by John Barr | Nov 6, 2021 | Data, Seattle Kraken, Uncategorized
Through the first eleven games of their inaugural season, the Seattle Kraken find themselves at 4-6-1 and at the bottom of the Pacific Division standings. We at Sound Of Hockey are bullish on how they have been playing lately, so we decided to dig into the numbers to identify any trends and validate if our sense of optimism is justified.
Goals scored and goals against
Through the Nov. 5 games, Seattle is tied for 19th in average goals scored per game and 25th in average goals scored against. Those numbers reconcile with their record, but the goals against number is a bit surprising considering most media outlets forecasted the Kraken would be a strong defensive team.
Let us look at these same numbers by game so we can evaluate if the Kraken are trending in the right direction.
Even with Edmonton scoring five goals against the Kraken in game 10, the average goals against has been declining. This is a trend we like to see.
The goals scored is slowly trending in the right direction as well. Up until the game on Thursday against Buffalo, the team was feeling snakebitten after getting Shesterkinned against the Rangers last Sunday and having several near misses against the Oilers on Monday.
Shots taken and shots against
At 29.5 shots per game, the Kraken are 25th in the league, which correlates very well with their goals per game. On the other side, they are averaging the lowest number of shots against in the league which should be considered a very positive sign.
Here are how those numbers project by game.
These trends should give Kraken fans a sense of optimism. Shots for (taken) per game have been on a steady incline while shots against have been on a steady decline. The Kraken have outshot their opponent over the last six games, and if they can maintain that trend, they should start winning more games than they lose.
High-danger and medium-danger shot quality
Of course, it is not always about quantity, as sometimes you need to consider the quality of the shots being taken. To analyze the quality of shots we’re going to look at five-on-five high-danger and medium-danger shot counts by game.
The concept is relatively simple. A high-danger shot has a higher likelihood to go in than medium- or low-danger shots. We’re probably stating the obvious here, but ideally you would have more high-danger shots than medium- or low-danger shots to increase your chances of scoring goals. This data was pulled from NaturalStatTrick.com, a valuable resource if you ever want to know more about the advanced analytics of hockey.
This trend should be considered positive and could be the impact of the “learning to play together” narrative of an expansion team that is compiled of a bunch of players that had limited playing time together before this season.
Wrapping it all up
This is a case where the eye test is aligning to the analytics. We do not like to get pollyannaish about the team’s outlook, but there are some objective signs in the analytics that this team is getting better and is trending in the right direction. It will be interesting to continue to monitor these trends and see if the goals and therefore wins start to follow.
If you have any questions, insights, or feedback, let us know in the comments section.
by John Barr | Jun 12, 2021 | Data, Uncategorized
One of the points of discussion during the Stanley Cup Playoffs is the quality of officiating. We have heard from columnists, coaches, and plenty of fans complaining about the calls going against them this year. This is a common complaint every year.
A few months ago, I published a Data Dump piece on penalties that evaluated how often they were assessed in games and when they were most frequently called during the regular season. For this week’s Data Dump, I wanted to revisit that piece with a focus on comparing regular season penalties called to those called in playoff games.
Penalties called per 60 minutes
I was a surprised to see that the rate of penalties called in playoff games is higher than regular-season games.
But whenever you look at data, it is important to add context to the figures. Like in my previous article, I used the goal differential to provide context to these penalties.
Penalties called where the goal differential is two goals or less seems rather consistent between regular season and playoffs. It is in the blowout game situation (goal differential of three or more) where things seem to get a bit out of hand. If you watch playoff hockey on a regular basis you can formulate why this happens, as teams tend to “send messages” or get frustrated and start committing penalties when the game becomes one-sided.
Penalties called by scenario: leading team vs trailing team
When we looked at the regular season trends a few months ago, there seemed to be some indicators that referees were more likely to call a penalty on teams that were leading a game versus trailing a game.
When we isolate the playoff games, there seems to be less variance between the penalties called on leading versus trailing teams.
The last two playoff seasons appear to have little to no difference in the number penalties called. Before we wrap this up and claim there is no bias in calling penalties if a team is leading versus trailing in the playoffs, let’s add some score differential context.
