Thursday, December 22, 2011

Understanding the Blue Jackets Early and Late Game Struggles

During Sunday's game against the Blues, Alex Steen scored just 58 seconds into the 3rd period to tie the game at three. This is a familiar sight for Blue Jackets fans. The Blue Jackets seem to constantly get scored on during the first couple minutes of a period. Also a familiar sight last weekend: with 1:48 remaining the 1st period of Saturday's game against Tampa Bay, Eric Brewer scored to give the Lightning a 2-0 lead. Inspired by these two goals, I decided to take a deeper look at what the Blue Jackets are doing differently at the beginning and end of every period.

To accomplish this, I went through the play-by-play's for every Blue Jackets game and separated all relevant statistics from the first and last two minutes of every period. After this, I went through all the scoring chance post's put together by the excellent Matt Wagner at The Cannon and separated the scoring chances for the same two minute stretches. The statistics I used were goals for, goals against, shots for, shots against, offensive zone face-offs, defensive zone face-offs, blocked shots, shots blocked, missed shots, missed shots against, penalties taken, penalties drawn, scoring chances for and scoring chances against. This information painted a very interesting picture of how the Blue Jackets play during the start and end of a period versus the rest of the period. I expected to see the Jackets getting outplayed in every way at the beginning and ends of periods. Surprisingly, this is not exactly what I found. Yes, the Jackets have been outscored. Big time. But they have not necessarily been outplayed.

The most startling statistics were the goals for and goals against averages (per 60 minutes). The Jackets average a measly 2.27 goals per 60 during the opening and closing of a period. Even worse, they have a 5.91 goals against average in this time frame. I probably didn't need to bold that number, as its terribleness stands on its own. 5.91. That is a crazy high number. That would be good for the worst goals against average by a goalie in modern NHL history. That is a higher goals against average than the worst NHL team in history, the '74-75 Capitals, who limped to a 8-67-5 record. During the other 48 minutes of the game, the Jackets average 2.44 goals per 60, and 2.70 goals against. While they would only jump ahead of Anaheim for 26th in goals per game (they currently sit 27th), this would vault their 27th ranked goals against average (3.27 GAA) all the way to 15th.

As the offensive production doesn't change much, the obvious thought is that goaltending is the problem. While the netminding has been fairly terrible, we can do better than just saying "Mason sucks" and moving on. The Jackets currently rank last with an 88.8% save percentage. During the other 48 minutes of a game, the Jackets tenders stop 90.5% of the shots they face, which would be good for 18th in the NHL. During the opening and closing minutes, they only stop 81.1%. Not good. Again, the obvious thought is "Mason sucks" and moving on. However, this is not necessarily the case. Mason has been bad during the opening and closing minutes, but Curtis Sanford has actually been (slightly) worse. Mason has a goals against average of 5.79, and a save percentage of 82.4%. Sanford checks in with a 5.77 goals against average and an 80.3% save percentage. Furthermore, the team has played better in front of Sanford as well, allowing 3.5 fewer shots per 60 minutes with Sanford in net, while posting a massive Corsi rating of 21.15 with Sanford (explained here, but simply its the difference between shots directed at the opponents net and shots directed at the Jackets net), versus a rating of 7.89 with Mason in net. The Jackets also take more penalties with Mason in net and draw more powerplays with Sanford. Every other statistic is approximately the same with each goalie in net.

Aside from a handful of outings this season, Mason has been terrible. Aside from a handful of outings this season, Sanford has been excellent. Yet both are terrible during this stretch of games. Clearly something is going on beyond the goaltenders. The Jackets actually do a better job of controlling the play during the opening and closing 2 minutes of a period. Their Corsi jumps from 3.87 during the other 48 minutes to 10.76 during the opening and closing 2 minutes. At even strength, this equates to a Corsi rating about 8.6 during the opening and closing minutes. This rate over a full season would have led the NHL in the 2010-11 season. Their even strength Corsi rating of 1.7 over the over 48 would have ranked 13th in the NHL last season, exactly the same place they ranked last year (with a slightly lower 1.4 Corsi rating per game). So the Jackets are getting more shots on net, but they are also missing the net and having their shots blocked with more frequency. They also allow more shots on net during the opening and closing moments of a period, but block much fewer shots and the opponent misses the net less. These statistics suggest the Jackets are playing with a more frantic pace during the opening and closing moments of a period.

Even with this frantic pace, the Jackets are playing in the offensive end. The difference in Corsi rating suggests this, and is backed up by the face-off information. The Jackets have an offensive zone start ratio of 55% in the opening and closing parts of a period, versus 49.8% during the other 48 minutes (zone start explanation can be found here). Furthermore, the Jackets draw 4.55 powerplays per 60 minutes during this period and only take 2.73 penalties per 60, compared to 4.17 and 3.27 over the other 48 minutes. These stats plus the high Corsi ratings suggest the Jackets are driving the play during the opening and closing two minutes of a period.

