Measuring Goalie Performance - Beyond the Goals Against Average

In this post I will delve into the basic metrics traditionally used to measure ice hockey goalies before introducing you to some of the more advanced metrics that have been created (and are required) to more accurately quantify goalie performance regardless of their age and skill level.

The ice hockey goaltender - the unsung hero (or heel) of any hockey team. While forwards and defensemen bask in the glory of scoring goals and making big plays, the goalie is the last line of defence, tasked with stopping pucks and ensuring their team has a chance to win. Lead your team to victory and your performance is praised, battle hard but lose and rarely is the goaltending performance discussed in a positive way. For the minor hockey goalie, this is never more true. The goal that the goalie should have saved is debated far more than the missed defensive assignment that lead to the scoring chance in the first place. So what is a goalie to do?

To truly elevate your game as a goalie, to measure the progress of your child as a parent, or evaluate the capabilities of your team as a coach, understanding and utilizing the right performance metrics are crucial.

What Are Goalie Performance Metrics?

Goalie performance metrics provide a quantitative way to assess a goalie’s effectiveness within a single game and across multiple games or the entire season. These metrics can help identify strengths and weaknesses, allowing goalies to focus their training on areas that need improvement as much as highlight the areas where they are doing well. They can also provide a way to demonstrate the relative skill level of a goaltender compared to their peers, no matter how good (or bad) their team might be performing. Take a few minutes to browse the NHL or PWHL goalie statistics. Two of the most common metrics you will find to measure goalie performance are:

• Goals Against Average (GAA) - Represents the average number of goals that the goalie gives up in a single hockey game.

• Save Percentage (SV%) - Measures the number of saves made versus the number of shots faced.

As we will explain, these statistics alone are not enough to properly quantify and compare goalie performance, particularly in minor hockey where team abilities can differ greatly. The creation of more advanced metrics and how they can supplement the GAA and SV% to give a more detailed picture of a goalie's performance will be discussed in detail in this post. Additional metrics we will discuss include:

• Expected Goals Against - Represents the number of goals that the goalie was expected to give up over the given period (season, game, etc.). This is a calculation based on characteristics such as where the shot came from, if it was off a rush or a rebound, if it was preceded by a pass across the zone, etc. This metric represents the quality of scoring chances that a goalie has faced.

• Goals Saved Above Expected - This metric measures the difference between the number of goals that the goaltender has allowed versus the number of goals that were expected. A positive value indicates that the goalie is saving more than expected, while a negative value means they are saving less.

Let's now take a closer look at how these metrics are captured and what they represent.

Goals Against Average (GAA)

A common measurement used to evaluate goalie performance is the goals against average (GAA), which represents how many goals on average a goalie is giving up in a single hockey game. Considering that a goalies' objective is to prevent goals, this seems like it would be the perfect statistic by which to compare goalies. The table below shows a pair of goalies with their GAA statistics.

Name	GAA
Barb	2.30
Vicky	2.78

Based on this metric, it would seem that Barb is performing better with a 2.30 GAA versus Vicky with her 2.78 GAA. Now recall that this statistic is an average across all of the games that Barb and Vicky have played. If we include the number of games played, the following table provides a slightly different view.

Name	Games Played	GAA
Barb	2	2.30
Vicky	21	2.78

Considering the number of games for which the GAA is evaluated, would you still consider Barb to be a better goalie than Vicky? Given that Barb has only played two games versus Vicky playing 21, it is not unreasonable to suggest that Barb hasn't played enough games yet to make it a fair comparison.

Now, consider the situation below where the goalies Steve and Ted have the GAA values presented in Table 2. In this case they are both averaging more than 4 goals against per game.

Name	Games Played	GAA
Steve	21	4.03
Ted	2	4.63

Comparing only the GAA as we did earlier, Steve is performing better relative to Ted but when we also consider the number of games played, we can argue that Ted has not yet played enough games to assess if the GAA of 4.63 is a fair representation of his typical performance. Perhaps Ted played his first two games against very good teams compared to his own team and over the next 20 games his GAA will trend downwards, maybe even lower than Steve's.

Let us look at one final goalie comparison as shown in the following table, which represents a common scenario in minor hockey:

Name	Games Played	GAA
Arthur	17	4.03
Ian	20	2.91

Based on what we have been saying about the number of games played and the GAA, this table would suggest that Ian is a superior goaltender compared to Arthur, because he has a GAA that is better by over 1 goal per game and has played more games.

Now, what if I was to tell you that Arthur is a very good goalie playing on a lesser-skilled team and he has won only 3 games, tied 2 and averages 4 or more goals against per game? Compare this to Ian, who is a less skilled goalie who plays on an elite team and has won 18 of the 20 games he has played while averaging less than 3 goals per game. Considering these factors, it might be the case that Arthur has played extremely well and helped keep his team in games even though the GAA does not necessarily reflect that.

When it comes to the GAA, the true skill of a goaltender and their development progression, whether playing house league or elite level hockey, cannot be measured simply by the goals against, or the number of games they have won.

Let's now look at another goalie metric, the save percentage and how this can compliment the GAA metric.

Save Percentage

It is clear that the GAA alone does not tell the whole picture of a goalies' performance and skill. This is even more true in minor hockey where the relative competitiveness of teams can vary significantly within a given age and skill level. Another common goalie metric is the save percentage, which measures the percentage of saves made versus the number of shots faced.

For example, if a goalie faces 30 shots and makes 27 saves in a game, their save percentage (SV%) would be:

SV% = 27/30 = 0.900 or 90%

In hockey, a goalie's save percentage (SV%) is typically presented in a decimal format (0.903) versus percentage (90.3%). For a professional NHL or PWHL goalie, a save percentage of 0.900 or higher is typically the benchmark for performance, but for a minor hockey goalie, a SV% percentage lower than 0.900 is not necessarily indicative of poor performance.

