Hockey Lexicon Spotlight Series – Game Score

[text_output]In the second edition of the Hockey Lexicon Spotlight Series, I will highlight one of my personal favorite advanced stats, Game Score. In the first post of this article series, I discussed Luke Solberg’s (uh, in light of this mind-blowing tweet, technically I should probably say identical twins Luke and Josh Solberg’s) teammate relative statistical model, and how it can be used to better isolate the potential impact that a player has on his team, relative to his teammates. As a reminder, the Blueshirts Breakaway Hockey Lexicon is a huge resource that attempts to explain all major advanced stats and concepts in an easy-to-understand manner, and I encourage you to check it out if you are interested in learning more about fancy stats. As always, I also encourage you to reach out to me on Twitter if you have any questions about anything.[/text_output][custom_headline type=”left” level=”h5″ looks_like=”h5″ accent=”true” id=”” class=”” style=””]What is Game Score?[/custom_headline][text_output]Game Score is a catch-all statistic created by Dom Luszczyszyn of The Athletic that quantifies the total value of a player’s productivity from a single game. Dom notes in his methodology that he got the idea for Game Score from noted basketball analyst John Hollinger (also the creator of PER), and that the original Game Score statistic is in fact from famous baseball statistician Bill James. Dom’s NHL version of Game Score incorporates the following stats in an attempt to quantify the overall performance of a player: goals, primary assists, secondary assists, shots on goal, blocked shots, penalty differential, faceoffs, 5v5 Corsi differential and 5v5 goal differential. Obviously not all stats carry the same importance, so Dom assigned weights to each of the metrics to come up with the following formula for Game Score:

Skater Game Score = (0.75 * G) + (0.7 * A1) + (0.55 * A2) + (0.075 * SOG) + (0.05 * BLK) + (0.15 * PD) – (0.15 * PT) + (0.01 * FOW) – (0.01 * FOL) + (0.05 * CF) – (0.05 * CA) + (0.15 * GF) – (0.15* GA)

As you can see from the formula, scoring a goal is the biggest thing a player can do to positively affect their Game Score, with a weight of 0.75, followed by registering a primary assist (0.7), with secondary assists carrying the third highest positive weight (0.55). Additional events that have a positive impact on Game Score (in order from greatest impact to smallest) include being on the ice for a goal for (0.15), drawing a penalty (0.15), registering a shot on goal (0.075), being on the ice for a shot attempt for (0.05), blocking a shot (0.05) and winning a faceoff (0.01). The items in the formula that carry a negative weight, and therefore decrease a player’s Game Score (in order of greatest negative impact to smallest), include being on the ice for a goal against (-0.15), taking a penalty (-0.15), being on the ice for a shot attempt against (-0.05) and losing a faceoff (-0.01).

Dom also created a Game Score model to assess goaltender performance that includes two stats—goals against and saves—which are also weighted according to importance. The goalie model is as follows:

Goalie Game Score = (-0.75 * GA) + (0.1 * SV)[/text_output][custom_headline type=”left” level=”h5″ looks_like=”h5″ accent=”true” id=”” class=”” style=””]How Can I Use Game Score?[/custom_headline][text_output]Game Score has multiple applications, and Dom frequently uses it in his writing when assessing player and team performance across single games as well as entire seasons (and everything in between). Like many stats, Game Score can be used in raw counting terms, or it can be depicted as a per-60 minutes of ice time stat. Dom notes in his methodology that, “there’s many applications for Game Score across hockey analysis that I think can further our understanding of the sport and how players work at the game level. Consistency, streakiness, clutchiness; whether they’re real or random is a question a stat like Game Score can help answer and one that we perhaps couldn’t answer properly beforehand.” For what it’s worth, I completely agree with Dom here, and I would go one step further and say that Game Score is an advanced metric that absolutely passes the smell test the vast majority of the time, which is an important element to getting buy-in from fans that are skeptical of these types of metrics.

