Welcome to the site! As previously mentioned in Objective Avalanche’s welcome post, this piece will focus on Nathan MacKinnon’s performance during the 2014-2015 season, and will be the first in a series of Season Reviews that I’ll be posting as we approach the beginning of the 2014-2015 NHL Season. I’d also like to specifically thank Emmanuel Perry and Micah McCurdy for their “contributions”, given that their work features prominently in this piece of writing. I’ve included significant portions of their methodologies/explanations below many of the visuals, seeing as this might be the first time some of you have seen them. As a result this post is a bit lengthy, but hey, there’s lots of pictures. Now, onto MacKinnon’s review!
If you’re reading this post, then there’s a very good chance you’re familiar with Nathan MacKinnon’s story. However, on the off chance that you aren’t, and for the sake of being thorough, I’ll provide a bit of a background before diving into the analyses.
Nathan was born on September 1st, 1995 in Cole Harbour, Nova Scotia, and grew up playing minor hockey for Sidney Crosby’s former club, the Cole Harbour Red Wings. Following seasons of 110 and 145 points for the Bantam AAA team as an underager, MacKinnon chose to follow in Crosby’s footsteps and enrolled at Shattuck St. Mary’s, where he spent two productive seasons before being drafted 1st overall in the QMJHL Draft by the Baie-Comeau Drakkar. However, MacKinnon was strongly considering the USHL/NCAA route to the NHL at the time, and fearing a situation where they were unable to come to terms with the young star, Baie-Comeau dealt MacKinnon to his hometown Halifax Mooseheads for a package that included three first round picks, and two players.
MacKinnon went on to score 153 points in 102 regular season games as a Moosehead, leading his team to the 2013 Memorial Cup in the process. At the Memorial Cup, MacKinnon exploded for 7 goals and 6 assists in a mere 4 games, a spectacular run that included a 7-4 drubbing of Seth Jones’ Portland Winterhawks in which MacKinnon notched a hat trick, and solidified his status as the top prospect available at the 2013 NHL Entry Draft. Halifax went on to win a franchise first Memorial Cup, with MacKinnon being named tournament MVP. While there was speculation that the Colorado Avalanche, proprietors of the 1st overall selection, would be addressing an organizational need and selecting Denver native Seth Jones with their pick, both Patrick Roy and then head scout Rick Pracey publicly stated that they were leaning towards selecting MacKinnon. Unsurprisingly, MacKinnon’s name was the first one called on draft day, and he became the newest member of a young and exciting Avalanche squad.
Both MacKinnon and the Avalanche experienced immediate success at the onset of the 2013-2014 NHL Season. MacKinnon managed 2 assists in a 6-1 opening night upset of the Anaheim Ducks and never looked back, quickly earning a regular spot alongside Paul Statsny and Gabriel Landeskog on one of the club’s top two lines. The upstart Avs continued their improbable run, buoyed by a combination of youthful vigour, timely performances and sheer dumb luck, and ended up as Central Division champions following a collapse by the St. Louis Blues over the final 10 games of the season. Unfortunately, the Avalanche were (painfully) eliminated by the Minnesota Wild in the first round of the playoffs, spelling an end to the Avs’ Stanley Cup dreams. MacKinnon tallied 24 goals and 63 points during his rookie season, breaking records belonging to Dale Hawerchuk and and Wayne Gretzky along the way, and was rewarded with the Calder Trophy for his efforts.
Data and Discussion
The analysis of various data and data visualizations will form the bulk of these posts, and is also the reason that I assume you’re here and still reading. Seeing as this is my inaugural Season Review, I will be explaining many of the charts and methods in full; expect my future pieces to be a little more to the point.
First, we have a game-by-game breakdown of MacKinnon’s 2014-2015 season, courtesy of the talented Micah McCurdy. A picture truly is worth a thousand words, and McCurdy’s fantastic data visualizations are worth that and then some.
The first strip shows MacKinnon’s 5v5 (even-strength) linemates. MacKinnon spent the majority of his time paired with Gabriel Landeskog, as was the case the year prior. After that, it gets a little more hairy. MacKinnon spent decent chunks of time with Iginla, Briere, Tanguay and Duchene early in the season before finding himself on the third line, toiling alongside Maxime Talbot and John Mitchell while in Roy’s proverbial “doghouse”. Eventually, Roy landed on the Landeskog-O’Reilly-MacKinnon combination that was highly effective for the Avalanche over the latter half of the season, and it’s no surprise that this corresponds with the period during which MacKinnon was the most successful individually. Unfortunately, MacKinnon broke his foot just as this trio was beginning to gel, sidelining him for the remainder of the season and limiting his campaign to just 64 games.
