Hockey Analytics

A friend recently reached out to me on Twitter, after I had emailed him a copy of my senior thesis, asking me if I’d be interested in writing a blog post explaining some of the advanced stats or analytics used in the NHL right now.  I was absolutely interested in doing so, and was really excited to, especially since I was already planning to write something on how the NHL is changing in regards to the youth that are in the league currently and how the game has changed in the past five or so years.

Let me start by saying that I am a big fan of statistics and numbers and my senior thesis was driven by various statistics, which I may at some point publish to the blog.  Hockey is one of the sports most driven by numbers, along with baseball, because understanding the numbers behind what works and doesn’t work on the ice or the field can have a major impact on a team’s success compared to other sports at this time.  For instance, in baseball, something as simple as playing a shift towards the first base side of the field against a left-handed batter who pulls the ball a majority of his at-bats can have a major impact by reducing that batter’s success at getting on-base.  Furthermore, baseball analytics have developed to the point where we can know what a pitcher’s spin rate is on specific pitches and now know how large of an impact that this number can have.  This number is measured in revolutions-per-minute and the faster the ball is thrown, the more revolutions there are, the increasing the swing-and-miss rate.  Baseball is a really easy sport to see how the numbers work, simply because it is a back-and-forth sport; the pitcher throws the ball, the batter attempts to hit the ball, if he makes contact, the defense tries to convert the batted ball into an out and, if that doesn’t happen, the ball is returned to the infield, then the whole process starts over again.  One would probably think that football would be an easy sport to apply numbers to and, in a sense it is, in that teams try to determine the probability of success for plays run on offense to try to exploit holes in the opposition’s defense.  For example, passes thrown between one and ten yards will always have a higher probability of success than those thrown fifteen yards and beyond.  Teams also try to gain a certain number of yards on first and second down, knowing that the more yards the team gains on those downs, the higher the probability that the team converts a first down.  However, football also isn’t a metrics-driven sport in that there is really no need for a possession stat and it isn’t as fast-paced as hockey.  It is more similar to baseball in that both are chess-like sports; the offense calls a play while the defense attempts to stymie that play and, when the ball is snapped, the team that had the better play call typically will have the desired outcome. Football has a lot of catching up to do in the analytics field and the application of them to the sport in order to be at the same level as baseball or hockey, but there have been movements in this direction, especially when the Cleveland Browns hired former LA Dodgers General Manager, Paul DePodesta, to be their Chief Strategy Officer.  A hiring like this could have great impact on how the Browns organization drafts in future years and how the team pursues free agents and could change the future of the sport as a whole.  However, it will still be different than the metrics used to measure statistics in hockey, a sport that is incredibly fast-paced and one in which teams try to quantify all aspects of the game.

Because hockey is such a possession-driven sport, ultimately everything a team does is in effort to increase a team’s puck possession.  The most basic understanding of hockey analytics has to surround the notion that the higher puck possession numbers a team has, the higher the likelihood that the team will be successful and win games.  This is simple, because if a team doesn’t have control of the puck, it is nearly impossible for that team to put shots on net, and therefore score goals.  There are two ways to measure puck possession and they are fairly interchangeable with each other and the of which metric to use is a personal decision.  The two measurements are Corsi and Fenwick.

Puck Possession

Let’s start at the most basic of ideas regarding puck possession:

  • Puck possession measures how much a team or a player possesses, or controls, the puck
  • This can be expressed as time of possession (the Chicago Blackhawks controlled the puck for 42:15 in tonight’s game) or as a percentage (the Vancouver Canucks possessed the puck 38% of the time in Tuesday’s game)
  • The NHL has only recently started tracking possession numbers and is still behind where it needs to be

Fenwick: Shots + shot attempts that miss the net (shots that hit the crossbar/posts, shots that went wide of the net)

Corsi: Fenwick + blocked shots

  • The idea behind Fenwick and Corsi is that the more time a team controls the puck, the more opportunities that team has to score
  • Both Fenwick and Corsi can be represented as a counting stat (Team X had 57 Corsi events in last night’s game) or as a percentage (Team X had a Corsi For of 52% in tonight’s game, meaning that they controlled the puck 52% of the time)
  • Both can be broken down into For/Against stats (Corsi For/Corsi Against), (Fenwick For/Fenwick Against) – abbreviated as CF/CA, FF/FA
    • Corsi/Fenwick For – the events that make up these metrics occurring for a particular team
    • Corsi/Fenwick Against – the events that make up these metrics occurring against that team
    • Having a higher CF/FF than CA/FA is what teams strive for – this is how a CF% above 50 occurs
  • Both can also be broken down by events per 60 minutes (or per regulation game) or by 20 minutes (per period)
    • This presents a more manageable number than a number that is in the thousands for an entire season

