How Can Detroit Improve Its Power Play This Season?

For a long time, power plays have been evaluated using PP%, calculated by dividing the number of power play goals a team scores by the number of power plays it has.

Using this metric, Detroit was 24th in the NHL last year, converting on 17.5% of its power plays. The Red Wings were even lower the previous year (27th out of 30). Seeing that the team was 13th and 2nd the previous two years respectively, the last two years have been a giant step back for what used to be a team strength.

In his article What the Red Wings can learn from an anatomy of a successful power play, Prashanth Iyer points out that the loss of assistant coach Jim Hiller corresponds to the decreased effectiveness of Detroit’s performance with the man advantage.

As Iyer pointed out in a previous article about Detroit’s zone entries on the power play, the loss of Pavel Datsyuk also hit the Wings especially hard in this area.

Returning to the idea of how power play success is calculated, it’s an evaluation that has some glaring flaws. For example, a full two-minutes on the man advantage counts the same as a 5 second carryover power play.

Fortunately, it’s not difficult at all to more accurately evaluate power play success. Simply using Power Play Goals per 60 Minutes or Power Play Shot Attempts per 60 Minutes will produce a much more accurate measure of effectiveness.

For this article, I’m going to look at what Matt Cane refers to as the “three major pillars of power play design” in his article in the 2017 edition of Hockey Abstract:

  1. Player Selection
  2. Structure and formation
  3. Regrouping and entering the zone

To avoid repetition, mention of Cane’s work will refer to this article unless otherwise stated.

Since Dan Bylsma is going to be joining the Red Wings as an assistant coach, most likely in charge of the power play, I will also be looking at what Buffalo’s power play looked like in 2016-17, when he was the head coach.

How Bad Is It?

So, if measuring power play success by using PP% isn’t a good way to evaluate power play success, maybe Detroit’s isn’t that bad.

Unfortunately, that’s not the case.

As Matt Cane references in his article: “Patrick D of Fear the Fin found that, much like at 5-on-5, shot attempts were a better predictor of power play success at the team level than any other metric, including past goal scoring and past shooting percentage...But these results were contradicted by the work of [Arik] Parnass, who in 2016 found that since the lockout-shortened 2012-13 season, past [power play] goals were often as good a predictor of future goals as past shot attempts were.”

Detroit was 26th last season in 5v4 Minutes Per Goal. They were 28th in 5v4 Seconds Per Unblocked Shot.

Clearly, Detroit’s power play needs a lot of work. Let’s see how Buffalo’s power play looked the last year that Bylsma was their coach. The first chart is Detroit’s 2017-18 5v4 Minutes / Goal vs. Buffalo’s 2016-17 of the same. The second chart is Detroit’s 2017-18 5v4 Seconds / Unblocked Shot vs Buffalo’s 2016-17 of the same

While Detroit was below average in both metrics for most of last season, Buffalo was at or above league average the majority of the 2016-17 season. That year, Buffalo finished one point behind Detroit in the standings, but their power play was clearly better.

Specifically, Buffalo’s 6.87 Min / PPG was 1st in the league, while its 45.99 Seconds / Unblocked Shot was 8th.

So, let’s take a look at Matt Cane’s three pillars and see how Detroit stacks up.

Player Selection

At this point, most NHL teams have moved to a four forward power play unit, at least for their top unit. Matt Cane shows that a four forward unit is more effective, and while it leads to an increase in 5v4 Goals Allowed per 60 Minutes (from 0.6 to 0.9), the increase in 5v4 Goals Scored per 60 Minutes (from 5.6 to 7.0) more than makes up for it.

Meghan Hall of the Balls & Sticks Podcast has been doing great work in evaluating power play effectiveness, building on Matt Cane’s work. She evaluated all units in the league that played more than twenty minutes together and scored at least one goal. The top 10 units in 5v4 Minutes per Goal are all 4 forward units:

You might notice that Buffalo’s #1 unit is 7th on the list. This continues the success they saw the previous year, in which the same players comprised Buffalo’s top unit.

On this next chart, we can see how Detroit stacked up. In the first section, we saw how poorly Detroit performed in the two metrics measured here, and there are only two teams, Ottawa and Chicago, that were worse in both metrics. At the bottom we can see that Detroit’s second unit performed better than their first unit for the PP time they received.

