Predicting Power Play Success Based On Analytics: A Test Run With The Minnesota Wild
Stats Up to date as of Monday, February 7th, 2022. Full 2021-2022 results will be posted at the end of the year.
Special teams play has largely been ignored when it comes to analytics despite accounting for approximately 20 percent of NHL game time. Considering the data is a small sample size compared to even-strength, it is somewhat understandable why some teams ignore using analytics.
Quite frankly, you could actually write an entire book based on using analytics and equations to quantify power play success. But I saved you some reading by condensing down data to just 2,700 words.
When I started this project back in October I wanted to achieve two things, I wanted to be able to use a decently-large sample of data accumulated by my own tracking and statistical analysis of the Wild’s power play. But I also wanted to be able to construct equations to quantify power play success.
In the middle of the All-star break, the Minnesota Wild’s power play currently ranks 17th in the NHL, scoring 19.4 percent of the time.
In order to come up with a project, you have to find a problem. The Minnesota Wild’s power play has been a bit of a problem for what seems the last two years now. Last season the Wild started the year 0-16 on the man advantage and followed that up by going 5-74 (.06%) just two months into the 2020 season.
To start the year the Wild sat under league average in terms of power play percentage. But since Matt Boldy joined the big club, the Wild’s power play has gone from 23rd in the league to a somewhat respectable 17th in the NHL, producing 30 percent of the time. Good for fourth in the NHL since January 7th.
During camp, Wild head coach Dean Evason focused a lot of time on improving the Wild’s power play. “It’s an area that is our priority, for sure,” Evason said. “It has to be better. We know that, everybody knows that, and there will be some different looks, and there’ll be some different people on it for sure.”
Following the conclusion of training camp, the Wild opened the preseason with a very strong power play. Constantly moving the puck, holding the zone for a minute on each power play, and directing pucks on the net, making it very difficult to defend. So what happened?
Introduction
As many Wild fans have been doing, I have been watching the Wild’s power play this season and wondering why it’s not producing above league average. With all the star power up front with Kevin Fiala, Kirill Kaprizov, Mats Zuccarello, Joel Eriksson Ek, and the captain Jared Spurgeon, one would expect the Power Play to be one of the leagues best. But that isn’t the case and hasn’t been for the Wild in quite some time.
After tracking the Wild’s power play this season through 223 videos, I have come up with a question.
How can we evaluate and predict power play success based on analytics?
Zone Entries
It all starts with the zone entry. A proper zone entry is one of the three ways to predict power play success. But what are the primary objectives of an effective power play entry?
- Generate a dangerous rush chance
- Get the puck into the offensive zone
- Keep control of the puck once entered
- Quickly set up
Four simple steps but let’s dive into each one.
The first and second steps go hand and hand. To generate a dangerous rush chance you need to get the puck into the offensive zone. You can do so many ways, with the most popular power play breakouts labeled single swing, single post, or double post. The single swing is an entry the Capitals have perfected.
Each are effective in their own way. One thing I noticed with the Wild is they like to use the drop pass method to break it out. To the eye, this creates more space for Kevin Fiala or Kirill Kaprizov to catch a pass from Jared Spurgeon and attempt to generate a dangerous rush chance.
The drop pass method for the Wild has given them a 31% successful zone entry rate. (zone entry is holding the puck for six or more seconds after crossing the blue line)
In terms of the formation of the breakout, I can’t really put my finger on which breakout they use more. Sometimes all five players drop low into the defensive zone for the breakout, other times two guys stay on the blue line (double post) as three guys drop into the defensive zone, and sometimes only one guy stays high on the blue line (single post) as the four others swing in the neutral zone and defensive zone.
Steps three and four listed above, also go hand and hand. In order to quickly set your power play structure, you have to have control of the puck when you enter the zone. Easy, right? What do you do when you get into the zone? That’s the real question.
Power Play Structure
Looking at the Wild’s power play this season you can see their current power play structure is a modified 1-3-1. The reason I say modified is other times the power play looks like an overload to the umbrella structure and other times it looks like a generic 1-3-1.
Essentially, a 1-3-1 can cater to either a left-handed unit or a right-handed unit. With only three right-handed shooting forwards, the Wild have been using an all lefty forward unit with the exception of right-shot defensemanJared Spurgeon.
On the main Wild power play, Evason goes with four forwards in Mats Zuccarello, Fiala, Kaprizov, Joel Eriksson Ek, and one defenseman in Spurgeon.
With the emergence of Hartman’s goalscoring ability and youngster Matt Boldy, Evason has re-scrambled the units. Still with four forwards but this time it’s Zuccarello, Boldy, and Eriksson Ek with Kaprizov and Spurgeon on the point.