2015-16 Season Highest Team Corsi For Percentage

    • Los Angeles Kings – 56.4
    • Pittsburgh Penguins – 52.7
    • Dallas Stars – 52.6
    • Nashville Predators – 52.5
    • Anaheim Ducks – 52.5
    • Tampa Bay Lightning – 52.2
    • St. Louis Blues – 51.7
    • Detroit Red Wings – 51.7
    • San Jose Sharks – 51.5
    • Montreal Canadiens –  51.5
  • Only Montreal failed to make the playoffs out of these teams and that was largely due to the absence of Carey Price in net.  The value of puck possession cannot be overstated

2015-16 Season Lowest Team Corsi For Percentage

    • Colorado Avalanche – 44.2
    • New Jersey Devils – 46.2
    • Arizona Coyotes – 46.8
    • Vancouver Canucks – 47.2
    • New York Rangers – 47.4
    • Buffalo Sabres – 47.5
    • Ottawa Senators – 47.5
    • Minnesota Wild – 47.7
    • Columbus Blue Jackets – 48.0
    • Calgary Flames – 48.1
  • Only one team, the New York Rangers, made the playoffs out of this group and they were dismantled by the Penguins in five games; The rest of the teams were largely considered some of the worst teams in the league

2016-17 Season Top Five Teams in Corsi For Percentage (As of 11/26/16)

    • LA Kings – 54.4
    • Carolina Hurricanes – 53.7
    • Philadelphia Flyers – 53.5
    • Boston Bruins – 53.4
    • San Jose Sharks – 53.2
  • Three of these five teams were playoff teams last season, with Carolina and Boston having missed them. Carolina has been trending upwards since Bill Peters took over as head coach and Boston has typically been a strong puck possession team under Claude Julien. Unsurprisingly, the Kings are leading the league again in this category.

2016-17 Season Bottom Five Team in Corsi For Percentage (As of 11/26/16)

    • New York Islanders – 44.9
    • Arizona Coyotes – 46.2
    • New York Rangers – 47.1
    • Detroit Red Wings – 47.3
    • Buffalo Sabres – 47.9
  • Both New York teams and Detroit were all playoff teams last season, but struggled once they got into the playoffs. This season, all teams but the Rangers are really struggling to win games, with the Rangers leading the Metropolitan Division. If the Rangers continue to struggle with puck possession the way they have been, it could be tough for them to continue to win games the way they have been.


  • Los Angeles continues to be a really good puck position team with Darryl Sutter as their head coach.  Since he became their head coach during the 2011-12 season, he has continually had the Kings near the top of the league in terms of Corsi For Percentage.
  • Under Bill Peters, Carolina has been trending upward since the 2014-15 season, his first as a head coach in the NHL, and it continued last year.  Even though the Hurricanes are struggling in the standings, the fact that they continue to be near the top of puck possession is a good sign for a young team.
  • Arizona and Columbus are both at the bottom of the standings in the league, which was expected of the Blue Jackets, and the Coyotes have seriously underachieved this season.

Shot Quality

  • An issue with Corsi and Fenwick is that neither consider the quality of shot, just the number of shots attempted
  • The idea behind shot quality is that some shots are more difficult than others for goaltenders to stop
    • Clear shots from the point without traffic are easier to stop than a rebound from 10 feet away with bodies around the crease
  • The question is, how possible is it to maintain shot quality throughout an entire season or across multiple seasons?
    • If a team or a player maintains a high shooting percentage across a large time frame, then shot quality exists
    • If a team or player has a high shooting percentage consistently, then that player or team has figured out a way to find locations that yield high returns for shots


  • One way to determine trends in a player’s shot quality is by looking at a team’s on-ice shooting percentage when a particular player is on the ice in 5-on-5 situations (we always use even strength 5 vs 5 situations because powerplay, shorthanded, or even 4 vs 4/3 vs 3 situations skew the statistics)