MacKinnon averaged 17:03 of ice time per game last season, down from 17:21 in 2013-2014, and his usage varied wildly throughout the season depending on Roy’s mood, his perception of MacKinnon’s level of effort, and perhaps the weather in Denver as well. This isn’t unusual to see for a young player; coaches will often decide their pupil needs a rest, or perhaps a swift kick in the butt to “jump-start” their play. MacKinnon’s special teams usage, however, remained relatively consistent throughout the season.
The third strip shows adjusted shots per sixty minutes. The black line shows all shots (including blocks, misses, saves, or goals) that were taken by Colorado at 5v5 while MacKinnon was on the ice. The red line shows the same thing for Colorado’s opponents. When the red line is above the black line, MacKinnon was mostly hemmed in his own zone, losing puck battles, not breaking the puck out, etc.; all things that lead to the opponent having the lion’s share of the puck. When the black line is above the red line, Colorado was taking the majority of the shots during MacKinnon’s shifts, suggesting that MacKinnon was driving play in these times; winning faceoffs, passing well, gaining the zone, doing the things that generate offence (Micah McCurdy, 2015). Shot metrics are, broadly, the most repeatable measures of skill for which we have good information. The adjustment method used is score-and-venue adjusted shots, where each shot counts slightly less when it is historically easier to generate; that is, when trailing, or when at home (Micah McCurdy, 2015). This metric is the single best, well-known predictor of future success for teams.
Unfortunately, Colorado has struggled mightily possession-wise over the previous few years, and consistently gives up more shots than they take themselves. Fortunately, MacKinnon has managed to mostly buck this trend, improving his Corsi For % from 47.4 in 2013-2014 to 49.5, and his Relative Corsi % from -0.7 to 7.5, all while starting an (essentially) equal number of shifts in the offensive (49.9%) and defensive (50.1%) zones. McCurdy’s visualization confirms these numbers; there are evidently stretches of games where the Avs are driving possession while MacKinnon is on the ice, particularly while he was playing with Landeskog and O’Reilly.
The fourth strip shows 5v5 goals per adjusted shot, in black for Colorado while MacKinnon was on the ice, and in red for Colorado’s opponents during his shifts. League average (not indicated) is around 5%, considerably lower than the league average for goals per shot-on-goal. The effect of individual players on either metric is extremely limited and is mostly explainable by luck, even over periods as small as several games. However, decisions about players are still routinely made on the basis of these results, so they’re important for understanding the history of a season. MacKinnon experienced several scoring slumps during the early portion of the season, often coinciding with his time spent on Jarome Iginla’s line. This perhaps suggests that their playing styles are marginally incompatible, but it’s not really indicative of anything noteworthy.
The final strip shows all of MacKinnon’s goals, primary assists, and secondary assists during the 2014-2015 season (not just 5v5). His goals are shown in dark blue, his primary assists are shown in light blue, and his secondary assists are shown in green. A different colour was used for secondary assists in order to differentiate them, as they’re commonly cited but are, in fact, almost totally random. Their primary value is as a measure of luck, or a lack thereof. MacKinnon’s 0.31 secondary points/primary point is fairly average.
MacKinnon’s seeming inability to score during the 2014-2015 season frustrated legions of Avalanche fans, who were hoping that their wunderkind was set to build on his 63 point rookie campaign. “He doesn’t drive the net anymore, he just pulls up inside the blueline and waits for the high man” was a common complaint, and was at times partially valid. However, MacKinnon was very obviously a victim of an unsustainably low shooting percentage. The average shooting percentage between 2005 and 2013 was 10.82% for forwards, and most elite forwards are able to sustain rates between 13% and 16% for entire seasons, with some outliers even approaching 20% (cough, Foligno, cough, Hudler). MacKinnon’s shooting percentage last year was a paltry 7.29%, and it’s very likely that he’ll have a few more bounces go his way this upcoming season.
In terms of shot type and distance, MacKinnon’s 2014-2015 season was quite similar to his rookie season. He favours his wrister when he shoots off the rush, and often employs his backhand when he’s in tight. Interestingly, MacKinnon’s shooting percentage during his rookie season was also below the historical forward average of 10.82%; are we primed for an outburst of scoring from MacKinnon this coming season? The statistics certainly suggest it.
Here we have a “With Or Without You” chart for the entire Colorado Avalanche team last season. The red numbers are based on a larger number of minutes than the blue numbers, and as such are more heavily clustered than the blue numbers. The legend also indicates how many 5v5 minutes the player played; players with fewer than 200 5v5 minutes (e.g. Joey Hishon) are shown faded to suggest uncertainty. Players with fewer than 100 5v5 minutes (e.g. Duncan Siemens) are not shown at all. All shots are adjusted, as is typical of McCurdy’s work.