The Umbrella used to be of the most common power play structures used in the NHL as teams thought it was the most dangerous. In my eyes, the 1-3-1 is one of the most dangerous structured layouts you can implant on your team, given the skill set of each player on the unit. What makes the 1-3-1 so dangerous is the combination of different plays that can be used to score.
Since 2018 out of the top three power play percentage teams at least two of the three use the 1-3-1 structure.
1-3-1 Examples
The 1-3-1 structure actually has five options with it. Option one is the structure the Boston Bruins, Edmonton Oilers, and the Tampa Bay Lighting use most effectively. Like most of the four options, the play is controlled on the right side of the ice.
With two lefty shooters down low, a right-handed shot in the middle of the high slot, and a right-shot defenseman at the point. This also leaves a right-handed shot way over on the left side of the ice open for a one-timer.
The puck is controlled along the wall, this will cause confusion for the penalty killers and open the ice for some power play magic. If one of the penalty killers does not take away the pass down low to the lefty below the goal line, you pass to him.
This allows the right-handed player to slide down for a one-timer. It is important that the far right-handed shot defenseman crashes the weak-side post. This gives a chance for a rebound off the slot shot or another option to pass too.
One great example is the Nashville Predators’ power play. In the last two seasons from 2019-2021, the Predators’ power play ranked 25th in the league, scoring at a 17.4% rate. It was even worse the two years before that, scoring at a 17.1% rate and finishing 29th in the league during the 2017-2019 time frame.
In those years the Predators had talented players like Ryan Johansen, Kevin Fiala, Fillip Forsberg, Roman Josi, and P.K. Subban to run their top unit. Even filtering in guys like Cale Jarnkrok, Viktor Arvidsson, and Kyle Turris to provide different looks but still, it was one of the leagues’ worst man advantage units.
The Predators rank fifth in the NHL in power-play goals with 35 making them seventh in the league in power play percentage, clicking at 25.18%. So how did they go from one of the league’s worst to one of the league’s best in such a short time?
Instead of running their overload to the umbrella power play with Subban/Ellis and Josi, they shifted to a 1-3-1 formation called the double screen. This option has the player at the top (Roman Josi) taking a shot from the point through a double screen.
The puck gets moved from low to high, when the puck is high you want one guy in the middle screening the goalie and one guy in front of the goalie blocking his vision so there is no way for the goalie to see the shot.
What makes this power play cash in so frequently is the ability and creativity the Predators’ captain Roman Josi has. Josi has seven goals and 13 assists on the power play alone. Most of his shots from the point either find the back of the net, are tipped by Duchene or Johansen, or create a rebound chance for the two guys standing in front (Duchene and Johansen).
What truly makes the 1-3-1 power play so effective is the ability to switch between all five options. You don’t have to commit to just one option or one set play. This is exactly what you see with the 2021-2022 Nashville Predators.
What Do The Analytics Say?
The 1-3-1 power play with the right five guys on the ice can be one of the dangerous tools out there. Shown above were examples and teams who use the 1-3-1 and use it to perfection. We have seen this year, Evason first stacked the top unit with all his talent to try and get something going.
What this creates is a log jam, at times it seems the Wild are trying to be way too cute with the puck. Trying to look for that back door tap isn’t as easy as it may seem.
As Evason may have found out, It’s difficult to tell whether a player is good on the power play without giving them minutes, but no one wants to give a player special teams minutes until they’ve proven themselves. How can a coach decide who they should be playing when they’re up a skater?
One solution, and maybe even the best solution, is looking at a player’s even-strength numbers. As one should expect, the same skills that lead to success at 5-on-5 should generally be useful with an extra skater. Right?
Being up an extra man creates more space for a team to make passes and generate offense. At times this causes teams to try and pass the puck too much and try to find the “cute play” or the guy standing alone in front.
Through various models and looking around the league, shot production is a better predictor of future power play success relative to raw performance or passing ability among players.
Through the Wild’s 129 chances on the power play, I have tracked raw analytics on every player that Evason has sent over the boards. I have also tracked stats on 5v5 hockey for every given player.
As a result, I came up with a formula that shows what players should be getting power play minutes and what players shouldn’t be.
After running our model, we find that the three even-strength variables that provide the best predictions of a player’s PPS% are a player’s Individual Corsi For Per 60 (iCF60.
How many shot attempts he personally takes per 60 minutes), expected FSh% (xFSh%, determines how well you can be “expected” to shoot based on shot locations), and Individual expected goals per 60 (ixGF/60, player’s individual expected goals for)
For forwards, our regression equation is:
PPS% = 1.07 * iCF/60 + xFSh * ixGF/60
For defensemen it is:
PPS% = 1.07 * iCF/60 + xFSh * ixGF/60 + 9
The value nine in the defensemen equation stands for the difference between the forward sum of PPS% and the defenders PPS%.