2011-2016, minimum of 4,000 minutes played, only forwards

  • Highest on-ice team shooting percentages when that player is on the ice, with number of goals scored by that player in that span (also a lockout occurred in the middle of this span, making goal totals seem lower than I thought they would be)
    • Jiri Hudler – 17.68% – 70 goals
    • Steven Stamkos – 16.79% – 111 goals
    • Evgeni Malkin – 14.01% – 80 goals
    • Milan Lucic – 13.77% – 72 goals
    • Martin St. Louis – 13.68% – 64 goals
  • Lowest on-ice team shooting percentages, with goals scored during span:
    • Daniel Winnick – 5.60% – 30 goals
    • Dustin Brown – 6.32% – 48 goals
    • Jarret Stoll – 6.47% – 22 goals
    • Ryan Callahan – 6.52% – 43 goals
    • Mikko Koivu – 6.54% – 31 goals

The difference between these two groups of players may not seem like a lot, but it is a substantial difference over the course of a season.  I looked at the difference between 6% and 13% over a season for a team, even though 13% is at the lowest end of the higher group and 6% is around the average for the lower shooting group.  I set the number of shots at 2,000 for a team, which is a fairly common (and level) number to work with.  The difference between these two shooting percentages is 120 goals at 6% and 260 goals at 13%, which is a 46% increase in goals, which is certainly substantial.  I understand that a 13% shooting percentage for a team over an entire season is fairly unsustainable, but most seasons there also aren’t teams with shooting percentages as low as 6%.  To not believe in shot quality means that there is virtually no difference between the two groups of players and the better shooting group of players achieved this shooting percentage by luck.

Shot Quality for Teams

Shot quality for teams is much more difficult to predict or determine trends across seasons than it is at the player level.  This is because a team’s roster can vary greatly season-to-season and it would be nearly impossible for a team to consistently dress a roster full of high shooting percentage players; this would also be very expensive.

2015-16 Top Shooting Teams, All Situations:

    • Dallas Stars – 10.10%
    • New York Rangers – 9.97%
    • Ottawa Senators – 9.89%
    • Washington Capitals – 9.88%
    • Florida Panthers – 9.84%
    • Calgary Flames – 9.55%
    • San Jose Sharks – 9.51%
    • New York Islanders – 9.41%
    • Chicago Blackhawks – 9.37%
    • Tampa Bay Lightning – 9.33%
  • All of these teams also ranked in the top ten in goals for last season except Ottawa (12th), San Jose (11th), and Chicago (20th)
  • A high shooting percentage doesn’t always correspond with strong possession, however, as only Dallas and Tampa Bay also ranked in the top ten in Corsi For Percentage last season

2015-16 Bottom Shooting Teams, All Situations:

  • Toronto Maple Leafs – 7.63%
  • Carolina Hurricanes – 8.00%
  • Vancouver Canucks – 8.05%
  • Buffalo Sabres – 8.23%
  • Philadelphia Flyers – 8.31%
  • Edmonton Oilers – 8.34%
  • Los Angeles Kings – 8.50%
  • Detroit Red Wings – 8.59%
  • Montreal Canadiens – 8.63%
  • Anaheim Ducks – 8.66%
  • Winnipeg Jets – 8.74%


  • 5 vs 5 shooting percentage + 5 vs 5 save percentage
  • The mean across the league would be 100% but slight variations take place
  • Can be used to measure team or player
  • Can take into effect puck luck, but one must consider the skill level of the goaltender a player is playing in front of
  • Ultimately though, PDO must be considered while recognizing the skill levels of all players on the ice
  • PDO takes into account the quality of a team’s goaltending or a skater with a high shooting percentage
  • Team PDO is easier to find statistics for

2015-16 Top Ten PDO:

    • Washington Capitals – 101.7
    • Florida Panthers – 101.6
    • New York Rangers – 101.4
    • Chicago Blackhawks – 101.2
    • Tampa Bay Lightning – 101.0
    • New York Islanders – 101.0
    • Ottawa Senators – 100.9
    • St. Louis Blues – 100.8
    • Pittsburgh Penguins – 100.7
    • New Jersey Devils – 100.5
  • 9 out of the 10 teams made the playoffs last season, with only Ottawa missing them.  The Senators high PDO was due in large part to their extremely high shooting percentage (nearly 10%) since they only had a .910 save percentage

2015-16 bottom ten PDO:

    • Toronto Maple Leafs – 98.1
    • Carolina Hurricanes – 98.2
    • Calgary Flames – 98.7
    • Edmonton Oilers – 98.8
    • Winnipeg Jets – 99.0
    • Montreal Canadiens – 99.0
    • Vancouver Canucks – 99.1
    • Columbus Blue Jackets – 99.2
    • Nashville Predators – 99.4
    • Detroit Red Wings – 99.6
  • Only Nashville and Detroit made the playoffs out of these teams last season

