The April Blues: Puck Possession?

In part one of our series, we took a look at the Blues' player usage in the 2014-15 playoffs and how it related to player usage during the regular season.  In part two we are going to look at the team's possession during the playoffs and breaking down different elements of their possession game.

The St. Louis Blues are known in the NHL as a puck possession team.  When aggregating the past three season, they are in the top 5 of corsi for.  This has translated to a top 10 spot in both goals for and in on ice shooting percentage.  However, they barely make the top 25 in on ice save percentage, but that is another story for another day.

The Blues continued this trend into the playoffs last season with earning over 50% corsi for in five of the six playoff games against the Wild.  The one game they were under 50% they had 48.9%.  So I think it is safe to say that the Blues dominated possession during their series against the Wild.

But if possession (as measured by shot attempts or corsi) is supposed to be a predictor of team success, how does a team with strong possession during the regular season and during their playoff series still manage to get knocked out of the playoffs in the first round?

Measuring puck possession in the NHL

Since the NHL doesn't track and publish the time a team has possession of the puck (or even zone time for that matter) we have to create a proxy to measure puck possession.  Several years ago some really smart guys who were into hockey and stats discovered that the shot attempts a team makes was a better predictor for team success than shooting percentage, or goal differential, or any other method that was being used to measure teams and players.  By shot attempts we mean all shots taken (misses, blocks, saves, goals), not just the ones the NHL and the game broadcasts report (which are primarily shots on net).  Since a team is required to have possession of the puck in order to shoot it, we can use these unblocked shot attempts (or corsi) as a way to measure puck possession for the team and to calculate the team's possession of the puck when a player is on the ice.  The stat was originally called corsi by the individual who created it.  However, as of last season, the NHL started to provide this stat on their website.  They refer to it as SAT or shot attempts.  Throughout this article you will see me referring to the stat as both corsi and unblocked shot attempts.

Player possession

The ultimate question about this team is if they play so well in the regular season, why the first round exits?  In the first part we compared usage and ice time between the regular season and playoffs and we will do that with possession as well.  On the "Corsi - Regular Season vs Playoffs" tab in the dataviz below, we are comparing two different components of corsi that determine a players overall corsi for percentage (the percentage of unblocked shot attempts in favor of the team when that player is on the ice).  The scatterplot compares the per 60 rate of a player's corsi for against their per 60 rate of corsi against.  The color of the bubble is their corsi for % and the size is time on ice.  The bubbles for the playoffs are a bit more spread out than those of regular season, but that is mostly because it represents only 6 games of data versus the 82 games of the regular season.

There are two ways a player can achieve a positive possession percentage.  He can either be on the ice for a lot of unblocked shot attempts or they can be on the ice when the team is limiting most of their opponent's unblocked shot attempts.  That is what I like about this chart.  You can see which elements are driving a player's possession numbers when they are on the ice.

For instance, when looking at forwards during the regular season, the only two that had sub 50% possession (meaning they were on the ice for more attempts against than for) were Ryan Reaves and Steve Ott.  This chart helps us understand why.  Even though their corsi against rate was the team average, when they were on the ice, the team was not generating as many unblocked shot attempts.  So they ended up with a sub 50% corsi for.  There were a lot more forwards who were on the ice for greater corsi against rates than Ott or Reaves, except when they were on the ice, the team's corsi for rate out paced their opponent's corsi for rate.

You can click on a player's name in the table underneath the chart to highlight that player's bubbles.  Go ahead and click on Pietrangelo.  Remember in part one how we highlighted Petro's increased ice time per game in the playoffs as compared to the regular season.  He sure made good use of that time.  His corsi for% rocketed from 50% in the regular season to 64% in the playoffs.  But more on that later.

First, I want to make sure you are sitting down.  This isn't going to happen too often, but we should always give credit where credit is due.  Go ahead and click on Steve Ott's name in the table.  His corsi for% jumped from 47% in the regular season to 61% in the playoffs.  Wow!  He accomplished this by reducing his unblocked shot attempts against.  His corsi for actually dropped between the regular season and the playoffs.  So say all the negative things you want to say about him, but he managed to be on the ice while the Blues suppressed the Wild's unblocked shot attempts.  There were some other unexpected jumps in possession for some players like Reaves and Porter.  But the disappointment was Jaskin.  Dmitrij Jaskin was the only Blues' forward who had a sub 50% corsi for during the playoffs.  People talk about players disappearing from the playoffs.  Jaskin nearly dropped 10 percentage points in his corsi for between regular season and the playoffs while still pulling the same time on ice.  Perhaps people are right in questioning whether Jaskin is ready for a top 6 spot?  You can explore the rest of the players in the chart on your own.  I look forward to reading your comments about what you discover.

Players' corsi contribution

So if the team overall had good possession numbers and almost all the players on the team had positive possession numbers and if possession is what we use as a predictor for success, what went wrong?  Was the driver of possession different in the post season as compared to the regular season?

To answer this question, I calculated what percent of unblocked shot attempts a player generated made up the total number of unblocked shot attempts that were generated when the player was on the ice.  In other words, what was the player's contribution to his and the team's corsi when they were on the ice?

The fourth tab in the data visualization below compares the player's corsi contribution in the regular season (thick blue bar)  to their contribution during the playoffs (narrow orange bar).  A player's corsi is a measurement of what he and his teammates do while he is on the ice.  This corsi contribution looks at the direct impact a player has on the team's possession as measured by unblocked shot attempts.

