A statistical analysis of the Calder Trophy race

Lucas Friesen is a student at Brock University. He also runs his own blog numberpuck.wordpress.com and can be followed on Twitter @LucasFriesen

With every team now 50 games into the season, I figured it was a fairly good point to start assessing the Calder Trophy race around the league. When it comes to this award, it really needs to be seen which rookies are creating value for their team. In the past, Calder winners have been assessed on the basic statistics (goals, assists, points, plus/minus, etc.). With the new advanced stats era, this award needs to be examined using new measures to determine a rookie’s true contribution to their team.

First, I needed to determine what rookies I should examine. To be considered for the Calder, you need to make an impact on your team. To narrow my pool of rookies, I took all rookie skaters, classified by NHL.com, who have played 30 or more games so far this season (35 players as of and up to February 1). This number of games should limit the amount of random variance I see in these numbers (given it already is a small sample). I figured that 30 games was a good starting point. Any rookie that looks to be considered for the Calder should have played this amount of games so far. Also, to be the most proficient rookie (which is the definition of the Calder Trophy winner), you must be in the lineup on a consistent basis.

Before I get into my analysis, I just wanted to point out one thing I saw in the data. Below is a graph comparing each players on-ice PDO and their P/60. This is showing a strong correlation between these two variables, with an r-squared value of 0.317. This can be interpreted as 31.7% of the variation seen in P/60 is determined by PDO. Why I wanted to note this is because it is tough to analyze statistics for rookies because of the small sample size. With such little data to draw from, this type of analysis must be coupled with the traditional “eye-test” to properly evaluate rookies over a single season. Also, this shows that using only P/60 isn’t the best way to analyze rookies, as it is determined a lot by PDO (or “puck luck”).

calder 4

For my direct analysis, I focused on looking 6 forwards and 2 defensemen from my list of 35 players. To get these select few, I looked at numbers across the board and found which players were consistent or “good” in many categories, were on a team that was struggling yet they were succeeding, or significantly contributing to a successful team. Not only did I look at raw numbers, but I wanted to focus on relative numbers to see how they are doing compared to their team. Also, TOI/gm and GP did factor in when narrowing down the list because a larger sample means less randomness. From here, I can then compare apples to apples (or “good” rookies to “good” rookies). I felt like this would be the best way to truly analyze the best rookies in the league this season. All data was retrieved from war-on-ice.com.

Here are the 6 forwards I selected and some of their numbers, in no particular order:

calder 3

All numbers at 5v5 play for the 2014-2015 season, up to and including February 1, 2015.

P60=Points per 60 minutes; CF%=Percentage of all on-ice shot attempts for; CF%Rel=On-ice Corsi-For Percentage minus Off-Ice Corsi-For Percentage; PDO=On-Ice Shooting Percentage plus On-Ice Save Percentage; ZSO%: Percentage of all on-ice non-Neutral Zone Faceoffs in the Offensive Zone; ZSO%Rel: Percentage of all on-ice non-Neutral Zone Faceoffs in the Offensive Zone minus the Percentage of all off-ice non-Neutral Zone Faceoffs in the Offensive Zone; SCF%: Percentage of on-ice scoring chances for; CorT%: TOI-weighted corsi-for percentage of teammates; CorC%: TOI-weighted corsi-for percentage of opponents.

What I first want to note is the two Senators players on the list. These two, Hoffman and Stone, have been impressive thus far on a struggling Ottawa squad. Besides Karlsson, they lack a defensive core and these two have still been able to drive solid offensive rates, much better than the rates when they are off the ice. Hoffman has been dominant, with 2.7 P/60, which actually ranks him 11th out of all skaters in the league with 400+ minutes at 5v5 this season and the best amongst all rookies I examined. While he does get more offensive zone starts than his team average, Ottawa as a whole doesn’t get many offensive starts (43.8%), which doesn’t necessarily show an offensive deployment tendency. The only thing I want to pick at is that he is riding an extremely high PDO number. 106.5 is not sustainable as we know, which doesn’t bode well for him in the future. However, if he can finish this season off strong and hold a high number, the Calder discussion wouldn’t be complete without him.

Stone plays a different type of game than Hoffman, but still puts up good numbers. You can see is zone start number is lower, possibly showing a larger defensive component to his game. He generates just less P/60 than Hoffman, while driving a higher possession number all in relatively the same amount of ice time. All six players have similar teammate and opponent numbers, which doesn’t show much for differences or substantial “stand-outs.” What Mark Stone has going for him, though, is a reputation as a strong possession driver, being top on this list in relative CF% and top among my 35 rookie skaters as well.

Now on to probably the most recognized name on that list: Filip Forsberg. This guy drives possession at a rate of 56%, has a good relative CF% on an already good possession team, and is substantially contributing to one of the more surprising team stories in the league this season. He gets the most ice time out of all these players, with 13.4 minutes of 5v5 play each game, showing his prominent role on Nashville this year. As you can also see, Forsberg is able to get chances of high quality, carrying the highest SCF% of these 6 forwards and third best in my list of 35. What I really want to see is some usage numbers from this young star, because he sure impressing me with his raw number production.