Sample Size

  • When consulting hockey statistics, one must be careful to consider the sample sizes of said statistical categories.  This is because when compared to other sports, (points in the NFL and NBA and runs in MLB) goals in the NHL are a relatively rare occurrence.  Therefore, a hot or a cold start or streak or even injuries can greatly affect a statistic.  Take this season, for example: Panthers forward Jonathan Marchessault has 9 goals scored in his 20 games played, but had only 7 goals all of last season.  Is he able to maintain a 40 goal pace? Probably not, but he is also relatively young still with limited NHL experience, so scoring between 25 and 30 could be achievable.  Another example is Michael Grabner who has 12 goals in 22 games, yet only scored 11 goals last year and hasn’t hit the 20 goal mark since the 2011-12 season.  Lastly, I will look at Auston Matthews, the first overall pick in this year’s draft, who had 4 goals in his debut but has just 4 goals since.  That is perhaps the most extreme example of a hot start to a season.  It generally takes an entire season or even multiple seasons to create a quality sample size to determine a player’s trend.  So if the three players mentioned above only score five more goals the rest of the season, then their hot starts don’t look quite as good at the end of the season. Applying this idea of sample size to analytics means that metrics such as Corsi/Fenwick are two of the most believable measurements because they utilize larger sample sizes than others (shooting/save percentages) and they can produce truer trends.

Game Score

Playing with the Lead

  • A game’s score can have a decent size impact on stats and how games are played.  A team with a lead typically gives up more shots than if the score was tied.  This is because often the leading team will play more conservatively, collapsing down around the net, in an effort to protect the lead.  This team, however, will also usually have a higher save percentage than when the score is tied or when that team is trailing.  This elevated save percentage occurs because the shots faced by a goaltender with the lead are often of low quality.  The higher save percentage occurs simultaneously with a low shooting percentage when trailing.

2015-16 5 vs 5 overall Corsi For Percentage, top teams:

  • Los Angeles Kings – 56.4
  • Pittsburgh Penguins – 52.7
  • Dallas Stars – 52.6
  • Nashville Predators – 52.5
  • Anaheim Ducks – 52.5

2015-16 5 vs 5 Corsi For Percentage when leading, same five teams and where they ranked:

    • Los Angeles Kings – 51.0, 1st
    • Pittsburgh Penguins – 47.9, 6th
    • Dallas Stars – 47.4, 7th
    • Nashville Predators – 45.5, 14th
    • Anaheim Ducks – 47.3, 8th
  • There was only one team in the league with a Corsi For Percentage above 50.0% when leading last year, the LA Kings, who are always one of the best puck possession teams.  There were only two playoff teams in the bottom ten of the league when trailing, the New York Rangers and Minnesota.
  • Shooting percentages, in general, increase when teams have a lead because they attempt less shots, a result of controlling the puck less.

2015-16 Average shooting percentage, all 5 vs 5 situations:7.49%

  • Top five shooting percentages, 5 vs 5
    • New York Rangers – 8.95
    • Florida Panthers – 8.84
    • Washington Capitals – 8.24
    • Calgary Flames – 8.15
    • Ottawa Senators – 8.15

2015-16 Average shooting percentage, 5 vs 5 when leading: 8.00%

  • 2015-16 Top five shooting teams, 5 vs 5 when leading
    • New Jersey Devils – 10.88
    • Florida Panthers – 10.55
    • Columbus Blue Jackets – 10.24
    • Ottawa Senators – 10.17
    • New York Islanders – 9.36
  • 2015-16 Bottom five shooting teams, 5 vs 5, when leading
    • Vancouver Canucks – 5.59
    • Detroit Red Wings – 6.35
    • Buffalo Sabres – 6.40
    • Chicago Blackhawks – 6.44
    • Minnesota Wild – 6.48
  • Conversely, the lowest save percentage a team had while leading was Edmonton with a .905, while 9 teams had a lower save percentage than that in all situations; the league average 5 vs 5 in all game score situations was .925

2015-16 Average save percentage, 5 vs 5 when leading: .926

  • 2015-16 Top five save percentages, 5 vs 5, when leading
    • Nashville Predators – .940
    • New Jersey Devils – .939
    • Colorado Avalanche – .939
    • Florida Panthers – .934
    • San Jose Sharks – .934
  • 2015-16 Bottom five save percentages, 5 vs 5, when leading
    • Edmonton Oilers – .905
    • Carolina Hurricanes – .907
    • Calgary Flames – .913
    • Dallas Stars – .915
    • Boston Bruins – .917