When MacKinnon was on the ice, the Avalanche generated shots at a rate of 56 per 60 5v5 minutes (a team high), and the Avalanche held their opposition to ~60 shots/60. When MacKinnon wasn’t playing, the Avs generated a mere ~45 shots/60, while their opposition was able to fire the biscuit at pool ol’ Semyon Varlamov ~62 times every 60 minutes. You don’t need a math PhD to see that MacKinnon’s presence on the ice is correlated with shot generation for the Avalanche, and he’s one of the team’s better shot suppressors to boot.
This WOWY shows every player with whom MacKinnon played at least 100 5v5 minutes, and offers more specific (and interesting!) insights. The legend lists all these players, with their jersey numbers, and the number of minutes each one played with MacKinnon and without MacKinnon. The lines connecting the blue-boxed numbers for “MacKinnon without player” and the black-boxed “MacKinnon with player” all intersect in a common point at the centre of the chart, this point represents the average of the team’s result with MacKinnon and the team’s result without MacKinnon for the year. The legend shows the number of 5v5 minutes played by each player without MacKinnon in the column headed “x-5”, the number of 5v5 minutes played by each player with MacMinnon in the column headed “x+5”, and the number of 5v5 minutes played by MacKinnon without the given player in the column headed “5-x”.
For instance, MacKinnon played 517 minutes with Landeskog, and Landeskog played 253 minutes without MacKinnon. When on the ice together (the grey 92 box), the duo are able to generate shots at an above average rate, and manage to eclipse their opponent’s shot generation rate as well, which was no small feat for last year’s Avalanche squad. However, Landeskog is less effective when MacKinnon is not on the ice (the red 92 box), and MacKinnon experiences an even more pronounced decline in effectiveness when parted from Landeskog (the blue 92 box). This indicates that the sum of the duo’s impact on the ice is greater than their individual parts, as it were, confirming a certain anecdotal “chemistry” that Avs fans are familiar with.
Broadly speaking, everyone who played at least 100 5v5 minutes with MacKinnon generated more shots while he was on ice than they did otherwise, and many saw an improvement in shot suppression as well. Something that jumped out at me right away was the positive impact that MacKinnon’s presence on the ice had on the play of Zach Redmond (the grey 22 box). The sample size is small (a mere 124 minutes of 5v5 playtime together), sure, but it’s still quite interesting. I was also surprised at how ineffective Iginla was without MacKinnon on the ice; obviously he made use of his heavy shot on our (29th ranked) power play that ran directly through him, but many metrics suggest that his net on-ice impact is no longer positive.
“Spider Diagrams” are McCurdy’s name for showing the 5v5 on-ice results obtained by a player, broken down according to their teammates. MacKinnon’s overall results for all 5v5 playtime are shown in the centre of the diagram with his jersey number, 29. The x-axis scale is shots for MacKinnon’s team (Colorado), and the y-axis scale is shots for Colorado’s opponent, hence, the best results are in the bottom-right corner. As always, the shots are adjusted. Every combination of players that had at least 15% of MacKinnon’s total 5v5 minutes is shown. For individual players, the percentage of MacKinnon’s ice-time spent with that player is shown in the legend.
As discussed earlier, MacKinnon and Redmond were a potent duo, generating far more shots than their opponents while on the ice. The 92-90-29 line appears to be have been adept at keeping the number of total on-ice events in check, while the 92-29-12 line seemed to have thought they were playing for the Gretzky-era Oilers. A closer look suggests that Iginla is the x-factor in this case. MacKinnon’s play was understandably hampered during his time spent with borderline NHLers such as Nate Guenin and Nick Holden.
The HERO chart doesn’t tell us much we hadn’t already guessed after working through MacKinnon’s game-by-game history, but it’s a terrific way to visualize some of the the core statistics that indicate player performance, and a useful method for quickly determining what “tier” a player belongs in based on said performance (Are they a first liner? A third liner? How good is x aspect of their game?). Note that the data used for the chart encompasses MacKinnon’s rookie season as well.
The first thing you’ll notice is that MacKinnon played a slightly higher percentage of his minutes while the Avalanche trailed than he did when they led, suggesting that he’s someone who was utilized in tough situations when the Avalanche desperately needed a goal. This metric is of course impacted by the Avs being in a trailing position in games more often than not, but it’s still interesting to see, given that MacKinnon wasn’t relied upon heavily in trailing situations during the earlier portion of his rookie season. The only real outlier on MacKinnon’s HERO Chart is his poor Corsi Against 60 +/-, which is understandable given how shockingly bad the Avalanche have been possession-wise during MacKinnon’s tenure with the team. Unsurprisingly, MacK produced at a mostly first-line rate.