Let's take a look at how this metric is typically used and how it compares and compliments a goalies' goals against average. The following table shows a set of metrics for a pair of goalies (the number of games played, the total shots, GAA and save percentage).

Name	Games Played	Shots	GAA	SV%
Wendy	21	628	2.85	0.912
Brenda	21	565	2.78	0.901

When comparing the performance of these goalies, there are some things to note:

1. Both goalies have played the same number of games.

2. Brenda has a slightly better GAA than Wendy

3. Wendy has faced a total of 63 more shots than Brenda across the 21 games.

4. Wendy has a slightly better SV% than Brenda.

Given the difference in shots, you might argue that Wendy is performing slightly better than Brenda and Wendy's team is likely not as defensively sound because they have given up more shots. Of course this assessment is subjective and just because a goalie has faced more shots, does not mean that those are "difficult" shots to stop. If Wendy is facing 2-4 more shots per game from the opponent's defenceman without traffic, then those extra shots would not be as significant.

Let's look at a more interesting example and return to the goalies Arthur and Ian from earlier, and include their total shots against and their SV% along with the games played and GAA.

Name	Games Played	Shots	GAA	SV%
Arthur	17	628	4.03	0.891
Ian	20	320	2.91	0.820

When we compare only the GAA of the two goalies, it appears that Ian is a far better goalie, giving up over one goal less per game than Arthur. But if we include the total shots for each goalie along with their SV%, the gap in performance seems far less convincing.

While Ian has a better GAA, he has faced only about one-half (1/2) the total shots that Arthur has faced in almost the same number of games. Ian averages 16 shots against per game while Arthur averages over double that at 37 shots per game. Arthur also maintains a superior SV% compared to Ian.

Now imagine you swap the goalies and they maintain their existing SV% and shots against on their "new" team. Their calculated GAA would look as follows:

Name	Games Played	Shots	GAA (Calculated)	SV%
Ian	17	628	6.64	0.820
Arthur	20	320	1.75	0.891

When presented with this data you would clearly consider Arthur to be the superior goalie. Unfortunately, it isn't practical nor truly accurate to swap statistics arbitrarily to compare goalies in this manner.

So, if GAA and SV% provide only a partial view of the performance of a goaltender, what is missing? The answer is in the measurement of shot quality and expected goals.

Expected Goals Against & Goals Saved Above Expected

As a goalie or a goalie parent, it is inevitable that you have watched a "best goalie save" video on the Internet. Search for "Top 10 Saves of the Week" (or check out this example or this example) and when you watch the video pay attention to what is common about all of the saves. Cross ice pass to the wide open player on the back-door, one-time slapshots, breakaways, net drives, goal-line grabs, or a crazy bounce. All are situations where it was "expected" that the player was going to score but the goalie somehow was able to make the save. Clearly, the goalie makes an incredible save, but pay attention to how that play develops and what leads up to that scoring chance.

The location of the shot and the actions taken prior to the shot contribute to how difficult the save is to make. A shot from the blue line that the goalie can see clearly has a lower percentage of going in the net, while the similar shot taken when there is traffic in front, the shot is tipped, or if the shot is a one-timer right after a pass from the opposite side of the ice would have a higher percentage of going in.

The higher that percentage, the more "expected" the goal is to go in.

With this in mind, let's look at the following table, which introduces some additional statistical information. The table definitions can be expanded below the table to help you understand the acronyms.

Name	GP	SOG	GA	xGA	GSAx	GAA	xGAA	SV%
Leigh	30	869	75	97.48	22.5	2.51	3.26	0.914
Sammy	30	756	62	72.17	10.2	2.14	2.49	0.918

Column Definitions

GP	Games Played
SOG	Shots on Goal across all of the games played.
GA	Goals Against across all of the games played.
xGA	Expected Goals Against across all of the games played. This metric represents the difficulty of the shots the goalie has faced.
GSAx	Goals Saved Above eXpected is the number of goals that the goalie prevented that were expected to have gone in across all of the games played.
GAA	Goals Against Average is the same as noted earlier.
xGAA	eXpected Goals Against Average is the GAA that the goalie would have had if all expected goals had gone in.
SV%	Save Percentage is the same as noted earlier.

From the table we can make the following observations:

1. Both goalies have played the same number of games (30).

2. Leigh and Sammy have almost identical SV% metrics (0.914 vs 0.918).

3. Sammy has a better GAA at 2.14 versus Leigh at 2.51.

4. Leigh has faced 113 more shots and has an expected goals against (xGA) that is 25 goals higher than Sammy.

5. Leigh has a goals saved above expected (GSAx) that is almost double Sammy's (22.5 vs. 10.2).

If we were to look at the typical metrics associated with goaltenders (SV% and GAA), we could argue that both goalies are about equal in performance, with Sammy sporting a slightly better GAA. Only after we include the expected goals, goals saved above expected and the expected GAA do we start to see that Leigh is a superior goalie across the 30 games played.

Conclusion

Tracking and analyzing performance metrics is essential for goalies looking to elevate their game. By incorporating more advanced goalie metrics into your evaluation you can better understand how the play of the team in front of the goaltender contributes to the quality of the shots they face and the types of saves that they are being asked to make. Using this data you can gain valuable insights into goalie performance, and can identify areas for improvement and implement practical strategies to enhance the required skills.

In my next post I will dive deeper into capturing and calculating these performance metrics across games and how this data can be presented to highlight goalie performance.