Breaking Down: The Avalanche Power Play

The Avs power play was the subject of much consternation among the coaches and fans alike. At times it looked unbeatable, most of the time it looked drab & lifeless. In the 2nd round of the playoffs it was horrible and in a series they lost by the slimmest of margins was one of the deciding factors against them.

Defenders point to the efficiency (PP%) as justification for all the other problems. It was 7th in the league! It must be good! Digging deep into how they arrived at that rank we see some issues crop up and the coaching staff’s reluctance to tinker with it either technically or philosophically allowed a molehill to become a mountain. What I’m going to do here is deconstruct the PP using tried & true metrics, modern shot metrics and systemic observation to get a handle on what ills befell them last season. As with most of the issues, a lot goes back to the major differences between the lucky first third of the season, the horrifying second third of the season and the solid final third.


Long ago it was decided that measuring special teams success should be done by simply dividing success by opportunity, or goals per chance with man advantage/disadvantage. There’s plenty wrong with that, mainly the fact that power plays aren’t always 2 minutes long nor are they always 5 skaters vs 4. Altitude’s Peter McNab likes goals per game, which is even simpler and actually a bit more logical than straight percentage. Here are some breakdowns by goals to set the stage for the deeper metrics.

Season: 63 goals in 286 chances, 22.0% (7th), 0.77 PPG/gm (4th)
First 3rd: 28 goals in 87 chances, 32.2% (1st), 1.04 PPG/gm
Second 3rd: 19 goals in 110 chances, 17.3% (18th), 0.70 PPG/gm
Third 3rd: 16 goals in 89 chances, 18.0% (19th), 0.57 PPG/gm

– 56 goals were at 5v4

– 20 goals (32%) came against the bottom 6 PK teams (CHI, EDM, LAK, DET, NYR, PHI)

– They were blanked on the PP in 34 games and 5 teams shut them out for the season, all in the East – TBL (1st PK), FLA (10th), MTL (13th), TOR (17th), WAS (24th)

– Low point of the season was a 9-game fruitless streak in late January into February. They went 28 days without scoring a PPG, going 30 straight chances and 48:25 of PP time in between goals. Yikes.

– 28 goals (44%) came in 3+ goal wins/losses
– 17 of 28 goals (61%) in the first 3rd of the season
– 6 of 19 goals (32%) in the second 3rd
– 5 of 16 goals (31%) in the final 3rd

– Only 2 goals were scored by defensemen, both by Tyson Barrie

– 38 goals (60%) were scored by the 3-Headed Monster. 16 by Mikko, 12 by Mack and 10 from Captain Gabe.

I doubt much of this is out of the ordinary save for the massive amount of PPGs scored in blowouts in the first 27 games of the year but it’s good context. The lack of scoring by defensemen is concerning since point shots were a huge part of their strategy, much to the team’s detriment. More on that later.

Shot Metrics

The Avs were considered the 7th best PP by efficiency (and shooting percentage) but far below that via shot metrics. By shot attempt rate they were 20th, unblocked shot rate 21st, shot on goal rate 16th and expected goal rate 18th. To put it simply, they didn’t shoot much but were ok when they actually did. I’ll get into some reasons why in the system section below. For now we need to see how and why the results changed so dramatically from the first 27 games to the rest.

Full Season

First, the season as a whole viewed in RainbowVomitChart© mode:

Legend: Dark blue is good, dark red is bad, lighter shades are less so. The charts are racked by TOI/gm so basically PP1 is at the top and PP2 at the bottom. Anyone outside top 10 TOI for the year is omitted which mainly does away with Andrighetto, Kamenev and Brassard’s limited contributions. These 10 are who hauled the mail for the season.

The top 4 guys (3HM + Barrie) played around 69% of the PP minutes this year. First thing that jumps out is holy cow does MacKinnon dominate. Looking at the shot/quality share numbers (iCF/CF & ixG/xG), he takes slightly more than a third of the on-ice shots and generates slightly less than a third of the on-ice expected goals. Mikko is next best in that regard and he takes 60% fewer shot attempts. Is that a good thing? I think it’s good that #29 produces that volume but it leaves the whole system vulnerable to his shooting luck, which isn’t good as we’ll see in a minute. You’ll notice one bit of pink in his row under G/ixG, which is goals per expected goal or a fancy kind of shooting percentage. It’s below average (as it was at 5v5 for the season as well) and that’s one big reason for the PP struggles. I don’t put that all on him, he’s the main guy so whatever isn’t working is going to manifest heavily through his numbers.

Mikko is the #2 option on the first unit and while not generating the same volume that Mack did he made up for it in finishing ability. His goal rate was 47% higher than Mack’s and his fancy shooting percentage was 164% better. Sounds like getting him the puck more would be a good thing, as we’ll see later… it’s complicated.

