Introduction

Before the NHL implemented the salary cap in 2005, greedy team owners could hoard talented players leaving other financially burdened teams left with rotten scraps. This tactic would unfairly allow financially rich teams to create long-lasting dynasties and, potentially, own the Stanley Cup. Now, talent is divided more evenly throughout the NHL with the salary cap rules making for a more fruitful game. 

The salary cap era is one of the most regarded eras in the history of the NHL. Many analysts have highlighted that since its implementation the game has become faster and more strategic.  This era is an excellent time period to discover a more pure analysis of the game, especially with goaltending – a position I hope to better understand with this project. 

Hockey has been a mainstay in my life since I first watched a game on a clunky RCA television with my dad. It didn’t take long after that moment before I learned to skate and demanded to play goalie. It’s been over 20 years since that day – both the sport and position continue to be the apple of my eye. 

Goalies are the most colorful, yet misunderstood players on a hockey team. Not only do they have big, medieval-like armor, but they play as an individual in a team-minded sport. There is no other player on the ice responsible for using their entire body to effectively thwart an opponent’s scoring chance and help win a game.

For this project, I am interested in seeing just how goalie careers have shaped out in the salary cap era. How old are NHL goalies? When are they the most successful? How do their individual numbers translate into a win for the team? 

Inspiration

Micah Blake McCurdy is very well-known around the hockey community for visualizing player and team data, which are archived on hockeyviz.com. There, one can see player statistics visualized on simple area charts, to the actual ice surface. 

McCurdy’s visualizations show years of effective color practices and storytelling techniques that communicate purpose without many words. The way he incorporates both story and color into the visualizations are strategies I hope to replicate in my own work visualizing the career of a goaltender.

• 

• 

McCurdy uses color coating as a tertiary piece of data to create a stronger understanding of the story he is trying to portray. In some cases, I think it could also have helped with a viewer’s pre-assessment of the data for a more immediate understanding. Throughout this project, I will add a tertiary color layer to help the viewer quickly realize the highlights of each visualization.

Tools

Dataset

I used a dataset from inaltic.com that was sourced from hockey-reference.com – a reputable website cited regularly by the hockey community. The dataset consists of season-by-season statistics on goalies and teams dating back to 1911. 

From this dataset, I narrowed in on goalies who played between the 2006-2019 NHL seasons and started at least 10 games. This way, I could focus my project on goalies who played within the salary cap era and who have been trusted to start games multiple times. 

Tableau

To create an effective visualization of this data, I chose Tableau as the main tool to help answer my questions. It’s an effective, free visualization tool that can intuitively represent the data I’m sourcing. Tableau helped me use color throughout the visualizations and aid in my decision-making on how to display the data.

Results

How old are NHL goalies?

I thought it would be best to not only find that average or median age but also answer how many goalies are the same age and what was the average age for a goalie on each team. This comparison of ages throughout the league is a ranking relationship that directed me towards using bar graphs. I relied on age and team as categorical data for two different graphs – frequency of age would be placed on an interval scale for one graph and teams would be placed on a nominal scale for the other. 

For the first graph on the frequency of ages in the NHL, the data shaped into a bell curve that showed goalies in the league were typically between the ages of 25 and 29 years old. I added a color gradient to the bars based on the frequency to help viewers immediately understand where the majority of goalies were. This way, they would not have to only rely on the bars for realizing which ages were the most common. I chose blue to represent higher frequencies to highlight the top ages and yellow as the lower frequency since it was similar to the background color. 

The second graph was designed on a nominal scale with teams as the categorical data. I ordered the bars based on the highest to lowest average age per team. I placed an average age reference line on the graph so the viewer could further identify which teams are further from the center. With this ordering and placement of a reference line, a viewer can easily compare teams on who has the older or younger goalies. Each team was highlighted in a different color to help viewers keep track of where their teams were on the chart. If each bar was the same color, viewers would have to take more time and effort to locate and relocate their teams on the chart.

With these two graphs, it’s seen that NHL goalies are typically between the ages of 25-29 years old and more teams have younger than the average goalies.

When are goalies most successful in their NHL careers?

Identifying success for a goalie is tricky – analysts track hundreds of data points and there is constant debate on how to define their success. Most analysts rely on save percentage (SV%) and win percentage (W%) as key statistics in identifying great goalies. Save percentage is calculated by saves made over shots against while win percentage is calculated by wins over games started. These are the two statistics we will focus on when mapping the answer to our main question.

For the visualizations, I will use age as an ordinal variable and conduct a time relationship comparison to both SV% and W%. Through this method, the data should show at what age goalies tend to play at peak performance. To help further identify when goalies are succeeding, following how analysts use SV% and W%, I placed a reference line of where the average of both those statistics fell on this graph. I added a temperature color scheme to help viewers immediately recognize where the high point of a goalie’s career is. 

These visualizations show goalies are most successful between the ages of 28-31. It’s at these points where we the age group has a higher than average SV% and W%. It’s also being shown that success has a continued downward trend as goalies age throughout their careers in the NHL based on the trend line. 

There are two extremely large spikes at the end of both of these lines that I believe are statistical anomalies. Considering there are less than 5 goalies over the age of 42 who have started more than 10 NHL games, it’s safe to say that those spikes do not show that a goalie will see an improved SV% or W% when they are over 42 years old. It’s also possible that those goalies had just one excellent season when they were that age. 

How does save percentage translate into a possible win?

To answer my final question, I organized the data on a scatterplot graph to visualize the relationship between SV% and W%. By adding in the average W% and average SV% to this graph, I could identify the quadrants of success. 

1. The top left quadrant contained those who won more than average but didn’t save as many shots. 
2. The top right quadrant contained goalies who both stopped shots more than average and won more than average. 
3. The bottom left quadrant showed those who stopped shots less than average and won more than average. 
4. The bottom right quadrant showed goalies who save more shots than the average but did not win more than average.

I was curious how age played a factor in determining who could win more and save more so I added in a color temperature range that showed green points as younger goalies, yellow as medium-aged goalies, and red as older goalies.

This visualization shows that not only does age not have a direct impact on equating SV% to W%, but also that there is no direct correlation between SV% and W%. It’s rational to think that the more likely you are to stop a shot, the more likely you are to win a game. However, hockey is a team sport, and winning a game also depends on the players, coaching, and management – not just the goalie. 

Reflection

It was incredibly fun to play around with my first effort on visualizing sports stats. I now have a better understanding on how age impacts a goalie’s success and that a high SV% does not at all effectively translate into a win for the team. 

One question I am thinking about now is the relationship between the frequency of goalie ages and at what age are goalies more successful. How is it that goalies have a higher SV% and W% than average between the ages of 28-31 but most goalies are between the ages of 25-29? If teams knew that those ages were more successful, wouldn’t there be more goalies between the ages of 28 and 31? This would call for some further reflection and visualization. 

For another visualization, I would like to compare which season the goalies are in (first, second, third, etc). Which seasons do goalies have the greatest success? Does more playtime equate to higher SV% and W%? At what season do goalies start to decline? Sometimes, goalies don’t begin playing regular NHL games until they’re 25 as seen on the frequency chart. This would call for seasons to be ordered through an ordinal scale and visualize a time relationship.

What is great about data visualization is that you end up having more questions than you came in with. I also found myself feeling bewildered at my own beliefs on where goalies had a higher success rate and that age was a strong predictor of the relationship between SV% and W%. To answer these questions further, I need to add more data to the existing dataset and understand what I can do better to help visualize the answers to these new questions.