Evaluating Passing Performance in Volleyball: From Averages to Expected Sideout %

Evaluating Passing Performance

Passing is the foundation of a successful offence. However, how do we evaluate passing performance? Observation and memory are limiting because you won’t remember the outcome of every passing attempt by every player throughout the season. Using raw performance data alone, e.g. most perfect passes, can be misleading because it punishes players who have limited passing opportunities. The best way to do this is through the use of performance indicators – specific, measurable statistics that reflect how well an athlete or team is performing. These indicators should convert raw data into normalised metrics, such as percentages, averages, or ratios, that allow for fair and meaningful comparisons across players and situations.

So, what are the common performance indicators used in volleyball to quantify passing performance? Read on to find out.

Evaluating Other Volleyball Skills

This is one in a series on how I evaluate volleyball skills using various performance indicators. Click on the links below to read more.

Overall PerformancePoint-ScoringAttackPassing Serving Defence Block Setting

Passing In-Play %

Definition The Passing In-Play % is the percentage of passing attempts successfully kept in play. It is the inverse and positive spin of a Passing Error %.

FormulaPassing In-Play % = 1 – (Passing Errors/ Total Passing Attempts)Passing Error % = Passing Errors/ Total Passing Attempts

Pros * Easy to understand by coaches and players * Simple to collect data and calculate it * Highly correlated to Expected Sideout % (see link) Cons * It doesn’t consider the quality of the pass * You can’t do any advanced analytics * May lack buy-in from players and coaches

Perfect (Positive) %

Definition The Perfect % is the percentage of passing attempts that result in perfect passes. The Positive % is the percentage of passing attempts that result in both perfect and positive passes. The Positive % is usually written in parentheses.

FormulaPerfect % = Perfect Passes/ Total Passing AttemptsPositive % = (Positive Passes + Perfect Passes)/ Total Passing Attempts

Pros * Easy to understand by coaches and players * It focuses on passes that enable a high-quality attack

Cons * It doesn’t consider the other grades of passing, especially passing errors

Passing Average/ Rating

Definition The Passing Average/ Rating is a numerical rating, typically on a 0–3 or 0-4 scale, that measures the quality of a player’s serve-receive passes, indicating how effectively they allow the team to run its offence.

FormulaPass Average (0-3 Scale) = [(Perfect Pass * 3) + (Positive Pass * 2) + (Negative Pass * 1)]/ Total Pass AttemptsPass Average (0-4 Scale) = [(Perfect Pass * 4) + (Positive Pass * 3) + (Medium Pass * 2) + (Negative Pass * 1) + (Overpass * 0.5)]/ Total Pass Attempts

Pros * Every action is taken into consideration * Probably the most common way to evaluate passing performance, so coaches and players familiar with volleyball stats will understand it

Cons * It can be hard to interpret performance levels, e.g. how good is a 2.10 passing score, and how much better is it than a 2.02? * Every action needs to be graded, which is subjective and time-consuming to record * The given weights assume the difference in quality between each grade is evenly spaced. This means the jump from 0 to 1 is considered as valuable as 3 to 4. However, this is definitely not the case

Expected Sideout %

Definition Expected Sideout % is the predicted probability of a team winning the rally after a specific pass, based on the quality of the pass and historical data of sideout success rates.

FormulaExpected Sideout % = [(Perfect Pass * SO%)+(Positive Pass * SO%)+(Medium Pass * SO%)+(Negative Pass * SO%)+(Overpass * SO%)]/ Total Reception Attempts

Pros * It is probably the best way to measure passing performance, as it is based on winning rallies, which is important to winning matches * It is data-driven. It provides a more accurate, objective measure of passing effectiveness by using historical data to predict actual outcomes rather than relying on subjective pass ratings * It reflects a truer impact. It directly correlates passing quality with the team’s likelihood of winning the rally, offering a clearer picture of how passing contributes to overall match success.

Cons * It is data-intensive. To calculate an accurate expected sideout percentage, you need large, reliable, and unbiased datasets. This can be difficult and time-consuming to obtain, especially in less competitive environments or without advanced analytics tools. * Expected sideout % is competition-specific as it varies by gender and competition level, meaning you can’t use the same values across different teams or leagues * Expected sideout % may also be time-specific because sideout rates in the same competition may change over time

Actual Sideout %

Definition Actual Sideout % is the actual sideout rate when one passing attempt was involved in the rally.

FormulaActual Sideout % = Rallies Won with at least 1 pass attempt/ Total Rallies with at least 1 pass attempt

Pros * Reflects the actual impact of passing performance * Expected sideout rates are not needed

Cons * Relies on other players for better stats * It favours better teams as better teams will have a higher sideout rate

Overall Thoughts

In a perfect world, I would use Expected Sideout % as the performance indicator to evaluate passing performance. However, it takes a crazy amount of work to do it. I’ve only ever done it once. It was when I statted every available match in the Women’s Super League, 2019-20 season. It took months, but I had the time since I did it during the COVID-19 lockdown. I eventually stopped using the Expected Sideout % because it was easier to have one match report for all my teams. I also worry that the expected sideout values may change over time, i.e. the 2019-20 values may not apply to the 2024-25 season.

The performance indicator I use in my match report is a passing average. It’s because it’s commonly used and most coaches and players understand it. However, I actually don’t use it analytically. It assumes there are equal jumps with each increase in pass quality, but that is not the case. Based on 2,803 service attempts by Team SideOut, the jump from a 0 to 1 pass is 60.2% sideout rate but the jump from a 1 to 4 is 2.9% sideout rate. It is way more important to keep the pass in play than it is to make perfect passes. This is not seen in a passing average.

Instead of using a passing average analytically, I actually use a Passing Error % to evaluate passing performance. The main reason is that it is highly correlated to Expected Sideout %, i.e. players with the lower Passing Error % tend to have the higher Expected Sideout %. It means I essentially get the general accuracy of using an expected sideout % without doing all the work of calculating expected sideout values. As I don’t include it in my match report, I have to mentally calculate it. This is easy to do because my benchmark is 10%. Anything over tends to be the poor passers, and anything under tends to be the better passers. Any player reaching 5% tends to be the very best passers.

Using the Actual Sideout % does intrigue me. I don’t need to do all the work of calculating expected sideout values, but I have an accurate reflection of how valuable a player’s passing is. It also doesn’t matter that players on better teams will have a higher Actual Sideout % compared to players on worse teams, because I only need to compare players within teams and not between teams. If I want to know who my best passers are, they are all on the same team, so the sideout rate should be the same. If I want to know who the worst passers are on opposition teams for potential service targets, they are all on the same team, so the sideout rate should be the same too. I would add it to my match report. However, I tried to do it, but I couldn’t make it work. I can have a formula for the total number of rallies won, but I can’t use the work-up properties when filtering it to specific passers.

I don’t use a Perfect (Positive) % at all. The expected sideout values show that it is more important not to be aced than it is to pass perfectly. The only situation where I may use it is if you have two monster middle attackers and I need passers with a higher Perfect (Positive) % so I have more opportunities to utilise them more.