For a single game application of Game Score, let’s take a look at Monday night’s game (March 26) against the Capitals. Using Corsica’s Live Game feature, you can view the all-situations Game Scores for all players on both teams. The Rangers’ top-three players in the game as measure by Game Score were: Jimmy Vesey (1.5), Filip Chytil (1.4) and Lias Andersson (1.4). When you look at the individual numbers that comprise Game Score, it makes sense why these were the top-three finishers. Lias Andersson tallied his first career NHL goal, which contributed a +0.75 to his Game Score, while Chytil and Vesey registered the primary assists for the Rangers’ goals on the evening, contributing +0.7 for to each players’ Game Score. Further, all three players had strongly positive Corsi for metrics on the evening, giving them positive contributions in the shot attempt differential portion of the equation, and both Vesey and Andersson were on the ice for one more goal than they were off the ice for. Despite having a neutral goal differential, Chytil’s Game Score was bolstered by a team-leading six shots on goal, good for a total contribution of 0.45 to his Game Score. All of these individual positive elements to each player’s evening add up to equate to the three of them leading the team in Game Score.[/text_output][image type=”rounded” float=”none” link=”true” target=”blank” info=”tooltip” info_place=”bottom” info_trigger=”hover” src=”2403″ alt=”Photo Credit: Bruce Bennett/Getty Images” href=”https://www.blueshirtsbreakaway.com” title=”Photo Credit: Bruce Bennett/Getty Images” info_content=”Photo Credit: Bruce Bennett/Getty Images” lightbox_caption=”” id=”” class=”aligncenter” style=””][text_output]In terms of a season-long application of Game Score, let’s take a look at the entire season’s worth of data for the Rangers. Across all situations, Mika Zibanejad leads the Rangers in total Game Score with a 53.16, followed by Vladislav Namestnikov (49.35), much of which came from his time on the Lightning, and Rick Nash (43.64). You can also filter the metric to only view 5v5 data. At 5v5 across the full season, Vladislav Namestnikov becomes the team leader in Game Score with a 32.84, followed by Rick Nash (31.54) and Mika Zibanejad (28.89). If we want to adjust for uneven playing time or games played, we can also view the Game Score as a rate statistic at both 5v5 and across all situations. For all situations (minimum 200 minutes player), Mika Zibanejad still reigns supreme on the team, with a Game Score per-60 of 2.7, followed closely by Chris Kreider (2.69) and Ryan Spooner (2.62). Lastly, at 5v5, Ryan Spooner leads the team in Game Score per-60 with a 2.48, followed by Chris Kreider (2.2) and Rick Nash (2.17).

Another season-long application example of how we can use Game Score is by analyzing the data for the entire league to supplement our analysis of the Hart Trophy discussion. To this point in the season (prior to the March 28 games), all teams have played between 74-77 games, and the arguments for who is most deserving of the Hart Trophy, the league’s MVP award, are rampant. In the past week, I have heard at least five names prominently discussed as deserving for the Hart from NHL pundits, including Nathan MacKinnon, Taylor Hall, Anze Kopitar, Connor McDavid and Nikita Kucherov. As a reminder, Game Score quantifies a player’s total productivity across each individual game, so in theory it should serve as a strong barometer for who has had the best statistical season. Obviously the Hart Trophy is not awarded to the player with the best statistical season, and instead is awarded to “the player judged most valuable to his team,” which adds another element to the discussion. That said, Game Score still is a valuable tool for helping to analyze players that belong in the discussion for the Hart.

Across all situations, Connor McDavid leads the NHL in Game Score by a comfortable margin, accumulating a total Game Score of 103.19 across the whole season. The other players that have been prominently featured in the Hart Trophy discussions these past few weeks all finished in the top-15 in the NHL in Game Score, with Kucherov placing 2^nd (97.57), Nathan MacKinnon 7^th (86.48), Anze Kopitar 12^th (84.01) and Taylor Hall 15^th (81.63). The top-10 in the NHL in this statistic also include Artemi Panarin (3^rd), Claude Giroux (4^th), Sidney Crosby (5^th), Alex Ovechkin (6^th), David Pastrnak (8^th), Evgeni Malkin (9^th) and Johnny Gaudreau (10^th). In terms of all situations Game Score per-60, Nikita Kucherov finished 2^nd in the NHL among players with at least 500 minutes played with a 3.99, while MacKinnon placed 4^th (3.86), McDavid 6^th (3.74), Hall 9^th (3.6) and Kopitar drops all the way down to 45^th (2.97). Other members of the top-10 include Brad Marchand (1^st), Patrice Bergeron (3^rd), David Pastrnak (5^th), Evgeni Malkin (7^th), Jonathan Marchessault (8^th) and Auston Matthews (10^th).