Playing from Behind

  • The opposite of how a team plays with the lead can generally be said about how a team plays when trailing.  Teams will typically attempt more shots, forcing bad shots rather than waiting for open lanes to develop, thus giving that team a lower shooting percentage.  However, a surprising statistic is that teams that trail typically have a lower save percentage that when leading or in tied games.  I think a reason for this is because the teams that found themselves down more often were also just bad teams in general and didn’t do a good job of “stopping the bleeding” when down.  Another potential reason I see this happening is because the shots a goaltender faces when trailing are of higher quality.  For example, when a team is trailing, they are playing a more aggressive offensive system and will give up odd-man rushes on occasion. This will certainly lead to two or three-on-one’s and the trailing team’s goaltender will face more difficult shots than normal.

2015-16 5 vs 5 Corsi For Percentage when trailing for top five teams in overall Corsi For Percentage and where they ranked when trailing

    • Los Angeles Kings – 62.4, 1st
    • Pittsburgh Penguins – 57.6, 6th
    • Dallas Stars – 58.2, 3rd
    • Nashville Predators – 61.0, 2nd
    • Anaheim Ducks – 57.4, 7th
  • Of the top ten teams in Corsi For Percentage when trailing, only one missed the playoffs – Carolina

2015-16 Average Shooting Percentage, 5 vs 5 when trailing: 7.43% (overall 5 vs 5 average was 7.49)

  • 2015-16 Top Five Shooting Percentages, 5 vs 5 when trailing:
    • New York Rangers – 9.98
    • Washington Capitals – 9.60
    • Minnesota Wild – 8.69
    • Colorado Avalanche – 8.67
    • Dallas Stars – 8.42
  • 2015-16 Bottom Five Shooting Percentages, 5 vs 5 when trailing:
    • Los Angeles Kings – 5.46
    • Anaheim Ducks – 5.51
    • New Jersey Devils – 6.01
    • Chicago Blackhawks – 6.27
    • Buffalo Sabres – 6.34
  • I’m not certain what to make of these two sets of numbers, because four of the five top shooting teams made the playoffs, but three of the bottom five teams also made the playoffs.  One certainty is that some teams are just better shooting teams than others; the Rangers and Capitals were good shooting teams in all situations last season, while Los Angeles and Anaheim were in the bottom ten, yet both still made the playoffs.  It could be that there is no real explanation behind these numbers.  A thought I had is that the Kings, Ducks, and Blackhawks didn’t play a lot of games from behind (they couldn’t have play too much from behind or they wouldn’t have made the playoffs), so they may not have been good teams at catching up.

2015-16 Average Save Percentage, 5 vs 5 when trailing: .920 (overall 5 vs 5 average was .925)

  • 2015-16 Top Save Percentages, 5 vs 5 when trailing:
    • Philadelphia Flyers – .947
    • Tampa Bay Lightning – .944
    • Pittsburgh Penguins – .941
    • New York Islanders – .933
    • Buffalo Sabres – .932
  • 2015-16 Bottom Save Percentages, 5 vs 5 when trailing:
    • Carolina Hurricanes – .893
    • Nashville Predators – .894
    • Anaheim Ducks – .901
    • Calgary Flames – .902
    • Dallas Stars – .907

Zone Starts

  • Zone starts correspond to the area of the ice – offensive, defensive, or neutral zone – that a team or player starts after a whistle.  They are written as OZFO% (Offensive Zone Face Off Percentage), DZFO% (Defensive Zone Face Off Percentage), and NZFO% (Neutral Zone Face Off Percentage).  This statistic is not the success percentage of faceoffs taken in those zones, but rather the percentage of faceoffs that occur in those zones.  Zone starts generally have minimal impact but can be used to further show how a team or player controls play; a team that starts more often in the offensive zone generally controls the puck more because play is frozen by the opposing goaltender.  The metric can also show which players get the easiest or hardest draws on a team because players that start more often in their defensive zone typically have a lower Corsi For Percentage.  Teams will often give their best faceoff takers (often third/fourth line centers) the hardest draws.