Interactive Version of this Chart

So, since this section is on player selection, let’s take a look at how that can be improved. Cane looked into what 5v5 player statistics are indicators of success on the power play.

As he says:

“In theory, many of the best offensive players at 5-on-5 should be good options for a team to use on the power play. After all, if a player is able to generate opportunities when they’re facing an equal number of opponents, they’ll likely still have that talent when space is opened up at the man advantage.”

Cane found that by using a player’s individual CF60 in combination with his teammates’ CF60 when he is on the ice, we can formulate a statistic called Expected Power Play Shot Attempt Differential Per 60 (xPPSAD/60). According to Cane: “This number represents our best guess of their team’s shot attempt differential on the man advantage, if that player were to be given a regular role on the power play.”

If you are interested in reading more about how this statistic is calculated, you can read about it here.

If we are looking at improving player selection, we can look at players with higher xPPSAD/60 than players who play more PP minutes than they do. Keep in mind this isn’t a guarantee that swapping those players out will lead to an improvement.

Here are the forwards. The color coding is as follows: The reds are the top four in 5v4 TOI, in descending order from darkest to lightest. The blues are the same for the next four.

This indicates that giving Andreas Athanasiou more power play time could lead to improved results.

The return of Thomas Vanek gives Detroit another potential PP option. He rated a 83.47 at this metric last season, almost exactly where Tyler Bertuzzi is on the chart.

Of course the potential of adding Filip Zadina and Michael Rassmussen on the power play is another avenue Detroit could explore this season, and I will return to in the next section.

Next, we have the defense:

Detroit really only used Kronwall, Green, and Daley on the power play (in descending order of TOI), with Joe Hicketts coming in fourth, even though he only played in 5 NHL games.

Moreso than the forwards, it looks like the Wings did not take advantage of players predicted to do better on the power play than those who actually played on the power play.

Looking at this chart, a Detroit fan may be tempted to dismiss it because Xavier Ouellet is second from the top.

I think even though it’s a rate stat, which accounts for ice time, Ouellet played less time per game than any Detroit defenseman other than Lashoff (including Hicketts when he was up). Also, he was pretty sheltered even with that little ice time. As always, decisions should not be made on one statistic, but rather in context. Additionally, Ouellet was basically at league average, so this is more a result of Detroit’s overall defense being poor than Ouellet being particularly good.

While Niklas Kronwall knows what to do, his body continues to not be able to do it, and Detroit has plenty of better options. That’s not even counting the potential of a player like Filip Hronek or Joe Hicketts. Simply put, if Kronwall continues to play on the power play this season, it will be a sign that the thinking has not changed enough from what led to such mediocrity the last two seasons.


Before moving to Section two, let’s take a moment to look at the importance of handedness when choosing players to play on the power play.

Cane found that “Players on their off-hand side shot 11.7% on the power play between 2005-06 and 2015-16, 1.4% better than players on [their] strong side.”

Many of the players Detroit has used on its power play recently are left shots. On the defensive side, right shot options from players projected to be on the Wings roster are Mike Green, Nick Jensen, and Filip Hronek (possibly).

At forward, Martin Frk is the only returning right shot forward who makes sense to play on the power play, although Thomas Vanek’s right shot makes it even more likely that he’ll see time on the man-advantage. This will be explored more in the next section.

Structure and Formation

Matt Cane’s second pillar of power play success involves structure. Simply, having a consistent structure and formation has a correlation with power play success.

The most obvious example of this is Alexander Ovechkin, who scores a plethora of goals from the same spot. Even though opponents know it’s coming, they can’t just key on it because Washington’s top unit has so many weapons, and the team is so creative at finding ways to get him the puck in that spot.

Detroit doesn’t currently have the same arsenal that the Capitals can bring to bear on an opponent’s penalty kill. That doesn’t mean that improving their structure can’t lead to more success.

Typically a power play that works well is one in which each player has a role, but they also have enough creativity that they force their opponents to react to different scenarios.

Detroit sets up in a 1-3-1, like many teams. Here’s how this formation works, with a stick tap to the classic book Hockey Plays and Strategies by Ryan Walter and Mike Johnston, which was recently updated for its second edition.

Right and Left Side Half-Boards Players: Right shot on the left boards, left shot on the right boards. “Both much be a threat to shoot or fake the pass ,while also being calm under pressure.”