In the chart above, the difference between penalties called on the leading versus trailing team seems very consistent in the regular season, regardless of the goal differential. Let us look at the playoff games.
In the chart above, the rate at which penalties are called is flipped in games where the goal differential is three or more. The data shows that penalties are more likely called on the trailing team when the goal differential is three or more. As previously stated, the reasons for this are the “message sending” and frustration-type penalties.
Penalties called in the third period
One of the most compelling visuals in the regular season post I did in March was the chart showing penalties called in the third period based on goal differential. Here is the updated visual with the breakdown.
One of the takeaways I see in the chart above is that the leading team seems more likely to be called for a penalty in the last five minutes of a one-goal game than any other time in the game. Now we can look at how this same data looks in the playoffs.
The large discrepancy of penalties in the playoffs in the last five minutes of a one-goal game between the leading and trailing team is no longer present. What jumps out at me in this chart is the spread in the two-goal game. In the first 10 minutes of the third period in a two-goal game, the leading team is much more likely to get a penalty called on them (about 61 percent) compared to the trailing team (about 39 percent).
Conclusions
Directionally, there are some interesting things in this analysis of penalties called in the regular season versus playoff games, but before we draw any hard-and-fast conclusions, here are a few other things to consider:
- Because only the top-16 teams qualify for the playoffs, there should be more parity in playoff games compared to regular season games, which could explain some of the variance when comparing the two types of contests.
- There is also a much bigger sample size when looking at the regular season versus the playoffs. The higher volume of regular-season games will smooth out any anomalies that could create higher variance with a smaller sample size.
- There are always other dimensions that could be added to the analysis to gather more intelligence about the data (i.e., type of penalty, player personnel deployed, etc.).
I hope this gives you a little deeper understanding of penalties called during the Stanley Cup Playoffs as opposed to just believing a coach’s remarks in a post-game press conference or a tweet you might have seen from a prominent journalist.
by John Barr | May 22, 2021 | Data
In my world, nothing beats overtime hockey in the Stanley Cup Playoffs and without a rooting interest in this year’s Playoffs, I can enjoy it without having heart palpitations. This year’s Playoffs kicked off with three straight overtime games and last night the hockey gods treated us with two more overtime games. For this week’s Data Dump Saturday, I am going to look at how this Playoff season is tracking in various categories to previous years.
Overtime games in the Stanley Cup Playoffs
As the chart shows, this year we have seen an above average number of overtime games through the first 22 games with eight. It is important to call out that the NHL categorizes last year’s play-in round as playoff games and for the purposes of this post, they are also categorized as playoff games.
Average goal differential
A playoff season with more overtime games is naturally going to have a lower goal differential compared a to playoff season with less overtime games so the lower goal differential this season should be expected.
Empty net goals in the Stanley Cup Playoffs
Historical goal differential might be a bit misleading considering the trend of pulling the goalie earlier in the third period to get an extra attacker when a team is down. By the down team pulling their goalie earlier provides more opportunity, i.e. time, for the team that’s up to score and thereby increase the differential.
Here is a look at the empty net goals scored in the first 22 games.
This increase of empty net goals will contribute to a higher goal differential that might mislead to show that games are not as close as they really were. Game analytics support pulling goalies earlier leading to overtime games and thereby improving a team’s chances of winning. For more on this I would suggest you check out the great work by Meghan Hall on pulling goalies.
Summary
Between the quantity of overtime games and the smaller goal differential, it appears there is a parity across the teams. The divisional playoff format might explain some of this parity. Traditionally, the best regular season team in the conference plays the lowest qualifying team in the conference, a one seed vs eight seed. The 2021 Stanley Cup Playoff format has the one seed playing the four seed inside the division. Mathematically this will create match ups of teams of closer ability. However, with a regular season schedule that was made up of exclusively divisional opponents, it is difficult to draw any conclusions from this format, but it makes you wonder if we should see this division playoff structure in the future.
