The Pittsburgh Penguins may have had a rough time against the Columbus Blue Jackets during the 1st round of the 2014 NHL Playoffs, but there were certainly some bright spots and some positives to take into the 2nd round against the New York Rangers tonight.
So as we have done numerous times throughout the season, we are going to look at how the players performed by comparing their relative numbers, which is to say how the team performs with them on the ice minus how the team performs when they are not on the ice. So we will be looking at the 5-on-5 numbers for Goals and Shots, as well as derived stats Sh% and Sv%. I also wanted to try something new, and with a small 6 game sample of data to sift through it was easy enough, so I also created a chart for relative Zone Starts. And then in order to further look at individual player contribution rather than just on-ice team based numbers I also have Individual Points Percent and Individual Shot Percent.
5-on-5 Goals Relative
The X-axis is Goals For Relative, so players with positive values on the right hand side are those that see the team score more often whilst they are on the ice, whereas players with negative values on the left hand side are those that see the team score less often when they are on the ice. The Y-axis is Goals Against Relative, so players with negative values on the lower half are those that allow less goals against whilst they are on the ice, whereas those in the upper half give up more goals against when they are on the ice. The size of the bubble is GF% Relative, with white bubbles indicating a negative value.
There are obviously going to be some strange sample size variations, which is quite evident in the fact that the 3 forwards who played just 2 games each form the 3 extreme peaks to form a triangle shaped area of our chart. Players in the lower right quadrant were the most successful, creating more goals for while allowing less goals against. Players in the upper right quadrant had struggled defensively but did manage to create more offensive output. Players in the lower left quadrant were solid defensive contributors but were lacking in offensive skill. And lastly, those in the upper left quadrant managed to struggled both offensively and defensively, seeing less goals for and more goals against whilst they were on the ice.
5-on-5 Shots Relative
The X-axis is Shots For Relative, with players who have positive values on the right seeing the team take more shots on goal whilst they are on the ice, whereas those with negative values on the left saw the team take less shots on goal when they were on the ice. The Y-axis is Shots Against Relative, with players that have negative values on the bottom allowing less shots against whilst they are on the ice, whereas those with positive values on the top give up more shots against when they are on the ice. The bubble size is SF% Relative, with white bubbles indicating a negative value.
As above, we see some sample size extremes emerging as two of the forwards who played in just 2 games each set the boundaries left and right. And as above our best players are those that are in the lower right quadrant as the managed to create more shots for while allowing fewer shots against. Those in the upper right quadrant are able to generate more shots themselves but unfortunately also allow more shots against. Those in the lower left quadrant are able to suppress the opponent's ability to get pucks on not but do not create much offensive opportunities themselves. And lastly those in the upper left quadrant see both fewer shots for and more shots against when they are on the ice, clearly losing the possession battle.
5-on-5 PDO Relative
Obviously you would prefer to keep the puck in the opponent's zone 100% of the time, if your opponent cannot shoot the puck they cannot score goals, but in the playoffs a slight gaffe can mean the difference between moving on and heading home, so in the end what matters is how many pucks go in the net. A player can control the flow of play by taking a ton of shots, but if they are relatively harmless attempts that are easily saved and gobbled up to stop play it is certainly going to lessen the impact it has on the game, I like to think of this as TK Syndrome. Or you could have a player that sees a ton of extra pucks come their way, but they manage to limit the opponents' ability to take quality scoring chances and as such wind up with very few pucks going into their own net. So in order to take in the contextual difference between outshooting and outscoring your opponent, we can look at Sh% and Sv%.
The X-axis is Shot Percent Relative, with those who have positive values on the right seeing the team find the back of the net a greater percentage of the time that they are on the ice, whereas those with negative values on the left see the team scoring at a lower rate with them on the ice. The Y-Axis is Save Percent Relative, with those that have a positive value on the top seeing the team allow the opponent a lower percentage of goals when they are on the ice, whereas those with negative values below see the opponent score are a higher rate with them on the ice. The bubble size is PDO Relative, with white bubbles indicating a negative value.