2015-16 OZFO% top five, 5 vs 5:

    • Los Angeles Kings – 36.8
    • Detroit Red Wings – 34.3
    • Pittsburgh Penguins – 33.2
    • Tampa Bay Lightning – 33.2
    • Nashville Predators – 33.1
  • All five teams were playoff teams

2015-16 DZFO% bottom five, 5 vs 5 – highest percentage of starts in the defensive zone:

    • New Jersey Devils – 34.3
    • Columbus Blue Jackets – 34.0
    • Buffalo Sabres – 33.8
    • St Louis Blues – 33.5
    • Colorado Avalanche – 33.5
  • Just one playoff team from this group – St Louis

Quality of Competition/Quality of Teammates

  • I’m listing these two metrics together because they pretty much go hand-in-hand.  Quality of competition measures exactly what it says, the level competition that player plays against during a game or throughout a season.  The problem with this measurement is that players don’t regularly play against a specific quality of competition across a season or even in a game to determine any sort of pattern.  This goes back to the sample size problem in hockey statistics. The only occasions when this is easy to measure is when a player is playing for the home team (therefore having the last change after a whistle) and the coach can put out the line that he wants.  In these situations, the coach often puts out his top defensive pairing against the opposition’s best line, along with his best checking line, usually a team’s most defensively responsible line.  In contrast, a coach usually attempts to get his top scoring line out against the opposition’s weaker defensive unit to exploit those weaknesses.  Regardless, this metric has minimal impact but is looked at on occasion.
  • The Quality of Teammates metric is measured as an individual statistic and can be of significant impact.  The idea behind it is that the quality of a player’s linemates or teammates can have a large effect on how that player plays overall.  This is especially evident when a player switches teams or lines and that player’s possession numbers are altered.  When I was trying to determine whether this metric was true or not, I looked at players who changed teams and how their Corsi For Percentage changed as a result.  A change in Corsi conveys the value of a player’s teammates, assuming that the player hasn’t had a rapid growth or regression in skillset or change in playing style.  If we can accept this assumption, then the only element that changed was the player’s teammates.  Therefore, if a player switched teams and his CF% increased, then we can gather that the quality of his new teammates are better than his previously team. Conversely, if the opposite occurs and his CF% decreases, then there is a good chance that the quality of his teammates have declined as well.  We do have to take into account with this theory the idea that a player’s new team could also play a system that values puck possession more than others.  This is absolutely true of the LA Kings, where if a player is traded to the Kings, his CF% typically increases.  With that in mind, let’s take a look at some players who have switched teams in recent seasons and how they fared:

Phil Kessel: 2014-15 (with Toronto) – 45.9% —– 2015-16 (with Pittsburgh) – 53.7%

  • Towards the end of Kessel’s tenure with the Maple Leafs, he had very little talent surrounding him outside of James van Riemsdyk, Nazem Kadri, and Tyler Bozak.  After being traded to Pittsburgh, where there is forward talent on all four lines, Kessel’s CF% jumped nearly 10%.

Milan Lucic: 2014-15 (with Boston) – 51.2% —– 2015-16 (LA Kings) – 59.3%

  • For most of his career, Lucic played for a strong puck possession team with the Boston Bruins before being traded to the Kings prior to last season.  While the Bruins are a strong puck possession team, the Kings are an elite possession team, nearly always in the top five in CF% since Darryl Sutter became their head coach.  Sutter has done this without always having the best talent, so a jump in Lucic’s CF% is no surprise.  This is an example of a move that saw a growth in CF% not necessarily due to a jump in teammate quality, but because of a strong possession system.

Kyle Palmieri: 2014-15 (with Anaheim) – 51.1% —– 2015-16 (with New Jersey) – 45.4%

  • This is an example of a player moving from a team that valued puck possession (under Bruce Boudreau) and had a talented roster to one that was young and didn’t have Ryan Getzlaf or Corey Perry.  It could also be that Palmieri just isn’t a strong puck possession player who was propped up in Anaheim.  He is a player to pay attention to the rest of this season and in the future as the Devils continue to develop and add more talent.

Carl Hagelin: 2014-15 (with the Rangers) – 50.4% —– 2015-16 (with Anaheim/Pittsburgh) – 55.9%

  • Hagelin went from a system in New York that has struggled to adapt to the modern game to Anaheim and, eventually, Pittsburgh.  Although Hagelin wasn’t as productive with the Ducks as he was expected to be, he was still a strong puck possessor and that most likely only got stronger with the Penguins.

With or Without You

  • This metric measures how much of an impact skating with a particular player, and line, can have on a player’s puck possession.  This is done by looking at a player’s overall CF% then comparing it to how he performed with and without a particular player on the ice with him.  I applied this idea to three of the NHL’s most prolific lines in the league last season: Pittsburgh’s HBK line, the Blackhawks’ line with Patrick Kane, and Dallas’ line with Benn and Seguin.