For now, let’s just root for more overtimes and hope for some game sevens.
by John Barr | May 20, 2021 | Data
Every NHL Stanley Cup Playoff season I enjoy digging into rosters to see how teams compare across relatively basic demographic information such as nationality, age, and acquisition type. As they say, sharing is caring. (Note: data sources are a combination of capfriendly.com and various statistical reports out of NHL.com.)
Average age across Stanley Cup Playoff rosters
Stanley Cup rosters by nationality
The league is roughly 45% Canadian, but variance across certain teams can be profound. Here is the breakdown of nationality by team.
How the players were acquired
This is a fun view on how these teams were built.
An interesting callout for Vegas is that Nicolas Hague is the first draft pick to make an appearance in the Stanley Cup Playoffs for the Golden Knights. This is his fourth season since being drafted and should be a good example on setting expectations for Seattle Kraken NHL Entry Draft picks in July.
Salary cap by playoff team
Technically there is no such thing as a salary cap in the Stanley Cup Playoffs, but this year playoff salaries are getting a bit more discussion. The defending Stanley Cup Champion Tampa Bay Lightning brought back Nikita Kucherov from long-term injured reserve just in time for the playoffs, pushing the team way over the limit of what would be the cap. But again, there is no salary cap for the playoffs.
I hope this gives you a little more insight into the teams in the Stanley Cup Playoffs this year. If you have questions about the data or some additional angles you would like me to consider, let me know in the comments section.
by John Barr | May 10, 2021 | Data, NHL Expansion, NHL Expansion Draft
Last week, James Mirtle of The Athletic mentioned that Chris Driedger is a goalie of interest for the Seattle Kraken. Mirtle is hardly the first person to mention Driedger’s name in the same sentence with the Kraken, but he was the first to reference “sources.” Driedger is having a good season as a platooning goalie for the Florida Panthers and will become an unrestricted free agent at the end of the season. This makes him an interesting candidate to be one of the inaugural season goalies for the Kraken.
My concern is Driedger has only seen limited action in the NHL and based on the small sample size, we might not have enough games to assess his long-term capabilities. When I think of older goalies with a small sample size, I immediately think of Scott Darling as a cautionary tale. Darling’s first 20-game season came when he was 26 years old.
Darling had two good seasons in Chicago as a platooning goalie over the 2015-16 and 2016-17 seasons. He then signed a lucrative four-year contract with the Carolina Hurricanes where he struggled mightily and played just 51 games over the next two seasons. His contract was eventually bought out after the second year of the contract. Worth noting, Seattle Kraken general manager Ron Francis was the one who signed Darling, and there’s no question that this deal has lived on in his memory.
For this week’s data dump, I want to look at how Driedger stacks up to comparable goalies that entered the league around the same age as he is now.
Sizing up Chris Driedger
Chris Driedger will be turning 27 years old next week with 37 career NHL games under his belt. This year he has played 22 games with a .923 save percentage, 2.17 goals against average, and a record of 13-6-3. Last year he split time between the NHL and AHL, and when playing with the big club performed well in his 12 NHL games. Driedger also played one game in each of 2014-15, 2015-16, and 2016-17 but remained in the minors for 2017-18 and 2018-19.
Goalie comparables
A goalie’s lifespan in the NHL is relatively short (that’s a post for another day). As such, coming into the league at 25+ years old is late in a goalie’s career. For the purpose of finding a sample of older goalies that entered the league late, I am going to isolate all netminders who were at least 25 years old at the start of their first season where they played at least 20 NHL games.
There were 33 goalies that fit this profile. Driedger has the highest career save percentage of the group, but that could be a bit misleading considering a lot of these players went on to create larger sample sizes and are either further along in their careers or are already out of the league. Naturally, goalie skill and therefore save percentage declines toward the end of a career.
Trends over time
Now that we have our comp list, let us try to visualize the trends of these goalies’ careers. First, we look at how many goalies would continue playing at least one game in the NHL after their first 20-game season.
Three goalies played 20+ NHL games in their first season and then none the very next season.
To put a finer point on it, let us look at the number of goalies that played at least 10 games in the subsequent seasons of the 33 comparable players.