Based on the statistics so far, the Blue Jackets look like a better team during the opening and closing moments. However, this does not take the quality of the shots into account. For this, we turn to scoring chances. Over the other 48 minutes of the game, the Jackets average about two chances for per 60 minutes more than their opponent, with 17.65 chances for per 60 minutes versus 15.77 chances against. At the beginning and end of periods this flips, as the Jackets average 15.76 chances for, while opponents get 17.42 chances per 60. This means that the Jackets get more shots on net but get fewer scoring chances, while the Jackets opponents gets more scoring chances per shot attempt. To further break down the scoring chance information, I looked at the save percentage and shot percentage of the Jackets on scoring chances. The Blue Jackets actually convert 14.42% of scoring chances, which is actually an increase over the 13.83% conversion rate over the rest of the game. On the defensive end, over the other 48 minutes Columbus goalies save 81.73% of opponent scoring chances. Unfortunately, during the two minute windows I looked at, the Blue Jackets goalies only save 66.09% of scoring chances.

So what does this all mean? Putting the pieces together, the Jackets are taking more shots, but getting fewer scoring chances. This leads me to believe they are forcing shots on net at the open and close of a period. This hypothesis is aided by the higher rate of missed shots and shots blocked by the opponent. And what happens with missed shots and blocked shots? They have a tendency to lead to odd man rushes for the opponent. Odd man rushes lead to scoring chances, which accounts for the higher rate of scoring chances allowed. Additionally, more odd man rushes for opponents explains the lower rate of blocked shots and missed shots against, as it is more difficult for the Jackets defenders to get into shooting lanes on odd man rushes. This also accounts for the higher conversion rate on scoring chances, as an odd man rush usually results in a better scoring chance than most standard shots from the scoring chance area (scoring chance area explained here).

So how can the Jackets solve these problems? The forwards need to stop forcing shots at the net, settle down, take their time and get quality scoring chances. They play this way over the other 48 minutes of the game, yet change drastically during the time frame in question. They are obviously capable of playing this way.  By lowering the number of missed shots and shots blocked, the Jackets will slow down the pace of the game and will lower the rate of odd man rushes. Lowering the number of odd man rushes faced by the Jackets goaltenders will lower the scoring chances faced and lower the opponents percentage of conversion, which will reduce the ridiculous 5.91 goals against average. The goalies are not blameless of course, as both Mason and Sanford must be better. Focus problems seem to be an issue and this is something that is fixable, and needs to happen. Weak goals at the open and close of a period are absolutely deflating. Fixing these 12 minutes per game will not solve everything, but they would be a major step in the right direction. If the team played a full 60 minutes like they did during the other 48 minutes of games, they would be sitting on a total goal differential of -8 instead of -31. A goal differential of -8 would tie them with Dallas and Winnipeg and be in the vicinity of Toronto, Buffalo, Calgary and Los Angeles. All five of those teams are either sitting in a playoff spot right now or in the fight for one.

Note One: There are some obvious sample size problems that I chose to ignore. Using the opening and closing two minutes of every game is only 396 minutes of hockey and therefore these numbers have been skewed a few different ways by a couple of very bad and very good performances. However, I felt the results backed up what I saw with my eyes, and that the good and bad performances essentially balanced out.
Note Two: Scorekeeper bias factors in with a few of these statistics, most notably scoring chances and blocked shots. Scoring chances shouldn't be an issue, as almost every single game was tracked by the same person. However blocked shots may be an issue with the different blocked shot numbers from rink to rink. I checked out out the Fenwick ratings (Corsi without blocked shots), and found the same results as with Corsi, so I have no issue with overall the results.
Note Three: There are some score effects issues as well, as the Jackets are usually trailing in games. Score effects (explained here) mean that trailing teams usually outshoot their opponents. This likely contributes to the Blue Jackets outshooting their opponents over the course of the season. To combat this, I would normally have looked at just the times when the game is within one goal, but this would have created even more sample size issues, so I ignored score effects for the purpose of this article.
Note Four: Empty net goals are obviously not counted. If they did, this would have looked much worse.
Note Five: I did not use penalty minutes for the powerplay/penalty kill statistics, but only looked at actual powerplays drawn, or powerplays given to the opposing team.
Note Six: Based on Pythagorean wins (goals for squared divided by goals for squared plus goals against squared), a -8 goal differential would give the Jackets 15 wins, or anywhere from 30 points and up with ot loss points. If you swap 6 losses for wins, would have exact same record as Los Angeles.

No comments:

Post a Comment