What Are The Results?
Most of the results were expected, others came as a surprise though. Using the equation I created, Kaprizov leads the way at 32% but quietly, Connor Dewar is tied with Dumba and Ek for fourth on the team with 24%.
Below are the full results.
Kirill Kaprizov 32%Kevin Fiala 28%Ryan Hartman 26%Matt Dumba 24%Joel Eriksson Ek 24%Connor Dewar 24%Calen Addison 23%Nick Bjugstad 23%Matt Boldy 22%Jared Spurgeon 21%Marcus Foligno 20%Jonas Brodin 19%Mats Zuccarello 18%Alex Goligoski 18%Brandon Duhaime 18%Jordie Benn 18%Jordan Greenway 17%Dmitry Kulikov 17%Frédérick Gaudreau 17%Nico Sturm 16%Jon Merrill 16%Victor Rask 13%
The results are certainly interesting. My model projects Dumba to be the best suited defenseman to run the top power play unit. Yet Dumba hasn’t been given the chance to do that since 2018-2019 when Dumba had six power-play goals in 32 games.
The knock on Dumba is he doesn’t hit the net as often as other defenders on the team and isn’t “responsible enough” to man the point.
Those may be valid reasons but as the model shows, Dumba actually is the best defender on the Wild in iFF which is a metric that calculates any unblocked shot attempt, and in iSF which is a metric that calculates a player’s total individual shots on goal.
Dumba also leads Wild defenders in iCF and ixG which are more shot and scoring based metrics.
This may start to seem like an argument to get Dumba on the top unit but I will show later what the actual units should be.
In the near future, the hope and plan is to come up with an equation that takes into account a player’s actual power play metrics and numbers. But for now, the only way we could come up with an equation that can thoroughly prove what players should get a shot on a power play unit and what players maybe shouldn’t is through their even-strength numbers.
This being said, should you take the top ten guys on the list above and construct two power play units? No. There is a strategy for deciding who should be on each unit.
Final Thoughts
For this example, if the Wild want to run two successful power play units based on the 1-3-1 formation they need to send the right guys over the boards.
In terms of the power play breakout, the Wild need to shift into a single swing breakout. This breakout has been perfected by the Washinton Capitals of the NHL and the Minnesota Golden Gophers of the NCAA.
Last season, Spurgeon ranked 12th in the NHL in entry efficiency at 44.4% pass entries and 20.9% individual carry entries. (InStat’s numbers used via The Athletic)
I believe Spurgeon’s elite zone entry ability can be perfectly shown if the Wild switched to a single swing power play breakout at all times.
The equations above suggest the Wild’s first unit should be, Kaprizov, Ek, Hartman, Dumba, and Fiala, and the second unit being Bjugstad, Dewar, Boldy, Spurgeon, and Foligno. With Mats Zuccarello’s elite passing ability, one would suggest he should be on the top unit. Or at least the second unit.
I think the Wild could and should have two very strong power play units. With the first unit of Dumba, Hartman, and Eriksson Ek, with Spurgeon and Kaprizov on the point. Now if you think the chemistry between Zuccarello and Kaprizov is too good to split up then you replace Erikkson Ek and Zuccarello. Finally, a second unit of Boldy, Dewar, Fiala, Brodin, and Zuccarello.
To compare, the Wild’s two units currently are Zuccarello, Eriksson Ek, Kaprizov, Boldy, and Spurgeon on the top unit. The second unit is Greenway, Hartman, Foligno, Brodin, and Fiala.
The only way to actually test the equation to see if it can actually predict power play success, it needs to be used. Could analytical equations be used to run a power play or penalty kill? Maybe someday, but that’s my hope.
Author’s Note
The point of this article was to create equations to quantify and predict power play success. I came up with my own equations based on even-strength numbers that I believe translate to higher danger chances. But I also wanted to touch on power play breakouts and what to do with the puck when not in the offensive zone.
Who knows, maybe I’ll come up with more equations to produce an effective breakout rather than equations that suggest who should hop over the boards on the power play. But that’s the great thing about analytics.
To some, it’s just fancy numbers on a spreadsheet but it’s much more than that. Analytics can tell a story about players and teams. They can be used in developing players, drafting players, signing players, trading, and so much more.
To this point of the article, it was used to predict and evaluate power play success.
I had a lot of fun with this as it’s safe to say we could very well see something similar to this in the NHL someday. Especially with the growing knowledge of analytics in NHL front offices.
(All Data and Information via Evolving-Hockey, InStat, NHL, and The Athletic)