Since we picked on Jaskin earlier, let's look at his contribution in the regular season versus the playoffs.  He had the second highest corsi contribution on the team during the regular season, and his contribution was greater in the playoffs.  If I was more savvy with my coding skills, I could easily scrape the data from NHL.com to figure out who Jaskin spent most of his time with on the ice during the playoffs, because it is difficult to see him being the cause of his poor possession numbers in April.

The other player I want to point out is the Washington Capitals' T.J. Oshie.  The fans were clamoring after the playoffs that the core disappears during the playoffs; Oshie disappears, Backes disappears, Steen disappears, on and on.  "TRADE THE CORE!" was the cry as soon as the game six clock reached zero.  Yet here we have Oshie actually driving possession more in the playoffs than when he was on the ice during the regular season.  But let's trade Oshie because "disappears" during the playoffs. [Yes I realize that there was more to the Oshie trade than just stats.]

If you want to find people who are disappearing during the playoffs, take a look at those individuals whose playoff contribution is considerably less than their playoff contribution.  Unfortunately these players are David Backes and Lehtera.  But could this be some explanation of why the Blues' possession numbers did not translate into playoff success?  I think an argument could be made that when your first line center (yes we all know he shouldn't be on the first line but work with me here) and the center of your flashy, high scoring second line contribute less to their team's possession in the playoffs than regular season, you are not going to be as successful.

Conversely, Petro, Zybnek, Schwartz, and Goc, all stepped up their corsi game in the playoffs.  For two out of those four that is a good thing.  For the other two, you have to question are these really the players we need driving possession in the playoffs?  I'll let you figure out which pairs I am talking about.

Oh, and here is a little something for the Berglund haters.  Yes we all know he isn't worth his contract, but he was also one of the players who brought his corsi game to the playoffs.  So can I ask you to back off just a little bit?  He has a role and he plays his role well.  He isn't a bad player just because his contract doesn't match his role.

Another look at corsi contribution

Now that we have an idea of how a player contributes to his team's possession while he is on the ice, we are going to compare that against their actual corsi for%.  This is the fifth tab in the data visualization below, "Corsi For % & Player's Corsi Contribution."  This is really just further examination of a player's possession numbers.  Petro stands out because of his individual corsi for rate.  It is obvious that he was driver of possession when he was on the ice.  Even though Tarasenko wasn't contributing as much to corsi in the playoffs as the regular season, he was still a significant driver of possession when he was on the ice.  The table underneath the chart has the raw numbers so it is easy to see which players were driving unblocked shot attempts when they were on the ice.  The fact that Goc (fourth line) was one of the tops, contributors of corsi while he was on the ice just doesn't sit well with me.  Not that I am anti-Goc or anti-fourth line.  But I would think you want players from your top six being in the top contributors of corsi.  After all, they are the ones you are counting on to score goals.

Corsi versus fenwick

So after corsi came along, someone by the last name of Fenwick did some work and suggested that you should remove blocked shots from corsi because blocked shots can be considered a skill.  However, if you are wanting to look at trends and patterns, more data is better than less.  So fenwick tends to be used for analysis involving longer periods of games while corsi tends to be used for shorter periods of game.  This is why we have focused mainly on corsi in this breakdown of the Blues' post-season possession.  However, we would be remiss if we didn't take a look at how many of these shot attempts were blocked by the Wild during the playoffs.  If blocking shots is a skill that can affect possession, then we should look at whether or not the Wild were able to significantly impact the Blues' possession numbers by blocking shots.

The final tab in the data visualization below shows the percent of shot attempts (corsi) made by a player that were blocked.  There are two columns:  regular season and playoffs.  The regular season bars are gray because we are just using them for comparison against the playoffs' bars.  I think we might have found another clue as to why the Blues' failed in the playoffs this year.

The bars show the percent of shot attempts that were blocked.  The playoff bars are shaded by the percent difference between the percentage of shot attempts blocked in the regular season versus the playoffs.  If you can't tell by looking at the chart I will tell you that the player that has the greatest percent change in blocked shots between regular season and the playoffs is Alexander Steen.  Petro, Carl Gunnarsson, Steen and Jori Lehtera all saw at least a 50% increase in the number of shot attempts blocked in the playoffs compared to regular season.  Chris Porter was the only player who actually saw a decrease in his shot attempts blocked.  Other players that stand out are Ott, Marcle Goc, and Reaves.  It seems that the Wild did their research.  They were going to let fourth liners take shots all day, but they were going to lay down their lives to block players like Steen and Lehtera.  This seems perfectly logical and what you would expect any coach to do.  So do you see where I am going with this?  Why did Hitchcock not anticipate this?  Why did he let his occur?

One of my theories all along is that the players kept playing Hitchcock's style of hockey.  However, the Wild knew what to expect and prepared for it.  It is kind of like playing poker but only playing one or two hands before moving to a new table.  If you keep switching tables the new players are not going to be able to pick up your tell after only one or two hands.  However, if someone has been watching how you play at each of those tables, they are going to be ready for you when you sit down with them and they are going to rob you blind.

But this is just a theory.

We still have to examine further evidence by taking a look at the Blues' shooting during the playoffs to confirm whether or not they were taking shots that are considered "quality" or if they were merely taking garbage shots that had no chance to score. We will be taking a look at shooting (including shot charts from War on Ice) in part three of The April Blues.

All data for the dataviz below comes from War on Ice and is 5v5 data only.