%6
ITALYAlessandro Michieletto.396548.1%118.4%2
IRANMeisam Salehi.392654.6%315.4%16
FRANCETrevor Clevenot.353752.7%617.4%19
BRAZILRicardo Lucarelli.344848.9%914.4%11
JAPANYuki Ishikawa.340947.1%1413.1%5
USATJ Defalco.3311048.1%1215.0%15
FRANCEEarvin Ngapeth.3291150.4%717.5%20
RUSSIAEgor Kliuka.3171346.4%1514.8%13
ARGENTINAFacundo Conte.3041444.8%1814.4%12
ARGENTINACristian Poglajen.3021549.1%818.9%15


* Kill % Rank and Error % Rank are taken from all Outside Hitters with a minimum of 50 attack attempts from the tournament. *

Outside of the top four players who rank well across these three metrics, we start to see a higher degree of variation between kill percentage and error percentage ranks. We can appreciate these variations even more by visualizing this table with a scatterplot.

hitting percentage in volleyball


The top 3 in efficiency form a cluster to the bottom right with high kill and low error percentages, 8-14 with the exception of FRA’s Ngapeth (higher error percentage) are grouped together with lower kill and seemingly average error percentages, and 4-7 differ more by increasing error percentage and kill percentage, with the exception of ITA’s Michieletto (lower kill percentage, lower error percentage).

One of the main advantages of considering kill and error percentage separately instead of combined in attack efficiency is understanding whether a player is scoring at an acceptable rate as well as if the player is making errors at an acceptable rate. ITA’s Michieletto is a prime example of a player ranking highly in attack efficiency, but without looking at his kill percentage, especially in relation to the other OHs in the tournament, we may not have teased out that his kill percentage was on the lower end of this group. As a coach, this helps narrow our focus to know if a player needs to find more ways to score, or make better choices to minimize errors.

This leads to expanding upon Point 2, getting a better understanding of attacking performance with more

context. Not all attack attempts are equal, and yet when we look solely at attack efficiency, kill percentage,

or error percentage, none of these metrics take the difficulty of the attack attempt into account. So how do we integrate a measure of difficulty into attacking data? There may not be one single “right” answer to this question, but let’s dive into an example of how one might be able to do so using the general concept of Joe Trinsey’s “expected FBSO” metric for rating reception.

To give a quick overview of “expected FBSO”, Joe substituted average first ball kill percentages for corresponding ratings of reception (e.g. 3 point/perfect pass = .50 kill%, 2 point/medium pass = .40 kill%, etc.) based on the level of play in question. An advantage of using a system like this, instead of an arbitrary 3 or 4 point scale, is having a data backed “baseline” of what kill percentage “should be” relative to the quality of the first contact. Over a large enough sample of attack attempts (and preceding first contacts), a direct comparison could be made between actual kill percentage and “expected” kill percentage. In the following example, we’ll use this concept to calculate an expected kill probability (pred_won), but generalized to all first contacts. A sample of the data is shown below for POL’s Wilfredo Leon.

POLANDWilfredo LeonAttackKILLGO38.6 %
POLANDWilfredo LeonAttackKILLOVERPASS58.7 %
POLANDWilfredo LeonAttackKILLGO53.5 %
POLANDWilfredo LeonAttackIN PLAYGO57.7%
POLANDWilfredo LeonAttackKILLBIC54.5 %


Now that we have attack result and a corresponding kill probability, we can calculate kill percentage as usual, but can also take the average of all the kill probabilities tied to each of these attacks. With an actual kill percentage and an expected kill percentage (EXP. KILL %), we can take the difference of the two (EXP. DIFF.) to see how far above or below a player scored relative to the quality of the first contact.

POLANDWilfredo Leon57.0%46.0%11.1%111
USATaylor Sander55.1%45.2%9.9%224
IRANMeisam Salehi54.6%46.1%8.5%336
CANADAGordon Perrin54.5%47.3%7.2%443
FRANCETrevor Clevenot52.7%45.9%6.8%567
ITALYOsmany Juantorena54.2%47.4%6.8%652
FRANCEEarvin Ngapeth50.4%43.8%6.6%7711
BRAZILRicardo Lucarelli48.9%45.5%3.4%898
ITALYAlessandro Michieletto48.1%45.8%2.3%9115
USATJ Defalco48.1%46.1%2.1%101210
ARGENTINACristian Poglajen49.1%47.3%1.8%11815
JAPANYuki Ishikawa47.1%45.4%1.7%12149
RUSSIAEgor Kliuka46.4%46.3%0.1%141513
ARGENTINAFacundo Conte44.8%45.6%-0.8%151814


Including expected kill % and expected difference to our original metrics add very insightful context to attacking performance that hitting %, kill %, and error % cannot show on its own. Of course, Leon’s kill % and hitting % are impressive in their own right, but the expected difference shows that he killed an excess of 11% of the balls than what he “should have”, relative to the difficulty of the attempts he took. The expected kill % metric also highlights the general difficulty of the attacks FRA’s Ngapeth took relative to other OHs in the tournament, and he still managed to outperform that mark by 6.6%. From a coaching standpoint, this can help shed some light on the kinds of situations players are getting set in more often, and how they are performing relative to that standard. This could also spark further investigation by breaking out attack types or in system/out of system designations to pinpoint particular strengths and weaknesses.

To wrap up, understanding the information metrics don’t tell us is just as important as understanding what they do tell us. Taking this concept into account can help us make apples to apples comparisons when using data, or at the very least, be more keenly aware of the limitations of the metrics we use.

Follow Us
Most Viewed
Training Setters

Perhaps no other skill in volleyball is taught so differently...

6-3 Offensive System

The 6-3 Offense in volleyball Young or new volleyball teams...