Tyson Barrie’s absence next year is a cause for concern by many. Lots of haters are glad to see him go, lots that appreciate his game think it’s going to be tough to ameliorate his departure. One thing that should be easy to replace is his 2 PPGs last year. I’m not a huge fan of shots from the point but more than that would be nice.

The story of the PP2 unit is interesting. They played about 31% of the time or 36 seconds of a 2 minute PP. Most of their stats are fairly uninspiring and it’s easy to see the guys that have some dark blue next to their names got that from time on the first unit anyway. As we’ll see in the next segment, the 2nd unit carried the team through their most productive time of the season then fell down a bottomless pit.

First 3rd

The Avs trounced everyone with the man advantage in the first 27 games of the season. RainbowVomitChart© #2:

Jost spent a lot of this period on the top unit but Colin Wilson also saw action there. Barrie was out for a few games, Sam Girard took his place and the top unit became much more productive. The top unit scored 18 goals while PP2 scored 10 in less than half the time. Compher missed all but 11 games in this stretch but he sure did plenty with the time he was available.

First thing that jumps out is very high goal scoring rates for all the forwards, all but Mack and Compher were around 2 goals per expected goal which is insane. The 2nd unit had great on-ice shot generation, and especially quality generation. Systemically, the way the 2nd unit played during this stretch was the most effective of any group at any time during the season. Vlad Kamenev isn’t on this chart but his rates are comparable to JT Compher’s, so very good. Even bit players like Andrighetto and EJ flourished here. Times were good, but they wouldn’t last unfortunately.

The Collapse

The second 3rd, or 27 game stretch of the season, was a disaster not only for the PP but in general. RainbowVomitChart© #3:

This is where efficiency left the building, going from 32.2% down to 18.2%. Also note that they were getting a lot more opportunities with 1:52 more PP time per game. JT Compher replaced Jost/Willie on the first unit for a lot of this stretch.

The story here begins with MacKinnon’s shooting luck. Most of the other guys were shooting unsustainably high in the first 3rd, Mack was the exception. He went from reasonable at 1.35 goals per expected goal all the way down to 0.47. Not good. His shot rates were fine, although slightly down, and his assist rate was basically unchanged. Barrie’s shot rate dropped 17%, which should be a good thing. Landy’s went up about 10%. Mikko was the big gainer with shot rate going up 28%. The problem was that his xG rate dropped 19% at the same time. Translation: He was shooting more but from less dangerous areas.

Another issue was that Compher wasn’t a good fit, his shot rates cratered playing with the top unit which made a net 70% loss from what Jost and Wilson were doing in the same role and his scoring totally disappeared. He went from 4.91 goals and 14.73 points per hour to 1.09 goals and 2.19 points per hour.

Meanwhile the 2nd unit had fallen off a cliff from what I suspect are system issues. They were nudged into playing the same style as the first unit and shot rates and dangerous chances evaporated. Sam Girard’s shot rate went up 36%, good for him and all that, but that’s not a solid strategy. The rest of the 2nd unit’s rates fell, especially Kerfoot and Soderberg. Carl failed to record a PP point in this entire stretch from December 3rd to February 9th (and beyond thanks to the aforementioned drought).

This was the coaching staff’s stubborn phase so they chose to wait it out and do nothing while the team imploded instead of trying other options. They ended up waiting a long time.

The Final Stretch

The team might have woken up during the final 28 games but the power play really didn’t, with a few exceptions. RainbowVomitChart© #4:

Compher, Kerfoot and Wilson all spent a fair amount of time on the top unit. Only 5 of these gentlemen scored goals in this stretch plus Derick Brassard who added a couple after the deadline. Tyson Jost scored no points in 43 minutes of PP TOI and joined Brass as the only regulars with no assists.

This period began in the throes of the 9 game drought but surprisingly other than lack of goals it really didn’t have much of an effect. Shot rates were actually better during the drought than they were during the rest of the stretch run. It’s tough to parse out how much of that is a continuation of ineffectiveness and how much is season effects with defense ramping up as the playoffs get closer. Relative to the league they were still around 20th place on average in shot metrics so safe to say they weren’t getting better.

The biggest change here was Nathan MacKinnon regained his shooting luck, going from 0.465 G/ixG back up to 1.344, which is essentially where he was in the first 27 games. Even better, his goal scoring rate went up to 2.81 per hour which was his best stretch of the season. Offsetting that was Landeskog getting no goals but he did dramatically improve his assist rate. It’s interesting to contemplate how and why this happened, data indicates that the focus went even more towards getting Mack to shoot with Mikko being an even more distant secondary option. That’s not a bad way to think, until you look at Mikko’s goal rates and fancy shooting percentage. He was the most effective and consistent shooter all season and as much as Mack’s shooting improved, Mikko’s goal rate was 41% higher down the stretch.

