This will also be one of the more openly critical posts I have written about the Anaheim Ducks and its coaching staff in particular. Most of you are aware that I oppose two of the systemic facets that Bruce Boudreau and his team encourage: shot blocking (as a means of defensive structure) and the deep cycle (as a means of offensive structure). I oppose them not because those are bad tactics in isolation, but because they are inefficient.
It is efficiency in playstyle that I am attacking in these posts. Recently, Micah Blake McCurdy, whose playoff probabilities are a feature of hockeystats.ca, released some slick chart images showing how each coach allots ice time by score. I asked if he would remove the noise (infrequent or irregular players) from the Ducks charts so I could present them in this post. I also asked him to color Fowler's line pink.
Today I am going to write about how Boudreau utilizes his defenders, according to score. The next post will be dealing with the forwards, which has equally puzzling trends. I was originally going to tackle both at once, but I got a lot carried away in examining the defense.
Here is the chart for the defense group, but with a caveat to start: the scaling on the chart is odd. The swing among the defense is about 39% of ice share available on the top end to about 30% on the low end, or only 9% (not that much). The chart scaling makes it look more aggressive than it is, although it is still pretty indicative of some bad decisions by Boudreau (in my mind). Here it is:
Spend some time with that chart before continuing.
The very first thought I had when looking at this information was: Francois Beauchemin is not Anaheim's best defenseman. However, Boudreau gives him the most playing time in score close situations. That's a problem, considering how little ice time Fowler sees in comparison during the same situations. I have an issue with this usage, which I will get into by starting broad and coming back to my narrow observations from these charts.
Given the player usage shown above, it is perhaps not a coincidence that the Ducks are 16th of 30 teams in score close possession, as ranked by all shots attempted ("Corsi," until Friday when it will be renamed by the NHL). McCurdy also published the math largely responsible (in my opinion) for pushing us away from score close as the best predictor of goals/success and toward score adjusted attempts, so how is this team there? According to puckon.net, Anaheim is 17th of 30 teams in score adjusted Corsi. That's bad, by the way.
The Ducks are currently playing with the puck among the bottom half of teams in the league, or slightly below average overall. This does not bode well for its playoff hopes, as the teams who advance the furthest in the postseason have all been in the top third in terms of controlling play in recent years. As of right now, the best measure we have for this is score adjusted shot attempts, and Anaheim's numbers there flat-out contradict its record (and contention aims).
Here's why I'm using shot attempts in this analysis: the goal of hockey is to score the most goals. Scoring goals happens by shooting the puck at the opposing net until the desired results are achieved. Probability favors teams who shoot more often, as those teams end up scoring more goals. As such, we favor teams who attempt more shots on average, because they "control play."
Defensemen are tasked with preventing goals more so than scoring them, but goaltenders are the players who stop pucks. In my mind, defenders are tasked with defending against attempts, chances, and shots on more than goals, all of which are under the umbrella of shot attempts. We can see how effective or ineffective a defenseman is by how many events occur during his shifts, the rate at which he "prevents" attempts against, and how often the puck is turned into offense on his shifts.
Anaheim is bad at controlling play but is also among the top teams in the league points-wise. This is possible because, once again, the Ducks rely heavily on the elite offensive output of two future Hall of Fame-caliber players instead of smart systems meant to boost their performance. It is *because of* Ryan Getzlaf and Corey Perry, not *in addition to* them, that Anaheim is still producing enough offense to win games when it doesn't control play well. And it has fallen on the outstanding performance of Frederik Andersen on the other end to preserve those wins, as Boudreau's own player usage is actively harming the team's ability to carry more play.
Consider the defense chart again. Beauchemin is playing the largest share of minutes when the team is leading/trailing by a goal or is tied. If there wasn't better choices below him, using him so heavily wouldn't be a terrible thing: his individual CF% (5v5) is 50.2, but relative to his teammates he's -0.6. He's also played nearly 20 games fewer than many players, which can inflate (or deflate) relative numbers.
In direct contrast to Beauchemin's usage is Cam Fowler, whose minutes shrink whenever Frankie's rise. Cam's individual CF% is 50.6, or as close to the zero point relative to others as the team has at 5v5. Most of us are of the opinion that Fowler is straight up better than Beauchemin, and this more or less sums up why. Fowler plays fewer score close minutes, posts similar possession numbers, and has outproduced Beauchemin in points. All things being equal, defaulting toward production players helps teams win more.
In this case however, Boudreau (and the entire organization really) is betraying his own biases, which aren't supported by stats or sound analysis. Bold statement, right? Not really, as I think Boudreau is remembering trends and going with the safest route. In today's game, safe = death (or a first round exit). Let's look more deeply at why Beauchemin sees more ice time than Fowler.
First up: partners. Beauchemin drives around the ice with Hampus Lindholm, who will eventually be the undisputed number-one defenseman on the team. Lindholm's individual CF% is 51, or +1.1 relative his teammates. He leads the team's defense in 5v5 points, so at least the team is getting production for his minutes. He also leads the defense (regulars) with an unsustainably high 103 PDO. The second-highest PDO on the team? Beauchemin, at 102.1.
Fowler drives around Ben Lovejoy, who had an excellent series against the Detroit Red Wings that one postseason but has been a somewhat unreliable defender since. Despite that, he puts up the fourth best points production among blue liners on the year (ironically two points more than Beauchemin), which likely inflates the opinion his coach has of his play. Lovejoy's PDO is exactly 100, and his individual CF% is 52.4 (+1.5 relative) in the sixth fewest games among regular defensemen.