Pictured above is the Bombay Chart for Nathan MacKinnon’s 2014-2015 season. Bombay Charts are one of my favourite data visualizations; they’re both a) intuitive (larger polygon = better performance/bigger impact) and b) include a wide variety of parameters (nothing beats being thorough). A full explanation of the methods (and madness!) behind the Bombay Chart can be found over at War On Ice courtesy of Emmanuel Perry. However, I’ll do my best to summarize.
Essentially, the Bombay Ratings make use of Euclidean distance to compare players to a theoretical player (Gordon Bombay) who possesses the best stats we can imagine without stepping outside the boundaries of what real players have been able to accomplish. The similarity calculation evaluates “distance” between players, each occupying a position in imaginary space. This space has as many dimensions as there are categories by which you choose to compare players, and the limits of each dimension are set by the maximum and minimum recorded values since 2005-2006. Gordon Bombay is simply a marker that was placed at the positive-most position in space — the position where the positive extrema of each dimension meet. In a three-dimensional plot, this is simply a corner. The Bombay Rating is the similarity between a player and Gordon Bombay (Emmanuel Perry, 2015).
The Bombay function essentially does what we all do when we pull up a player’s statistics. The advantage is that it’s more precise, quicker, and returns a single number – Emmanuel Perry
The Bombay app Perry developed uses a variety of 5v5 stats to assign ratings to skaters based on the selected weights and generate charts comparing players to Gordon Bombay in each of the chosen categories. The outer edge of the chart represents a 100% similarity to Bombay in that measure, and is only achieved if a selected player-season possesses the best recorded value in that metric among regular skaters since 2005. The dashed grey polygon represents another fictional player – one whose stats are all equal to the league average for regular skaters at that position. At the default weights, this hypothetical average forward has a Bombay rating near 46 (Emmanuel Perry, 2015).
Here we have a comparison between MacK the rookie, and MacK the “slumping sophomore”. Hmm…
Here’s how MacKinnon compared to some of the other precocious forwards from his draft year.
I was particularly interested in the comparison between MacKinnon, who allegedly suffered from a serious sophomore slump, and Monahan. I’ve seen a lot of chatter online (mostly from Flames fans) about preferring an elite two-way centre with “intangibles” over a one-dimensional winger, but it’s obvious that assessment of MacK misses the mark by quite a large margin. Don’t get me wrong, I like Monahan, and I like his game a lot. But there’s certainly a reason MacKinnon was picked first overall just two years ago, while Monahan fell to the Flames at the sixth spot.
Here we have a comparison between MacKinnon, and a couple of other recent first overall picks.
Also interesting! How you doing, Nuge?
In fact, based on Bombay Ratings, MacKinnon had the most impressive season of anyone on the Colorado Avalanche. Here’s a look at how he compared to Matt Duchene, Gabriel Landeskog and Ryan O’Reilly, in particular.
Moving forward, I would expect MacKinnon to experience the salto de qualità (leap of quality) that often occurs during the early portion of a player’s career; think Vladimir Tarasenko last year. Obviously success is contextual, but I don’t think 30 goals and 70 points are unrealistic targets for MacKinnon this season, barring some sort of catastrophic collapse by the entire Avalanche squad. MacK’s shooting percentage was unsustainably low during the 2014-2015 season, and historical evidence suggests that that won’t be the case next season.
MacKinnon is undoubtedly the Avalanche’s most dynamic talent, and should be given the opportunity to influence and impact the game as much as possible.
Ideally, this season will also mark the beginning of MacKinnon’s transition back to his natural position at centre. While MacKinnon is slated to line up on the right wing alongside Landeskog and Soderberg to start the season, I think it’s safe to assume that Soderberg likely doesn’t figure into the Avs’ long-term plans for their top two lines. In the future, having MacKinnon become the Avalanche’s 1C would allow Duchene to assume more sheltered minutes and start an even larger percentage of his shifts in the offensive zone, a role in which I believe he would be most successful. MacKinnon is undoubtedly the Avalanche’s most dynamic talent, and should be given the opportunity to influence and impact the game as much as possible. I don’t expect to see Roy “throw him to the dogs”, per se; the NHL is a tough league, and young centres often struggle. However, it would be reassuring to see the transition begin.
Anyway, that’s all for now. Thanks for reading, and toss me a follow on Twitter to stay up to date with all things Avalanche!