PP2 was an afterthought at this point. They still plodded along with their 30% share of TOI but as far as I can tell they didn’t score at all as a unit. Mack double shifted more during the stretch and I have vague recollection of maybe a goal or two scored with the 2nd unit. Mostly they mopped up and handled the transition back to 5v5. The strength of this group from the early season was completely absent.


I didn’t tear this apart much because it’s a different animal and they were so bad in the 2nd round it just creates anger. Against the Flames they were not great but didn’t have to be. Shot rates were low but quality was decent, scoring rate ended up 9th of the 16 teams.

San Jose completely obliterated the Avs power play in round two. They had 2 goals in 23 opportunities, which was similar to the Sharks (2-for-20) and Blues (2-for-22) in the round, but the shot rates were ghastly. They had less than one shot attempt per minute, which wasn’t just last it was 35% behind next to last. Unblocked shot attempts and shots on goal were similar. Expected goal rate was 53% of the Hurricanes in 7th. It was disgusting, they were severely outcoached and with the series as close as it was it might have lost it for them. At the very least it was a contributing factor.


There are entire books and hour-long videos devoted to power play systems so this is going to be description for the sake of context rather than analysis. What the Avs do is common in the NHL and while not extremely modern or creative it is tried and true. This is a fair example of a routine setup:

This is a typical 1-3-1 power play setup. MacKinnon has the puck and this is generally the strong side for Colorado. Barrie is off-camera at the point, Mikko is on the weak side and Landy is in front of the net. The guy in the middle is the “bumper” or slot shooter. The PK here is in a fairly conservative triangle plus one.

One reason this formation is so popular in the NHL is that it’s versatile. Many teams use it but there are also many ways to go from here. It comes down to philosophy.

Passive vs Active – A passive philosophy boils down to waiting for the opponent to make mistakes and capitalizing, active teams try to force mistakes. The Avs are generally passive, which isn’t as negative as it sounds. Teams will eventually make mistakes and with the amount of skill Colorado has they’ve been decent at turning that into goals. The problem comes from a team that doesn’t make a of of mistakes in a playoff series like we saw vs San Jose. Without the ability to force mistakes they were shut down hard.

Vertical vs Horizontal – This describes the general motion in the offensive zone. The Avs are very vertical and became even more so this season. They move the puck high to low, low to high looking for openings. Horizontal teams have more “East/West” motion and passing, which is more dangerous for both them and the PKers. The thinking with low to high is opening up a gap in the defense and slipping a royal road pass through to the weak side. When it works it’s deadly but again it’s more of a waiting game than forcing trouble.

Perimeter vs Probing – This mainly has to do with how often a team is willing to put the puck in dangerous areas and how comfortable they are playing in the slot. The Avs are very much a perimeter team but when we saw PP2 having success early this season they used a probing philosophy extremely well. When that went out the door they became totally ineffective. It depends on having a skilled bumper and on PP1 that’s the guy that almost never touches the puck. They’ve experimented with putting a skilled guy there, usually Landy, rarely Mikko, but it was usually Jost/Compher/Wilson/etc and other than rebound shots they were not involved. More than anything else, changing who is in the middle and how involved they are could transform the Avs PP into something more consistent and effective.

In each of these three areas, the coaching staff tends to the conservative side which again probably sounds a lot worse than it really is. Everyone remembers years past when the Avs would give up massive amounts of odd-man rushes on the power play, now they are one of the stingier teams in the league in that area along with shorthanded goals against. There is merit in that for sure but the next step is the ability to add a little risk at times when things aren’t clicking.

Key Factors

MacKinnon’s shooting percentage – In the first 3rd of the season Mack was scoring 1.35 goals per expected goal, during the collapse that went down to 0.47 and jumped back up to 1.34 down the stretch. It’s interesting that his collapse coincided with the PP’s collapse but when he regressed back to normal it didn’t have an effect on efficiency. Basically whatever they did fixed Mack but that came at the expense of everyone else’s production. He’s the most dynamic shooter on the team but being in the situation where when he sneezes everyone catches a cold makes the PP too one-dimensional. This isn’t an easy problem to tackle. Mack shooting crazy volume is a good thing so finding a way to make that benefit the other 4 players on the first unit has to be a priority.