If you want to look at only the 5v5 Game Score data, McDavid still leads the league in total season Game Score with a 74.22, while Kucherov places 3^rd (63.02), MacKinnon 6^th (55.73), Hall 19^th (50.56) and Kopitar 28^th (48.52). The top-10 also features Artemi Panarin (2^nd), David Pastrnak (4^th), Johnny Gaudreau (5^th), Dougie Hamilton (7^th), Brad Marchand (8^th), Vladimir Tarasenko (9^th) and Claude Giroux (10^th). Finally, in terms of 5v5 Game Score per-60, Connor McDavid finished 4^th in the NHL among players with at least 500 minutes played with a Game Score per-60 of 3.55, while Kucherov finished 5^th (3.35) and Mackinnon 7^th (3.33); however, Taylor Hall drops all the way down to 31^st (2.86) and Kopitar comes in at 70^th (2.43). The top-10 in the NHL also includes Brad Marchand (1^st), Patrice Bergeron (2^nd), David Pastrnak (3^rd, and holy shit was that line good this year), Auston Matthews (6^th), Evgeny Dadonov (8^th), Jaden Schwartz (9^th) and Craig Smith (10^th).[/text_output][image type=”rounded” float=”none” link=”true” target=”blank” info=”tooltip” info_place=”bottom” info_trigger=”hover” src=”2404″ alt=”Photo Credit: Codie McLachlan/Getty” href=”https://www.blueshirtsbreakaway.com/” title=”Photo Credit: Codie McLachlan/Getty” info_content=”Photo Credit: Codie McLachlan/Getty” lightbox_caption=”” id=”” class=”aligncenter” style=””][text_output]So, in terms of Game Score, each of the five players that are most commonly being mentioned in the Hart Trophy conversations by NHL pundits have had strong seasons. Connor McDavid, Nikita Kucherov and Nathan MacKinnon all placed in the top-10 in the NHL across all four Game Score measurements I discussed, and have statistical cases that easily should prominently place them in the discussion for the Hart Trophy. While Hall and Kopitar have weaker statistical arguments, each of them are on teams competing for the playoffs and have incredible gaps between themselves and the nest best player on their respective teams, thus creating strong arguments for the “most valuable to his team” portion of the discussion.

In terms of just all situation total Game Score, Nico Hischier is 2^nd on the Devils and 30 points behind Hall with a 51.21, while Dustin Brown is 2^nd on the Kings and 21 points behind Kopitar with a 63.32. To be fair, while Leon Draisaitl accumulated a Game Score of 69.02, significantly higher than Hischier and Brown, he trails McDavid by a whopping 34 points, while Stamkos has posted an impressive 81.89, 16 behind Kucherov, and Mikko Rantanen’s 71.17 trails MacKinnon by 15.

Long story short, if you want to disqualify Connor McDavid from your own personal Hart Trophy ballot because his team was an absolute dumpster fire this year, that is your right. However, he has perhaps the strongest statistical argument of any player for the league’s MVP award, and has the largest gap among any of the contenders between himself and the next best player on his team in terms of Game Score. Personally, if I were voting for the Hart, my top-3 would include McDavid, MacKinnon and Kucherov, and I am still unsure of exactly what order I would place them in. If I absolutely had to decide right this instant, I would probably give the award to Nathan MacKinnon, as I think he is at the heart (no pun intended) of the remarkable turnaround that Colorado has had this season, is clearly the best player on his team, and has among the strongest statistical cases in the league.

Enough of the Hart Trophy talk and back to just purely discussing Game Score. Dom also created another stat based off of Game Score, which he calls Game Score Value Added (GSVA). GSVA is a three-year version of Game Score that is translated to its value in wins. In other words, GSVA is similar to WAR, GAR and wPAR, in that it communicates player value in terms of wins, as opposed to points or any other production metric. One particularly useful application of GSVA is Dom’s use of the model in the pre-season to project individual player performance, and then aggregating these performances by team to project team performance.