Penguins HBK line of Carl Hagelin, Nick Bonino, and Phil Kessel:

  • Carl Hagelin overall CF% – 55.9, Nick Bonino overall CF% – 51.3, Phil Kessel – 53.7
    • Hagelin with Kessel – 57.9
    • Hagelin without Kessel – 54.7
    • Kessel without Hagelin – 51.5
    • Hagelin with Nick Bonino – 59.1
    • Hagelin without Bonino – 55.3
    • Bonino without Hagelin – 48.9
    • Kessel with Bonino – 57.8
    • Kessel without Bonino – 52.7
    • Bonino without Kessel – 48.4
  • This was the Penguins’ best line last year after it was put together and was their best line during their Stanley Cup run.  They haven’t played together as much this season and the team is struggling, could this be a contributing factor?

Blackhawks line of Patrick Kane, Artem Anisimov, and Artemi Panarin:

  • Patrick Kane overall CF% – 51.7, Artem Anisimov overall – 50.0, Artemi Panarin overall – 51.8
    • Kane with Artemi Panarin – 52.7
    • Kane without Panarin – 48.5
    • Panarin without Kane – 46.8
    • Kane with Artem Anisimov – 51.1
    • Kane without Anisimov – 53.2
    • Anisimov without Kane – 43.4
    • Anisimov with Panarin – 52.4
    • Anisimov without Panarin – 39.6
    • Panarin without Anisimov – 50.1
  • This Blackhawks line, one that included the Calder Trophy, Hart Trophy, and Art Ross Trophy winners, was one of the best in all of hockey last year.  The interesting thing is that Artem Anisimov, the line’s center, was the line’s worst puck possessor, when center is typically a line’s most defensively-responsible position.

Stars line of Jame Benn, Tyler Seguin, and Patrick Sharp:

  • Jamie Benn overall CF% – 53.7, Tyler Seguin overall – 55.0, Patrick Sharp overall – 53.8
    • Jamie Benn with Tyler Seguin – 56.4
    • Benn without Seguin – 43.7
    • Seguin without Benn – 46.2
    • Benn with Patrick Sharp – 54.6
    • Benn without Sharp – 52.7
    • Sharp without Benn – 52.7
    • Tyler Seguin with Patrick Sharp – 56.8
    • Seguin without Sharp – 53.2
    • Sharp without Seguin – 50.9
  • All three of these players are strong puck possessors overall, but worked best when they all skated on a line together.  Patrick Sharp played fairly well regardless, but Benn and Seguin should always skate together as their puck possession numbers dropped significantly when apart.

Individual Points Percentage (IPP), Individual Goals Percentage (IGP), and Individual Assists Percentage (IAP)

  • All of these tend to show the same information – how productive a player is relative to the number of goals that were scored while he was one the ice.  They are all relatively interchangeable, I just like using the Individual Points Percentage (IPP) metric because I like to measure total points rather than just goal or assist production.  IPP is calculated by taking the number of points a player scores over a season and dividing it by the number of goals scored while he was on the ice.  The best way to truly understand this metric is by jumping right into the numbers and explaining them.

2015-16 Highest IPP Among Forwards, Minimum of 750 Minutes Played:

    • Kyle Palmieri, NJ – 87.9 % – this means that Palmieri registered a point on 87.9% of the goals that were scored while he was on the ice, including the goals that he scored
    • Matt Duchene, COL – 87.0%
    • Patrick Kane, CHI – 85.1%
    • John Tavares, NYI – 84.9%
    • Leo Komarov, TOR – 84.6%
  • If you consider the quality of the names listed above, the fact that they ranked at the top of the league in IPP makes sense.  Duchene, Kane, and Tavares are premier goal-scorers and points producers and have all received MVP votes in the past.  Meanwhile, Palmieri and Komarov, while both extremely talented, played on less talented teams than the others and drove the offensive play while their lines were on the ice.

2015-16 Lowest IPP Among Forwards, Minimum of 750 Minutes Played:

    • Alex Chiasson, OTT – 40.0%
    • Brad Richards, Det – 47.8%
    • Johan Larsson, BUF – 48.0%
    • Antoine Vermette, ARI – 48.5%
    • Brett Connolly, BOS – 48.6%
  • All of these players are lesser-known forwards who typically spend a large percentage of their minutes on the third and fourth lines, where not a lot of offensive production occurs.

