BLOCKBUSTER! Subban v Weber - Part IV
Photo: Philippe Bouchard/Icon Sportswire
Results compiled from 13/14, 14/15, and 15/16 seasons
WHO WON !!!
Before we reveal who won the NHL Deal of the Century…. Gretzky was traded last century…. we should clarify how we determine who the better player is.
When we evaluate on a play-by-play basis we identify all plays that influence the game, either positively or negatively and on offense or defense. Breaking down the resulting data is straightforward and reveals what each defenseman contributes on offense, and on defense. We can go into much more detail, but for our purposes this level of resolution is sufficient. Here is a quick recap:
On offense PK Subban is the clear leader. He creates 43% more net offense at even strength than Shea Weber and ranks 11th. If we add in his Power Play contributions he jumps to 3rd. He makes sensational offensive plays and is one of the most exciting players to watch, ranking among the top 3 defensemen in Fan Quality Score. His net offense at even strength would be even higher if not brought down by his high Cost of Offense. Only five defensemen at even strength have a higher level of negative offense (essentially weighted turnovers). Weber’s offensive contributions trail Subban’s by a sizeable margin, but he still ranks in the top 30 offensive defensemen at even strength, placing 24th. His ranking climbs to 22 with Power Play included. They are both considered #1 offensive defensemen.
On defense Weber is the overwhelming winner. He ranks 11th at even strength and 7th when Penalty Kill is included. He makes positive defensive plays with the best in the league, ranking 12th. He will not be mistaken for Marc Edouard Vlasic as he makes some defensive mistakes, ranking 87 in negative defense. His biggest defensive advantage over Subban is 85.3% fewer Bad Defensive Plays. Weber is a top ranked defensive defenseman while Subban ranks near the bottom in most defensive categories.
This is what each player contributes on offense and defense. How do we mix the two to come to an overall contribution?
We operate on two basic premises when balancing offense and defense:
Goals for and against are of equal value.
Our platform treats goals for and goals against as being of equal value. This is no surprise but as an extension of this, similar chances for and against have similar value - and one step further - all offensive and defensive plays have a corresponding opposite but similar value. This is just a complicated way to say - if a player makes an offensive play that creates a positive contribution for his team, he should receive a similar but negative defensive contribution if he allows the opposition a similar advantage. For example, if a player creates a 2v1 for his team, he should get the same negative contribution for allowing a 2v1, with consideration given to the unique circumstances of each play. There is obviously more complexity to this process but the concept, if executed properly, should appropriately account for offense and defense.
Contributions should correlate with winning.
To confirm the validity of the platform, we evaluated the data’s relationship to the most important measures of success. If we add up all players’ net offensive and defensive contributions per game, we get a Team Contribution Score (TPI). If we use these per game TPI scores to measure a correlation to winning, or to goal differential, we have a good indication of the value of our offensive, defensive and combined scores.
Using goal differential per game as a measure of team success, and all the games played from the 15/16 NHL Playoffs, we found a categorically strong correlation with TPI, which scored a Pearson correlation coefficient of 0.75. Using the same sample size, classic enhanced statistics, such as Unblocked Shot Attempt Share (close), showed almost no correlation to goal differential, scoring a Pearson of just 0.1. Using a near complete regular season data set from 2014/15, a similar analysis of TPI produced a Pearson correlation of 0.82. Note: Correlation analysis does not include goalie performance data, which should bring the correlation higher.
Without diving into more detail, these results confirm what we already suspect from years of playing experience, that the data captures the keys to winning and that offense and defense are already well weighted. We use this correlation approach to make minor adjustments to optimize the balance between offense and defense. Teams may place a higher or lower value on offense or defense, but this approach is guided by what every team is trying to do…. win.
Including all game situations, Weber leads Subban in contribution by 15.9%.
Weber (rank 8) Subban (rank 16)
At even strength, Weber leads Subban by 35.6%.
Weber (rank 10) Subban (rank 37)
Based on the individual contributions of both defensemen over the past three years...
Montreal made the better deal, at least based on each players contribution to winning.
Whether Weber and Subban’s offensive and defensive contributions will increase or decrease with their new team will depend on a series of factors. It is important to note the analysis does not account for age of the player. Performance several years from now could be significantly different from the past three years.
A few final notes:
Have Weber’s contributions been declining? Play-by-play analysis reveals the 15-16 season, in all game situations, was the strongest of his last three seasons. He had a drop at even strength, but his special teams play pushed his overall contribution ahead of his 14-15 season. He has performed consistently over the past three seasons which should help predict what Montreal can expect from him.
Subban has the capacity to make high impact plays and change games. He was a force in the 2014 playoffs vs Ottawa. This noted, his even strength rankings drop him out of the top 30 defensemen. This indicates he may be expensive relative to his contribution ranking, but not necessarily relative to his Fan Quality Score. From a trending perspective, Subban’s overall contribution climbed from 13/14 to 14/15 but in 15/16 his overall contribution dropped to the lowest level in three years. Hopefully the move south will help drive him back up to his 14/15 level and beyond but based on his last three years of performance, predicting his 16/17 contribution level will be significantly more challenging than for Weber.
NOTE: Some of the data claiming Weber has been regressing is based on shot attempts. There is a massive difference in the probability of a goal within the range of a 'shot attempt'. Many have less than 1% chance of going in while some have a 95% chance. At any given time a 'shot attempt' is made, think about the offensive or defensive influence any given player on the ice had on making that wide variation of shot attempt occur. Actually doing this exercise will shed light on the process.
In the alternative we systematically analyze each game on a play by play basis which takes between 8-10 hours. Each play is analyzed based on the circumstances at the outset of each play, and therefore a player is not penalized or credited for his team mates putting him in good or bad situations. He is evaluated based on what he does, with what he is given. Strength of team and team systems will have an influence on how many good or bad situations a defenseman encounters, but with each new situation he will have an opportunity to make a positive play or a negative play. Even though our data shows a majority of players perform remarkably similar when they change teams, we can track the rate of positive to negative plays to account for more or less positive and negative situations.
Read BLOCKBUSTER! Subban vs Weber - Part I here
Read BLOCKBUSTER! Subban vs Weber - Part II Offense here
Read BLOCKBUSTER! Subban vs Weber - Part III Defense here
Learn more about OUR METHOD