Projecting The 2022-23 Eastern Conference Standings Using Goals Above Replacement (GAR)

Photo: @Capitals

In a series of articles last season, we projected the potential standings of each NHL division using Goals Above Replacement (GAR), a metric that encapsulates player value for every game situation and compares that performance to a replacement level player. The advantage of GAR as a means for measuring overall player value is that it gives us a real insight into the holistic performance of a player, unlike than the typical box score statistics like goals, assists, points, and plus-minus.

At the end of the day, goals, assists, and points are important in measuring offensive contributions that results in actual goals scored. On the other hand, defensive value is really important to measure in terms of player value outside of just the overall “eye test”.

In this post, I’ll project what the overall Eastern Conference standings will look like using GAR. I’ll talk a bit about the methodology used for calculating the teams’ overall GAR projections, as well as compare some of the projections we made last season to see just how accurate the model is.

The statistics used in this post are courtesy of Evolving Hockey. If you’d like to learn more about GAR or some of the other statistical terms used in this post, please check out our NHL analytics glossary.

Methodology

For each team’s GAR projection, I used the NHL roster listed for each team on CapFriendly’s team page. Players who have a chance of returning from the long-term injury list were also included, like Nicklas Backstrom, Tom Wilson, and Carl Hagelin. Other players like Carey Price and Brent Seabrook, who are expected to miss the entire season were omitted.

Also, for players who either did not play in the NHL last season or did not have a projection in Evolving Hockey’s data set were given a replacement level score of zero so they wouldn’t overly affect team’s projections. Players like 2022 first overall pick Juraj Slafkovsky were not given a projection, so he was one of the few given a GAR projection of zero. That’s not saying that’s his expected level of performance, but it’s a standardized route for a league wide projection.

For goaltenders, we had previously used their GAR performance from the prior season to project out their upcoming season. This time, we took the average of the past two season’s performances to help even out one-hit-wonder seasons and make some of the projections a bit more accurate.

How accurate were last season’s projections?

In the Metropolitan Division standings projections from last season, here’s how the division standings looked:

  1. New York Islanders (92 GAR)
  2. Pittsburgh Penguins (92 GAR)
  3. New York Rangers (83.1 GAR)
  4. Washington Capitals (79.9 GAR)
  5. Carolina Hurricanes (73.6 GAR)
  6. New Jersey Devils (69.9 GAR)
  7. Philadelphia Flyers (56.4 GAR)
  8. Columbus Blue Jackets (46.7 GAR)

And here’s what the actual division standings panned out to be last season:

  1. Carolina Hurricanes (116 points)
  2. New York Rangers (110 points)
  3. Pittsburgh Penguins (103 points)
  4. Washington Capitals (100 points)
  5. New York Islanders (84 points)
  6. Columbus Blue Jackets (81 points)
  7. New Jersey Devils (63 points)
  8. Philadelphia Flyers (61 points)

Obviously, the big miss was on the Carolina Hurricanes. The projection model did not value Frederik Andersen very highly (-1.3 GAR from the season prior), but in this season’s updated projections using the last two year’s averages, he has 18.6 GAR.

The large differentiator in the projections is the quality of goaltending. Teams that have high-end goaltending, or two very good goaltenders (looking at you, Islanders), tend to have higher GAR projections in this model.

The model from last season did get the top 5 teams correct, but the Islanders had a really disappointing 2021-22 season after being pre-season Stanley Cup contenders after a few deep runs in the playoffs.

For comparison’s sake, let’s look at the Atlantic Division projections from last season compared to actual results:

  1. Toronto Maple Leafs (110.4 GAR)
  2. Boston Bruins (98.2 GAR)
  3. Tampa Bay Lightning (74.5 GAR)
  4. Florida Panthers (70.6 GAR)
  5. Detroit Red Wings (67.7 GAR)
  6. Montreal Canadiens (67.7 GAR)
  7. Ottawa Senators (48.2 GAR)
  8. Buffalo Sabres (26.7 GAR)

And the actual standings:

  1. Florida Panthers (122 points)
  2. Toronto Maple Leafs (115 points)
  3. Tampa Bay Lightning (110 points)
  4. Boston Bruins (107 points)
  5. Buffalo Sabres (75 points)
  6. Detroit Red Wings (74 points)
  7. Ottawa Senators (73 points)
  8. Montreal Canadiens (55 points)

The Florida Panthers were the large aberration here, similar to the Carolina Hurricanes in the Metro. They made a big jump in the standings due to a very solid goaltending performance by Sergei Bobrovsky, as well as loading up at the trade deadline by acquiring Claude Giroux and Ben Chiarot. The Bruins had a step back due to some center depth issues, but still had solid goaltending and defensive play.

