Goals Above Replacement and the 2017 Stanley Cup Playoff Teams

Published in

How to Win

7 min readApr 13, 2017

Alex Ovechkin takes the ice for the Washington Capitals in the Stanley Cup Playoffs (Courtesy: Flickr)

During the season, most people involved in hockey tend to focus their attention and game-watching on specific teams, whether that be a fan’s hometown team or the club that various personnel (team officials, reporters, scouts, etc.) are tasked with observing. Once the playoffs roll around, however, that focus expands to the remaining 16 teams. Regardless of your relationship to hockey, this is likely the time of year when you begin to wonder what some of the other teams in the league look like. Personally, I am excited to watch the Ducks and Flames, two teams whom I haven’t had the opportunity to see much of this season.

It can be daunting to learn about so many unfamiliar teams at once, especially when the action is fast and furious with multiple first round games being played each night. You may be conscious of the big names but it can be difficult to evaluate the ability of the more obscure players in the lineup. One way to do this is with a single, all-encompassing metric. Baseball fans are familiar with a statistic called WAR (Wins Above Replacement), which attempts to measure all contributions made by a player in terms of the number of wins he adds to his team over a replacement level player at his position (think about the guy who your team could sign right now for the league minimum).

Dawson Sprigings, who is more famously known as @DTMAboutHeart on Twitter, created a WAR statistic for the NHL which he released at the beginning of the season and has updated throughout the year. You can find the first page of his multi-part explanation here and I highly suggest that you check it out. He does a pretty fantastic job of explaining his sources of inspiration, how exactly the model works and some of the limitations it faces. With that said, I will attempt to provide a simplified explanation for quicker consumption.

Sprigings has recently modified his model to be measured in Goals Above Replacement instead of wins. I have not seen any statements from him on the reasoning behind this but I personally feel that it is more intuitive to use goals as the barometer for success in hockey. Each player’s GAR is made up of six components:

EVO — Even Strength Offense

EVD — Even Strength Defense

PPO — Power Play Offense

DRAW — Penalties Drawn

TAKE — Penalties Taken

FAC — Faceoffs

Even Strength and Power Play Offense are each made up of two smaller parts, XPM and BPM. XPM is focused on a player’s impact on shot differentials with a huge list of variables accounting for teammates, competition, coaching, zone, score and more. BPM, meanwhile, is focused a player’s impact on the box score, primarily with points scored. This is an attempt to both reward players for scoring but also highlight those who generate offense even when it doesn’t directly result in a goal.

Even Strength Defense is based purely on defensive XPM. Current box score measures for defense (plus/minus, hits, giveaways and takeaways) are not necessarily indicative of strong defensive play and so defensive BPM is not used. It is also important to note one of the major limitations of this metric: the lack of Penalty Kill Defense. Upon review of his testing, Sprigings determined that the currently available data for shorthanded defense was neither consistent nor impactful enough to include. That is not to say that shorthanded play isn’t important; rather, it’s simply not something that we can accurately measure at this point in time.

Finally, penalties and faceoffs are included. One of my favorite parts of this model is how it includes these aspects and weights them in terms of goals compared to offense and defense. On the whole, the importance of faceoffs tends to be overstated in public discussion while the ability to draw penalties and refrain from taking them is understated.

Since the creation of Goals Above Replacement, there hasn’t been much criticism or discussion of the metric, save for a little blip in the Twittersphere a few weeks ago. I find this unfortunate because GAR is both a fantastic measurement of a player’s contributions and a deeply flawed one. It completely ignores the penalty kill and there are valid concerns about the way that each component is weighed (i.e. the value of point production compared to shot differential). However, the GAR results that have been made publicly available do a great job of both confirming common wisdom (Connor McDavid and Sidney Crosby top the list of forwards in GAR this season) and providing surprising results (Oscar Klefbom and Ryan Ellis leading the way for defensemen).

I have spent a lot of time looking at GAR this season and, while I think that improvements should and will be made to the formula, I believe that it is the best currently available way to evaluate overall player performance. Even for the best scouts and team officials in the business who have an incredible eye for the game, the value of this metric is obvious when considering the sheer number of hours required to feel confident in the visual evaluation of just a single player. Looking at the GAR components for each player on a team can be a great starting point to be later supplemented with video or in-person review.

It is for that reason that I have compiled GAR charts for each of the 16 teams in the 2017 Stanley Cup Playoffs. I also used the work done by another great member of the community, @EvolvingWild, who calculated the average time on ice for all players involved in the sample, categorized by situation (even strength, power play and total). The charts that I have made display the goals above replacement that a player provides when given a league average amount of even strength, power play and total ice time, adjusted for position and situation. Finally, I have included benchmark lines in order to help visualize the production expected of certain roles. The forward lines are groups of 90 (30 teams multiplied by three players per line), indicating that a player above the 1L line is among the top 90 forwards in the league while a player between the 2L and 3L lines is among the 181st-270th best. Defense pairings are in groups of 60.

Hopefully this explanation is helpful. If you have any further questions I encourage you to reach out to me on Twitter (@AndrewDellapina) or in the comments below. Without further ado, the Goals Above Replacement charts for every team participating in the 2017 Stanley Cup Playoffs:

Chicago Blackhawks