Mid Ice Point Man: The defenseman in a four forward power play. “He slides along the line with deception while looking to find an open lane to the net.”

Slot Player: “...from one side he must be ready for a quick release shot and from the other side a shot pass.”

Net Man: “He may release to the strong side for a low pass and the potential to make a quick inside play.”

Here’s what it looks like:

And here is how Detroit’s first unit and second unit typically set up last season:

Let’s see that in action:

In the above video, we can see that some of the players can switch spots, specifically Zetterberg, Nyquist and Tatar on PP1, but you can see the setup. From watching video of last season, Detroit’s top unit saw the “3” of the 1-3-1 on the top unit move around more than usual because they didn’t have a right shot to play on the left half-boards.

Matt Cane introduced the concept of a Structure Index, “a rough measure of how consistent teams are in generating options from the same location. A team’s Structure Index is the weighted average of each player’s average distance from their central shot location.”

The lower the Structure Index number, the more consistent teams are where each player takes his shots from. As Cane puts it: “a higher Structure Index indicates that teams are having trouble setting up plays, and that their players aren’t getting looks from the same spot on each power play.”

Meghan Hall has done great work in creating visualizations for this as well. Before we take a look at Detroit’s shot locations, here are the individual shot locations for Toronto’s top PP unit, which was the best in the NHL in minutes per 5v4 goal:

Here are Detroit’s two main power play units from last season:

Zetterberg took his shots from either side of the ice, while Tatar was shooting from everywhere and Nyquist shot mostly from the center.

Athanasiou and Larkin looked like they are pretty interchangeable in the slot position and the right half-board spot. Frk is pretty consistent, and Mantha and Green are as well.

Hall took the top 30 power play units by shot total and ranked them by structure index. Detroit’s top unit was 18.5. It’s second unit was 15.2, which is not surprising, when looking at the shot locations. 15.2 puts the second unit in a tie with the top SJ unit (Pavelski, Thornton, Couture, Hertl, Burns) although it’s important to note that’s with much less TOI than the SJ unit.

For context, here are the top 30 units by PP Shots sorted by Structure Index. Detroit’s top unit is right at the bottom of the group, which indicates that they were not getting set up consistently.

To close this section out, here is a clip of how this formation can work well. Nice structure and creative passing leads to a beautiful goal for Gustav Nyquist:

Regrouping and Entering the Zone

The third of Matt Cane’s pillars of power play success is entering the zone effectively. Other than allowing a shorthanded goal against or taking a penalty, an opponent clearing the puck all the way down ice is the worst thing that can happen to a power play. It allows the opposing team to change its lines and takes precious seconds off the clock.

While a clear can happen off of a blocked or missed shot, a poor zone entry will often lead to the team on the man advantage having to regroup and try again.

Here’s an example where Niklas Kronwall tries a stretch pass from behind his own net, but the failure of this entry leads to precious time ticking off the clock:

Much work has been done in recent years in the area of zone entries and exits. Cane summarizes what has been found so far:

“Ever since Eric Tulsky first published his zone entry work in 2012, analysts have constantly driven home the importance of carrying the puck into the zone and avoid dump-ins at 5-on-5. On the power play, however, he found something different—dumping the puck in had more utility on the man advantage, as the extra man provided a boost in puck recovery. The question is whether that extra utility is enough to recommend a dump-in over a controlled entry.

“Parnass took Tulsky’s zone entry work one step further in his Special Teams project, breaking down controlled entries by the type of control that the player had when they entered the zone. His work found that while dump-ins and chips had the highest success rates (as Tulsky’s work showed), controlled entries are preferable.”

The reason for that last point is that while a dump or a chip-in can gain entry to the zone, a controlled entry allows the attacking team to quickly set up.

Here is an example from last year of how a Mike Green dump and chase entry succeeds in gaining the zone but fails in leading to sustained zone time:

Corey Sznajder has been doing tremendous work in providing the community with data through manual tracking. Last season he tracked 21 of Detroit’s games. Here is the power play zone entry data:

Nyquist, Larkin, and Zetterberg all had more than 20 entries, but 67.9% of Nyquist’s were Carry-ins, as opposed to 96.2% for Larkin.

Athanasiou’s 85.7% Carry-In% (6 Carry-ins on 7 entries) provides further evidence for the idea of giving him more power play time this coming season.