Mikko’s consistency – Over the season he scored 2.72 goals per expected goal and there wasn’t much deviation from that in the three sections we’ve studied. Even though he’s not the primary option he’s quite deadly. What was interesting to me was how his shot & quality rates fluctuated. Early on he was low-volume/medium-quality, during the collapse he was high-volume/low-quality and down the stretch he was medium-volume/high-quality. Best goal rate happened at the end and it would appear that’s when he found the best balance between volume and quality. Wow, actual learning on the job. The fact that he was scoring at a fairly consistent G/ixG rate while going through these changes says a lot about his skill. It’s scary to think he’s got upside left but it seems like he’s still not near his ceiling.

Point shot ineffectiveness –  Avs defensemen scored 2 goals on a total of 56 shots on goal and 76 unblocked attempts. They generated exactly 4 expected goals for an aggregate 0.5 G/xG rate. I am decidedly not a fan of point shots, especially on the power play, but that’s no reason for the D’s to generate so few scoring chances. I don’t blame them for the most part, they generated scoring chances at higher rates than 5v5, especially Sam who generated chances more than twice his normal rate. Of the two regular PP QB’s he was more able to move into space and get meaningful shots off towards the net while Barrie often let loose low-quality shots from far away. QB is a poor analogue for what a D should do in this situation, point guard is more apt. Other than the completely ignored bumper spot, the D was the most under or misused position on the Avs PP.

No bump from the bumper – Speaking of the bumper, this needs major work. Colin Wilson laid the foundation for how this position functions with his excellent run in the playoffs but it’s only a start. His ability to tip and screen was fine but someone that can handle the puck and make shoot/distribute decisions quickly would be another level up. This should be a highly involved player in the PP, not a mop-up man.

Conservative philosophy – The staff finally got rid of the hideously awful rush chances against but it came at the expense of creativity and options. The hope is that with confidence in the personnel some of that can be added back in when appropriate. There’s no reason why horizontal motion and more use of the slot shooter can’t be part of their system going forward. I think most importantly they should have more of an active mentality, it’s something that’s difficult to turn on and off. Often times a power play will destroy momentum because the style is such an anathema to how they play at 5v5. Everything about the power play is slow and timid from the entry to how scoring chances are developed and that carries over into normal play far too often.

The Drop Pass – Aside from the safe & deliberate play inside the offensive zone, the aspect of the PP fans seem to hate most is the default zone entry. The perception is that they lose lots of faceoffs (they don’t, 54% on the PP) and have to start from behind their own net most of the time. From there the initial puck carrier casually brings the puck forth and once they hit the red line, drops it off to the real puck carrier with the charming spontaneity of a rocket launch. When it first appeared several years ago it was fresh and new, now it’s gotten a little stale. In all honesty it’s still effective and it’s tough to imagine some other strategy making a meaningful change in success rates. PK coaches have realized that denying the zone is a great way to avoid getting scored on and have put a lot of emphasis on doing so. The Avs do ok with it and the main complaint I have is that they execute it with the same slow, passive mentality that infects the entire PP system. Unless that changes there’s not a whole lot to complain about as far as how they enter the offensive zone.

A note on Vladislav Kamenev – His injury happened right after the Avs collapse began in early December. At the time he was 10th in forward PP TOI with around a minute per game. These are all small (22:34) sample sizes but he was 2nd among forwards in iCF/60 at 23.93 and SOG/60 at 13.29 and had a goal to his credit. It’s almost certainly coincidence that the 2nd unit fell off a cliff when he destroyed his shoulder but it sure didn’t help. He’ll probably start the season in the press box while an aging grinder takes his spot but look for him to make a difference this season if he ever gets a chance.

⇐     ⇔     ⇒

The Avalanche power play was lucky then bad then one-dimensional during the regular season. In the playoffs it was mediocre against a bad opponent then got rolled over by a team that had a terrible penalty kill in round one. None of these is particularly reassuring about next year being any different. I have less faith in Ray Bennett than I’d like as a power play coach if for no reason other than Pierre McGuire thinks he’s a wizard which means that he’s probably bad and at least 5 years behind the times. At times it seems like Coach Bednar puts a lot of the PP issues on the players, which is ok but when the problems are chronic it means either they haven’t bought into what’s being coached or what’s being coached isn’t rational or possible. The struggles that the Eagles went through on the PP (last in the AHL at 13.8%) with the same basic system reinforce that there’s more than execution issues here. Whether it’s the coaches’ system/philosophy or the players’ mentality/execution doesn’t really matter. It’s an area the team has the talent to succeed and fell short of expectations.

This summer was an opportunity to make a change and bring in someone with modern views on developing offense but they stood pat. Enough players were swapped into the roster that won’t be an excuse next season. In my opinion there was plenty of talent to make a competent man advantage system work last season, now there’s even more so if it doesn’t happen it’s 100% a coaching issue.

Thanks to the NHL and especially Natural Stat Trick for the raw statistics.


Scoring LW, punchy climber for the Ardennes classics, spirit guide

%d bloggers like this: