Reader Zach Fabi, a second semester senior at the US Military Academy at West Point, majoring in Engineering Management (Systems Engineering) came to us willing to explain some advanced stats. Over the past two years he has done significant work with statistical analysis, primarily analyzing baseball statistics, done two internships over the past two summers with the New York Yankees working in their scouting department running regressions and analysis for their minor league system.
So who were we to turn him down. Zach's first post on this subject tackles the Pens streak and what the advanced stats are telling us. ( The post was written as the Pens were on an 11 game winning streak) It is one of the better reads you'll find on advanced stats, and if you have some time, check it out.
Fun with Advanced Stats: Explaining the Streak with Fenwick and Corsi
By Zack Fabi
Here’s some news that if you’re reading this site you probably weren’t aware of, heading into this Sunday, the Pittsburgh Penguins have been on an 11 game winning streak. Sarcasm aside, the 10 games prior to that win streak saw the Pens play out to a 5-5-0 record, taking only 10 points out of a possible 20.
So what gives? Why have the Pens suddenly doubled their points per game rate and what is driving this winning machine? Aside from the failed Zach Boychuk experiment and the absence of Bortuzzo from the lineup, to try and answer that, I’ve taken a look at some of the 5v5 statistics the Pens have put up both prior to, and during the streak to see if we can glean any information from them. Here is a fun chart I put together:
This chart shows the total stats over the 10 game winning streak by player. It looks at Defensive, Neutral, and Offensive Zone faceoffs they were on the ice for (all 5v5 situations), as well as Def-Off zone faceoffs. It also looks at the stats that correlate with Fenwick (+/-) and Corsi (+/-). These include goals for and against, saved shots for and against (For in this case means a shot by a Pens player that was blocked by the other team), missed shots for and against, and blocked shots for and against.
To go any further, I’m going to pause and explain exactly what Fenwick and Corsi actually are. Fenwick describes all of the shots taken while said player was on the ice, not counting shots that were blocked. This includes shots that missed the net, shots that were saved, and shots that ended up as a goal. Corsi takes those same numbers, and adds to them the shots that were blocked prior to reaching the net.
At this point you might be asking yourself why should anyone care about these Fenwick and Corsi guys? Well, for starters, shots tend to indicate the flow of the game, particularly puck possession. And as some complicated studies that I’m only going to link to suggest, puck possession plays a massive role in predicting who is going to win the game (http://objectivenhl.blogspot.com/2011/02/shots-fenwick-and-corsi.html, http://jibblescribbits.com/2013-articles/march/an-introduction-to-fancy-hockey-stats.html).
Now that you understand why we’re looking at Fenwick and Corsi, you might be wondering why I included the information about zone faceoffs taken. These stats do not correlate with winning nearly as much, in fact hardly at all, but they do give us some valuable insight into how Bylsma is using his players (i.e. Does he trust Tyler Kennedy on the ice to start in the defensive zone very often or does he prefer to start Kennedy on the other side of the ice where he can do less harm?). This will let us see whether a change in strategy has played a role in the Pens win streak, or if I just painstakingly wasted everyone’s time.
The first stats I’m going to look at are the Team totals, computed by simply taking the totals from when Fleury was on the ice and adding them to the totals when Vokoun was on the ice, not counting empty-net situations (pretty complicated stuff here). Looking at the streak (which mind you is 11 games, the pre-streak stats only span 10), the team totaled a +13 Fenwick, and a +26 Corsi, while pre-streak the team totaled a +16 Fenwick but a measly -1 Corsi. What does this mean?
The Pens were better at Fenwick before the streak, but they were way worse with regards to Corsi. Essentially, the biggest take away from this stat is that over the past 11 games, the Pens have suddenly remembered how to block shots.
During the 10 games prior to the streak, the Pens blocked a total of 114 shots, during the streak they’ve blocked 143.
That’s a pretty significant difference that can’t be explained away by the one game they played. This doesn’t explain the disparity in Fenwick, however. Fenwick could be explained by more time spent playing special teams, but it can also be explained by the fact that teams have been giving up the body and blocking Pens shots more and more frequently as well.
It’s not that the Pens have been firing the puck less, or even possessing the puck less, it’s that teams are getting more and more eager to lay down in front of their shots as the streak has gone on. Pens’ opponents have blocked 156 shots over the past 11 games, whereas during the 10 games prior, they had only blocked 97.
This all might be easier to wrap your head around in the form of per game averages, so here are two more charts I put together:
Back to the Team’s performance, prior to the streak the Pens were averaging a Fenwick of +1.6 per game, and a Corsi of -.1 per game. The pens were blocking more shots than their opponents, but were still barely outshooting them.
Puck possession was nearly an even match, as can also be evidenced by the fact that the number of in-zone faceoffs were nearly identical for the offensive and defensive zones. During the streak however, the Pens have controlled the play far better, posting a Corsi of +1.93. The Fenwick average is down, but that is largely due to the fact that the Pens opponents are simply blocking more shots, and the increase in Corsi is directly related to the fact that the Pens are pulling the trigger more often, and are carrying the play more as well.
But what about the performance of the individual players? Who’s stepped up their game over the past 11 games. For starters, let’s look at the best line in hockey, Crosby, Kunitz, and Dupuis. Prior to the streak, those three were posting a 2.9, 3.2, and .3 respectively with regards to their Corsi, and a 1.8, 2.4, and a 1.1 respectively for Fenwick.
During the streak, these numbers have jumped to 4.73 and 3.18 for Crosby, 4.73 and 2.9 for Kunitz, and a 4.64 and 3.18 for Dupuis. That’s a pretty staggering increase across the board. Essentially, in the absence of Evgeni Malkin, the Crosby line has dramatically increased their ability to possess the puck and use that advantage to fire shots at opposing nets. What’s more, this line is accomplishing this while slightly increasing the number of Defensive zone faceoffs, and decreasing their offensive zone faceoffs. They’re starting in their own zone more often, and are still taking close to 5 more shots a game than whoever they are on the ice against.
The last person I’m going to look at is everyone’s favorite Norris Trophy candidate, Kris Letang. Before going down earlier this week, Letang had been a driving force during the winning streak. During the winning streak, Letang increased his Fenwick and Corsi from 3.88 and 3.75 respectively to 4.33 and 6.67. When he was on the ice during the streak, Letang was a force to be reckoned with; driving the Pens to almost 7 more shots a game than his opponents on the ice.
Does the loss of Letang mean the streak is doomed? Will the Pens be able to continue to use puck possession to wear down and beat their opponents? Does anyone actually think Tanner Glass is a useful player (seriously look at those stats)? Those are questions for another day, but as it is, enjoy the streak, and the fact that Sidney Crosby is not human.