On January 10, 2018, Dom published an article on The Athletic that discussed how each team and player had performed to that point in the season, compared to Dom’s pre-season expectations based off of his model. The following infographic shows how the Rangers had performed through the first 41 games of the season. It includes the team-level data at the top, and a table beneath that includes the 2017-2018 average Game Score for each player in the column labeled “2017-2018 Game Score”, the pre-season GSVA projections for each player, their current GSVA through the first 41 games of the season, and the differential between the two. An image beneath the table explains the color-coding of the non-differential data columns: the darkest blue indicates top-tier production, while red represent replacement level performance.

Mika Zibanejad had the highest average Game Score through the first 41 games of the season, Pavel Buchnevich had exceeded his pre-season GSVA projection by the greatest amount, and Kevin Shattenkirk had been the greatest disappointment in terms of GSVA. It should be noted that despite being a disappointment, Shattenkirk still ranked as the second-best defenseman on the team, and graded out as a “top-62” defenseman (i.e. low end top-pairing guy) in the NHL.[/text_output][image type=”thumbnail” float=”none” src=”1893″ alt=”” href=”” title=”” info_content=”” lightbox_caption=”” id=”” class=”aligncenter” style=””][image type=”thumbnail” float=”none” src=”1894″ alt=”” href=”” title=”” info_content=”” lightbox_caption=”” id=”” class=”aligncenter” style=””][custom_headline type=”left” level=”h5″ looks_like=”h5″ accent=”true” id=”” class=”” style=””]How Can I Access Game Score Data On My Own?[/custom_headline][text_output]First and foremost, I definitely recommend following Dom on Twitter and reading his work at The Athletic, which as I mentioned above, frequently includes his Game Score data. However, if you are looking to access Game Score data at any point during the season, you can do so from Corsica. If you want to view single-game Game Score data, then Corsica’s Live Games feature is the tool you need. You can access the Live Games feature from a link within the More dropdown menu in the site’s navigation menu, or from an image near the bottom of the homepage.[/text_output][image type=”thumbnail” float=”none” src=”2405″ alt=”” href=”” title=”” info_content=”” lightbox_caption=”” id=”” class=”aligncenter” style=””][text_output]The Live Games landing page lists the games for the current day at the top; if you are accessing the feature in the morning, the top will display the previous night’s games as well as the slate for the current day. If you are access the feature while games are playing, you can click the game to view the near-real-time data from the game. From this section, you can also input a date in order to view the game data from a specific game from that day.[/text_output][image type=”thumbnail” float=”none” src=”2406″ alt=”” href=”” title=”” info_content=”” lightbox_caption=”” id=”” class=”aligncenter” style=””][text_output]Selecting a game brings you to a screen like the one pictured below, which includes a table at the top that summarizes the team-level key metrics, followed by a chart that depicts how each team performed in a selected metric throughout the game; the dropdown to the left allows users to select what metric to view.  The bottom of the screen provides the summary data for each player on each team, and it includes a GS column, with houses the Game Score for each player. The screenshot below is from the Rangers vs. Capitals game that I discussed in the previous section, and I have sorted each team’s player table by the Game Score columns.[/text_output][image type=”thumbnail” float=”none” src=”2412″ alt=”” href=”” title=”” info_content=”” lightbox_caption=”” id=”” class=”aligncenter” style=””][text_output]To view the season-long Game Score data, go back to the Corsica homepage and either select the Skater Stats section in the middle of the page or select the Skater Stats link within the Skaters dropdown menu at the top. Either link will bring you to the Skater Stats landing page; please note that by default the page will display only 5v5 data and it will limit the player data to only those who have played at least 50 minutes throughout the season.

You can use the filters provided at the top of the screen to set the criteria you want, including the abilities to filter the player data by select season(s), player(s), team, position, game state (e.g. 5v5, any situation etc.), venue, session (e.g. regular season, playoffs etc.) and time on ice minimum. A Report dropdown menu also allows you to toggle between various data screens; for the Game Score data, you want the Summary report, which is displayed by default. The table beneath is sortable by every column, and includes GS and GS/60 columns, which house the Game Score and Game Score per-60 data, respectively. The screenshot below depicts all situations data for the New York Rangers players that have accumulated at least 200 minutes of ice time throughout the season, and it has been sorted by Game Score (largest at the top).[/text_output][image type=”thumbnail” float=”none” src=”2413″ alt=”” href=”” title=”” info_content=”” lightbox_caption=”” id=”” class=”aligncenter” style=””]