The drop between year one and two does not seem material, but the drop in year three is significant. Only 36% of the 33 players would play 10 more NHL games just three seasons later. Given this information, any deal you sign with a goalie from this age group should have a limited term of under three years.
Adding save percentage
We will now add save percentage into the mix to see how well these goalies performed in the games played.
Now we can start to paint a picture of who was/is successful following their first season of 20 games or more. For some of the players that are still active, we may not have enough service time to get a true determination of whether they have been successful following their first 20-game season.
I went ahead and bucketed all 33 players into four categories:
- Success
- Ok
- Cautionary tale
- Too early to tell
This was completely subjective and was based on how many seasons the goalies played a minimum of 10 games, the number of games they played, and their save percentage.
Here is the count of goalies in each category.
If you strip out the “Too early to tell” group, only five out of the remaining 22 goalies were a success. More succinctly, only 23% of the goalies in this category are determined to be successful.
This analysis still lacks statistical context and goalie scout insights. The broader point is that with only 37 NHL games in his career, Driedger’s long-term success is not a slam dunk for Seattle and the team should proceed with caution, should they draft and sign him to a longer-term deal.
Let me know your thoughts. How would you predict Chris Driedger’s viability on being a No. 1 starter in the league?
by John Barr | May 1, 2021 | Data, Uncategorized
There has been a lot of talk about the battle for the Hart Trophy, awarded to the NHL’s MVP, between Auston Matthews (TOR) and Connor McDavid (EDM). For this week’s Data Dump Saturday, I want to dig into the numbers between these two players to demonstrate the thought process when evaluating who is truly the best.
The Basics
Points, goals, and assists are the most basic level of statistics when evaluating a player so let us start there to establish a baseline for comparison.
Next, let’s consider the number of games played. As of now, Toronto has played 50 games while Edmonton has played 48. Matthews also missed four games due to injury so we should consider average points, goals, and assists per game to make it a more level playing field.
There is no material change, but averages are better for comparing players with a different number of games played.
By Opponent
Now let us peek at what teams these players are feasting on this season.
Both players are taking advantage of the Ottawa Senators, averaging two points or more per game. Additionally, McDavid must love facing the Winnipeg Jets, while Matthews does pretty well against Montreal. These matchups might be worth noting because coincidentally the first-round Stanley Cup Playoff matchups appear to be Toronto versus Montreal and Edmonton versus Winnipeg.
Add more context
Any goals are valuable, but there are certain scenarios where it is easier to score goals, like on the power play. To even this out let us look at the point production without the power play points.
This starts to level out the scoring, but McDavid still leads Matthews in points when we just isolate even strength points.
Another way we can add context to the point production of these two players is by looking at what period of the game they are scoring their points.
One may think that scoring more at the end of the game could be considered more meaningful. As such, one might think that McDavid is more clutch by scoring more points late in the game, right? Not necessarily. We are missing context. What if Toronto tends to enter the third with a bigger lead than Edmonton does in its games? If Toronto establishes a lead in a game early then this would enable them to play a more defensive-minded game later, thereby limiting the opportunity for Matthews to score. I do not know if this is the case, but these are contextual questions that might be asked when evaluating players.
Game situation
The last comparison I am going to look at is the game scenario in which the player scores a goal. This is not perfect, but I want to try to isolate more meaningful goals based on what is the situation in which the player scores.
To do this, I am going to show the goal differential after the player scores. For example, if the Toronto Maple Leafs are down 3-1 and Auston Matthews scores a goal to bring the score to 3-2, the goal differential is -1. It would be considered a more meaningful goal than if Matthews were to score when the game is 5-0.
Assuming you believe the smaller the goal differential at the time the goal is scored, the more meaningful the goal, Auston Matthews scores more meaningful goals. 58% of his goals tie the game up or put his team up by one goal compared to 50% of McDavid’s goals.
My intent of this week’s data dump is just to showcase the thought process in evaluating players and their commonly reported stats. Context is very important to evaluate players and will be very important for the Seattle Kraken when considering who to select for the Expansion Draft and beyond. I hope this sheds a bit of light on the topic and would love to hear any ideas or thoughts on how you would evaluate players in the Expansion Draft.