While the drop pass entry has been the bane of many fans’ existences the last few seasons, it’s a type of entry that can be used very effectively, especially with players like Larkin and Athanasiou. The basic idea is that the first player carrying the puck gets far enough towards the opponent’s blue line to push the defenders back, so that when the puck is dropped to the trailer, a fast, shifty player can find an opening through the defenders. If you want to read more about the way it’s supposed to work (as opposed to how Detroit has been doing it recently), Arik Parnass (surprise, surprise) took a look at it in 2016.

As we can see here, even when the drop pass entry works, the player entering the zone has to quickly make a play to get set up. People arguing for a penalty call certainly here have a case, but Larkin has to make the pass quicker:

When the drop pass entry hasn’t worked for Detroit lately, it’s been because the drop pass has taken place too far from the offensive zone, the trailing player has not been someone fast and shifty, or both.

That’s not to say that Detroit should use the drop pass entry all, or even the majority of, the time, but if they are going to keep using it, they need to do a better job of it.

What Improvements Can Be Made?

Buffalo used a very similar setup to Detroit during the 2016-17 season, when incoming Red Wings assistant coach Dan Bylsma was their head coach. The difference, and it’s a pretty big one, is that Buffalo’s top unit had both Jack Eichel and Ryan O’Reilly on the half-boards.

In the first clip, Buffalo does a good job maintaining its basic structure, then the talent of Eichel and O’Reilly combines for a pretty goal:

Similarly, the second goal comes from Sam Reinhart (net front player) knowing where Jack Eichel will be and getting the puck to him off a rebound:

So, the point is that Bylsma knows how to make this formation work. Even though Detroit doesn’t have Eichel or O’Reilly, they can put out players that can be effective.

The obvious barrier to improvement will be the probable loss of Henrik Zetterberg. For a team that needs help gaining the zone consistently on the man advantage, losing a player who can do this is not helpful, to say the least:

So with the players Bylsma will have to choose from, which players would give Detroit the best chance to succeed? Let’s run through the 5 positions in a 1-3-1.

For the point, Nick Jensen and Mike Green make the most sense for current roster players. Depending on which defensemen, if any, make the jump from Grand Rapids out of training camp, the likely candidates would be at least deserve a look.

Abdelkader has been a net-front mainstay for a while, but that should come to an end. Assuming Rasmussen makes the team, both he and Tyler Bertuzzi would be better choices. Mantha could also play there, but I think he fits better in the slot position.

That leaves the slot and both half-boards positions.

As the roster is currently constructed, Detroit will have two right shot players for the left half-boards player: Vanek and Frk, so it makes sense to pencil them in here.

The rest of the players who make sense to play on the power play are: Larkin, Athanasiou, Mantha, Nyquist (in descending order of xPPSAD/60) and Filip Zadina, if he makes the team.

Since Larkin and Athanasiou are currently Detroit’s best players at carrying the puck into the zone on the PP, they should be split up.

Without Zadina, it could look like this:

If Filip Hronek makes the team, it might be worth trying either him or Jensen at Frk’s spot on the left half-boards. While Frk’s huge slapshot is dangerous when it’s on target, if he continues to miss at the rate he has been, it could be better to have a better passer at that spot. Four forward power play units are generally better, but coaches need to choose the best option with the players they have.

If Zadina makes the team, it would create a good problem in that there would likely be a quality offensive player not on the power play. Given his skill set, he would make the most sense on the right half-boards. One option would be to have Mantha take Bertuzzi’s spot on the 2nd unit and move Larkin to the slot on PP1, although Larkin’s best spot seems to be on the left half-boards.

While this article is already very long, there is one more thing to point out. Detroit has had difficulty with aggressive penalty kills like Tampa’s the last few seasons. In this last clip, you see the difficulty Detroit has getting set up with the Lightning pressure:

Just after the clip ends, Detroit ended up with a good scoring chance, but more often this pressure leads to a defensive clear.

The potential units listed above have the passing skill to make quick passes and smart reads to beat this pressure.

By using the best available players for the power play formation, focusing on maintaining structure, and having their zone entries run through the best players at gaining the zone with possession, Detroit may not see its power play produce as well as Buffalo’s has, but it can certainly rise out of the power play basement.