According to these stats, Lovejoy doesn't look that bad. But in roughly 60 minutes of 5v5 time apart from Fowler, Ben's individual CF% is 44.9 (Clayton Stoner-level awful). Cam apart and in isolation is 47.7, so clearly Boudreau sees that the pairing does lift both players' numbers, which is a probable reason for them as partners. Except in limited minutes, Fowler with Josh Manson posts a 51 CF% and with Sami Vatanen posts a 52.8 CF%, so options exist that should be explored outside of "more Lovejoy." That's part of my issue here today: this is yet another inefficiency put on one of the team's more dynamic players due to Reasons.
Back on track. Of Lovejoy's 10 5v5 points from the blue line, only four are "direct production" (1 G, 3 FirstA). Remember, secondary assists are far less repeatable a stat, so most of Ben's points are not necessarily from plays in which he generated the offense. Of Fowler's 12 5v5 points, eight are direct production (3 G, 5 FirstA), which means that most of his points tend to be from his own creation.
Backing up on just piling on Lovejoy, who is a great complimentary player in limited minutes but just shouldn't be keeping an excellent player back, consider this: of the "regular" defensemen in question, Beauchemin is sixth in primary assists, with two. He does have three goals, but in terms of production efficiency given his minutes played, Frankie offers the team next to nothing. (Again, Beauchemin isn't bad in a vacuum, he's just bad when considering he sinks Fowler's use.)
Imagine Fowler playing Beauchemin's minutes WITH Lindholm: in the 26:29 5v5 minutes of it that exist this year, the two put up 54.2 CF%. If Lindholm with Beauchemin is 51.4 CF% with five goals and eight primary assists for Hampus, imagine how much more production is there if Fowler is there tilting the ice even further in the Ducks favor. And that doesn't even touch on Fowler's own production, which would improve too.
So again, this is inefficient usage of players, in my mind. I understand it somewhat, because Lindholm with Fowler leaves four positions to fill than there are comfortable solutions. But the main reason Boudreau is running Beauchemin out for the most minutes in score close situations to begin with is because his on ice goal differential, a stat that individual players aren't solely responsible for, is more favorable than any regular option aside from Lindholm. Fowler, conversely, is among the worst, and you can go right back up to Lindholm's and Beauchemin's respective PDO numbers to check out how that happened.
That tells me that Boudreau has noticed that Beauchemin gets scored on less than Fowler, ergo more trust. (Never mind that Fowler's been on the ice for more goals for than Beauchemin has, which is the risk-reward of dynamic players.) In terms of effective pairings, the logic behind what's there is obvious: one safe player, one "risky" player per pairing. The safest pairing, as determined more by randomized luck than actual defensive skill, gets the most ice during close situations, and so on.
An aside: ever since Mike Babcock coached the Canadian Olympic team in 2014, the NHL has been obsessed with the handedness of defensemen and playing guys on the "correct" side of the ice. I default back to the older mindset of guys being skilled enough to play either side of the ice, with handedness and sides primarily benefiting the defenders who aim to shoot more than anything. So with Lindholm and Fowler as a pairing in my example, I discount the "but they both play the same side." Bah. Fowler ends up playing two positions with Lovejoy, he could theoretically handle his off-side.
Shortchanging Fowler's upside for 5v5 caution is precisely why the Ducks sit below average in controlling play. Without digging deeper into shot-based numbers, the most apparent reason for it is A) to maintain a balanced blue line (which is an organizational issue, because please stop signing bad defenders) and B) an assessment of the "best defender" based on, largely, luck. Fowler with a 102.1 PDO (like Beauchemin) instead of a 98.9 (what he has) probably puts his production even higher than Lindholm's, so there you go.
This is part of why Anaheim can't consistently hold onto leads and end up barely scraping by in most wins. This can actually be better seen by going across both the defensive and offensive usage charts, which you'll get a better picture of in the next post. In fact, I'll publish the charts side by side then. For now, I wanted to address the biggest inefficiency I can see on the roster: Boudreau rewarding a merely good-to-average player with the best defensive minutes when his most dynamic one sits on the bench, not helping the Ducks be the best team on the ice.
Consider: for every 60 minutes played, Fowler is among the better defenders (along with Lovejoy, but again in limited minutes) at having lower event rates against! Fowler is 53.6 Corsi-against (CA) per 60 and 39.5 unblocked CA per 60, which is better than Beauchemin at 54.3 per and 40.4 unblocked per. And Fowler has played more minutes total on the season, which favors his numbers more than Frankie's in fewer minutes.
Boudreau has Fowler and Vatanen sitting more than he should in close situations. While the two, along with Lindholm, usually end up playing the most minutes each game after taking all score situations and strengths into account, the inefficient 5v5 usage upsets me. The best I can figure: both Fowler and Vatanen tow around boat anchor players and suffer reduced "trusted minutes" because of it, while "safe" Beauchemin has the good fortune to have received better goaltending when he's on the ice. (Not to mention, I'm sure Beauchemin is much more willing to adhere to Boudreau's lane blocking defensive zone scheme.)
I can't wait until Anaheim uses statistics to boost its player evaluation and analysis.
The second post in this series exploring forward usage will be up next week.
Follow Kid Ish on Twitter: @kid_ish