Maybe the two surprises (to be considered for the Calder) I included in my list are Anders Lee and Andre Burakovsky. With a negative relative zone start number, Anders Lee is still able to be a positive possession player, and a positive relative corsi player for that matter, on a shockingly good Islanders squad this year. Generating over 2 points per 60 minutes, Lee is finding a nice spot with the Islanders. I included Burkavosky on this list mainly for his offensive threat and ability to put up points and good possession numbers on an average Washington Capitals team. With the lowest PDO in the group, he is not seeing the “luck” that some other rookies are seeing which could be playing as a detriment to his numbers.

I felt like I had to include Johnny Hockey on this list. He’s been one of the driving forces on a Flames squad that is in a playoff spot while being way below average possession-wise. With a 2.7% CFRel and a 47.1% CF, he is generating more possession than his team, while playing with fairly poor teammates (45.8% Teammate Corsi). While he may be a long shot for the Calder Trophy, based on the others considered, he is showing to be a good player as far as fancy stats are concerned.

I want to mention a few notables that were left off my list of 6 forwards that some may think deserve to be here. Tobias Rieder has been great for Arizona thus far, and was mainly left off the list because of his lack of offense so far (1 point per 60 minutes). He is finding a nice role on the team and is very valuable to that franchise. Another is Jonathon Drouin; this guy has been snake-bitten all year, riding a 2.9% personal shooting percentage at 5v5, scoring only 1 goal. His possession numbers are good, but has been sheltered by offensive zone starts so far. He has also played with good linemates (54.5% CorT%), and against relatively weaker opponents (48.5% CorC%). He just hasn’t stood out this year and been as proficient as others, which is why he was left off this list.

calder 2

All numbers at 5v5 play for the 2014-2015 season, up to and including February 1, 2015

Now on to my defensemen, which are showing in the chart above. Mainly, there are two standout rookie defensemen this season: Aaron Ekblad and John Klingberg. So far, Ekblad has been crucial to the transformation of the Panthers. Hovering around the 8th spot in the East, the Panthers have been very much improved so far this season. Getting solid TOI/game at 5v5, Ekblad has been able drive possession at 53.7% and post a CF%Rel of 2.8%. His team is significantly better with him on the ice. His zone start numbers show he is driving play into the offensive zone, and coupled with this possession numbers, this makes him very valuable. Limiting scoring chances is something Ekblad does well also, controlling these high quality chances at a 56.5% rate for while on the ice. Overall, Ekblad has played more like a veteran defenseman rather than a rookie so far in 2014-2015.

John Klingberg has made a nice case for himself as of late as well. The winner of the Rookie of the Month for January has been strong on an inconsistent Dallas squad so far this season. His CF%Rel at 3.2% shows he is capable of the large minutes he is getting (17.4 minutes at 5v5 per game). His CorC% (competition corsi) is greater than Ekblad’s, showing he may be facing strong opponents. But with his lower zone start number, and a negative relative zone start percentage, this shows he may be deployed more for defensive situations than Ekblad. Klingberg has been stellar in his time in the league so far and could transition into a solid top pairing defenseman if he keeps it up.

After looking at each rookie and comparing them, it’s only fair that I make my early season choice for the Calder Trophy winner. As said, both the eye-test and the stats need to be used together when looking at first-year skaters because of the small sample size and large variation in numbers for only one season. My vote for Calder goes to Filip Forsberg; this guy has not looked out of place in the NHL and is posting solid numbers. While Ekblad and Klingberg have been important to their respective defensive cores, and even with Michael Hutchinson showing value and impressing in the net for Winnipeg, Forsberg has been simply outstanding for the Predators.

As a final note on Forsberg, I want to look at his usage, in the form of Domenic Galamini’s (@MimicoHero) new “HERO” charts on ownthepuck.blogspot.ca. Forsberg, while getting low second line minutes at 5v5 per game, has been outperforming his TOI in almost every category Galamini displays. His production (G/60, A/60, and P/60) is a first line rate, and he is definelty serving as a great top-6 forward in terms of possession (Usage Adjusted Corsi %). Overall, Forsberg has been proficient so far in 2014-2015 and would be a justifiable choice for the Calder Trophy at year’s end.

calder 1


  • nhl.com
    • Statistics Page
    • Rookie Qualifications:
      • To be considered a rookie, a player must not have played in more than 25 NHL games in any preceding seasons, nor in six or more NHL games in each of any two preceding seasons. Any player at least 26 years of age (by September 15th of that season) is not considered a rookie.
      • war-on-ice.com
        • Skater Statistics Page
        • ownthepuck.blogspot.ca
          • Tableau Visuals Page


One thought on “A statistical analysis of the Calder Trophy race

  1. Pingback: Calder Trophy Race Analysis: dCorsi and Usage Comparison | NumberPuck

Leave a Reply

Your email address will not be published. Required fields are marked *