2015-16 Highest IPP Among Defensemen, Minimum of 750 Minutes Played:

    • Erik Karlsson, OTT – 57.9%
    • Brent Burns, SJ – 55.6%
    • Ben Hutton, VAN – 51.6%
    • Roman Josi, NSH – 50.0%
    • Cody Ceci, OTT – 50.0%
  • There is a significant gap between the highest IPP for defensemen and highest IPP for forwards and that is not due to talent.  Karlsson is one of, if not the best, defenseman in the NHL, who is a perennial Norris Trophy favorite and potential MVP candidate.  Brent Burns and Roman Josi are two more of the most offensively-talented defensemen in the league that help drive their teams’ offenses.  Defensemen typically play the highest amount of minutes for their teams and that is especially true for the elite defensemen, often playing over half of their games.  Because of this, there are bound to be more goals scored while they are on the ice, increasing the number of chances for the defenseman to contribute to the goals, but also decreasing the likelihood that they’ll be able to maintain a high IPP because of the large number of opportunities.  A defenseman’s IPP also suffers because they typically play near the blue line, the furthest position away from the goal.  I feel that the closer a player is to the goal crease, the more likely that player is to touch the puck before going into the net.

2015-16 Lowest IPP Among Defensemen, Minimum of 750 Minutes Played:

    • Mark Stuart, WPG – 4.3%
    • Mark Borowiecki, OTT – 5.6%
    • Patrick Wiercioch, OTT – 7.4%
    • Fedor Tyutin, CBJ – 10.7%
    • Barret Jackman, NSH – 14.3%
  • These defensemen are on the opposite end of the spectrum from the players with the highest IPP.  Part of the reason for this is that they are simply less talented defensemen who play a more defensive style of game, limiting the number of offensive chances.  They also received considerably less ice time (Stuart averaged 16:21, Borowiecki 14:38, Wiercioch 17:20, Tyutin 17:35, and Jackman 13:51) compared to the upper group, restricting the number of potential chances there are to score.

Future of Analytics

  • There has been a lot of research done recently and numerous articles have been written on the future of hockey metrics and analytics.  One area that has been of interest is the idea of zone entries and exits and their importance to hockey success.  The reason for this is how teams enter the offensive zone generally determines puck possession for the sequence that follows.  For example, dump and chase, or chip and chase, zone entries have lost popularity in recent seasons because the team gives up possession with the hopes of re-acquiring it.  This simply doesn’t work in a league that values puck possession more than ever before.  Instead, teams have gone to a zone entry system that allows for players to carry the puck into the offensive zone, hoping to draw attention from more than one of the opposition’s players in order to open up passing lanes.  Similarly, how effectively a team exists the defensive zone after gaining possession can impact how the team will perform in the offensive zone because it can create odd-man rushes and help set the team up.
  • Shots off the rush is another category that has gained attention in recent years.  The reason for this is that shots off the rush are generally of higher quality than when there is a traditional zone entry and both teams have an opportunity to set up their systems.  Shots off the rush can include odd-man rushes where there is only one defenseman back and rushes occurring at higher speeds, making it harder to defend.  Typically these opportunities are harder to defend too because the shorts are taken from a closer range, giving the goaltender less time to react.  David Johnson, who is at the forefront of this metric, has defined shots off the rush as a shot that occurs within 10 seconds of a shot taken by the other team, within 10 seconds of a faceoff at the opposite end or in the neutral zone, or within 10 seconds of a hit, giveaway, or takeaway in the opposite end or neutral zone.
  • Lastly, the most recent development has been in a category trying to measure the number of crossovers per stride in an order to measure skating efficiency.  Belfry Hockey founder, Darryl Belfry, has quantified this metric and has determined that the most ideal ratio is four strides to one crossover.  Meanwhile, skaters on the third and fourth lines typically crossover once every twelve to fourteen strides.  This type of crossover isn’t one that occurs when circling out of the zone but rather one that happens in a straight line and are called linear crossovers.  Linear crossovers are actually used as a tool for acceleration, while at top speed, in an effort to “go to the next gear.”  Some of the league’s best players, including Sidney Crosby, Connor McDavid, Nikita Kucherov, and Patrick Kane have actually reduced their ratio to more than one crossover per three strides.  Belfry has taught clients including Crosby, Kane, John Tavares, Matt Duchene, and Nathan MacKinnon the idea of the linear crossover and these players are among the most elite skaters in the league.  I think that the more successful these players become and the more attention this metric receives, it could become a legitimate tool for gaining an advantage over the rest of the league.




Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s