We still see some slight sample size discrepancies, once again one of our 2-game forwards being the extreme outlier on the chart, but for the most part we are seeing some good data. Players in the upper right quadrant have been our best performers, scoring at a higher rate and suppressing the opponents' ability to score whilst they are on the ice. Those in the lower right quadrant have continued to excel offensively, but they are less than ideal when it comes to defensive situations. Those in the upper left quadrant do not bring much to the table offensively, but they do manage to be quite proficient in defensive situations. And lastly those in the bottom left quadrant have been struggling and see a lower percentage of scoring while allowing the opponent to score more often when they are on the ice.
5-on-5 Zone Start Relative
In an effort to somewhat explain why certain players have higher or lower numbers than others, I decided to look at the difference in Zone Starts. It doesn't completely excuse poor results, but it does help explain why some players may look worse than they actually are. If a player sees significantly more time in the D-zone then we would expect them to be getting outshot and possibly outscored, so we can forgive somewhat lower numbers. However, those that manage to excel in this situation are just that much more impressive. Conversely, players with significantly more time in the O-Zone should be dominant offensively, so if they happen to be outshooting and outscoring their opponent that is just them doing what they were put on the ice for, thus lessening the impact. However, those that get put in these preferential situations and still struggle should be somewhat of a cause for concern.
The X-axis is Offensive Zone Start Relative, so players with positive values on the right hand side start in the O-Zone more often than their teammates, whereas players with negative values on the left hand side see fewer O-Zone starts. The Y-Axis is Defensive Zone Start Relative, so players with positive values on the upper half take more D-Zone starts than their teammates, whereas players with negative values on the bottom half see fewer D-Zone starts. The bubble size is Zone Start Percent Relative, with white bubbles indicating negative values of those that see tougher starts whereas the positive yellow bubbles are more cushy sheltered minutes.
Players in the top left quadrant play the toughest minutes, seeing more D-Zone starts and less O-Zone starts than their peers. Players on the lower left quadrant see fewer D-Zone starts but also fewer O-Zone starts, whereas players in the upper right quadrant see more D-Zone starts as well as more O-Zone starts, so these players may or may not be taking a higher proportion of tough minutes. Players in the lower right quadrant, however, are most certainly seeing more sheltered starts are they see fewer D-Zone starts and more O-Zone starts than their peers.
5-on-5 Individual Percent
This is the one that unfortunately sees the most sample size issues, as there are so few individual events that occur in any given game. However, it is important to look at individual player contributions, as the on-ice numbers above are team-based stats. Since most of the players skated with the same teammates for most of the series, we are unable to pinpoint who on a particular line is driving play, or conversely dragging their teammates down with them, and who is just a product of being on the ice with those who do. So by looking at Individual Points Percent we can see who is responsible for scoring the goals that occur whilst they are on the ice, while Individual Shots Percent allows us to see who is driving play by taking a larger proportion if the shot attempts that occur when they our on the ice.
The X-axis is IPP, with players further to the right having a greater hand in the team's offensive production whereas those further left are more likely to have just been on the ice without contributing to the goal. This is where we see te biggest impact of sample size, as there are two players who weren't on the ice for any Goals For and as such have 0%, an addition 6 with no Points and as such a 0%, and 5 that managed a perfect 100% by having a hand in all the goals they were on the ice for. The Y-axis is ISP, with players higher up having a higher percentage of the shot attempts that occurred while they are on the ice come from their own stick, whereas those lower on the chart are more just riding on the production of others. The bubble size is 5-on-5 TOI%, so the larger bubble indicate players which the team trusts with a greater number of minutes.
Since we don't have positive and negative values like we did with the Relative numbers, we don't have a nice easy 0 value to center everything around. So instead the axes cross at the team average of 55.0% IPP and 20.0% ISP. However, because there is a difference in how much forwards and D create offensively, I felt compelled to also include separate averages. As such, the blue lines are the F average intersecting at 69.4% IPP and 26.3% ISP, whereas the red lines are the D average intersecting at 33.3% IPP and 10.5% ISP. So you can get a general idea of how players are doing based on whether or not they exceed these average values.