2022-23 GAR Standings Projections

Below are the GAR standings projections for both the Atlantic and Metropolitan Divisions for the 2022-23 season:

The Bruins take the top spot in the projections in the Atlantic Division and the Eastern Conference, largely due to a lot of top-end talent in Charlie McAvoy (12.9 GAR), David Pastrnak (12.2 GAR), Brad Marchand (10.2 GAR), and Patrice Bergeron (8.9 GAR). On top of that, two of their top five GAR projection values are their goaltenders Jeremy Swayman (16.3 GAR) and Linus Ullmark (9.2 GAR).

The most interesting piece about the projections were how low the Toronto Maple Leafs came in. While they do have some immaculate star power up front in Auston Matthews (20.8 GAR), Mitch Marner (14.1 GAR), William Nylander (9 GAR), and John Tavares (6.2 GAR), their goaltending duo definitely weighs their total value down. Former Capital Ilya Samsonov (-0.85 GAR) and Matt Murray (2.25 GAR) are among the lowest projections for goaltenders for a playoff contending team.

For the most part, the Metropolitan Division projections seem like they are rather realistic, outside of the New York Islanders potentially being buoyed a bit higher in the standings due to their extremely solid goaltending tandem. Semyon Varlamov (16.3 GAR) and Ilya Sorokin (23.9 GAR) were the highest two GAR values on the Islanders. Mathew Barzal was their highest projected skater at 7.8 GAR. The Islanders are rather deep in defensive GAR value, which also buoys them a bit.

It stands to reason that the Pittsburgh Penguins replace the Islanders in the playoffs. Their core is getting older, but until they actually miss the playoffs, it’s hard to count them out of the race.

For the Capitals, they benefit from adding one of the league’s top goaltenders in Darcy Kuemper (15.4 GAR). The top five skaters in terms of GAR value were Alex Ovechkin (9.7 GAR), Dylan Strome (7.2 GAR), Anthony Mantha (6.6 GAR), Dmitry Orlov (6.6 GAR), and John Carlson (6.2 GAR). The projections were rather low on Evgeny Kuznetsov (3.3 GAR) after a 6.4 GAR performance last season.

Potential Playoff Bracket

If these projections were to hold up, here’s what the Eastern Conference playoff bracket would look like:

These are some intriguing matchups. You have the Battle of the Sunshine State between the Panthers and the Lightning, the hockey edition of the Subway Series between the Rangers and Islanders, and a rematch of the 2021 playoffs between the Bruins and the Capitals.

Conclusion

Overall, GAR projections are just predictions based on prior performances and age regression. There is room for variance just because no one model can predict immense breakout seasons, as well as in-season trades and other means of acquiring or adding players to the NHL roster. Rookies are also a huge wild card that are hard to predict.

These predictions do set a baseline for which teams we can expect to be in a position to battle for the playoffs.

By Justin Trudel

About Justin Trudel

Justin is a lifelong Caps fan, with some of his first memories of the sport watching the team in the USAir Arena and the 1998 Stanley Cup appearance. Now a resident of St. Augustine, FL, Justin watches the Caps from afar. Justin graduated with a Bachelor's of Science in Political Science from Towson University, and a Master's of Science in Applied Information Technology from Towson University. Justin is currently a product manager. Justin enjoys geeking out over advanced analytics, roster construction, and cap management.
This entry was posted in News and tagged , , . Bookmark the permalink.

4 Responses to Projecting The 2022-23 Eastern Conference Standings Using Goals Above Replacement (GAR)

  1. Diane Doyle says:

    Very interesting piece.

    How does the model take into account players who are out for a good portion (2-3 months) of the season but not the whole season? While the Bruins have high-end talent, one factor adversely affecting them this season is the fact they have many key players out for long stretches of time. Marchand, McAvoy, Mike Reilly, and Matt Grzelcyk are all projected to miss the first two months of the season.

    Like the Caps, Boston is also a team with relatively old key players (Marchand, Bergeron, and Krejci who is coming back).

  2. Jon Sorensen says:

    The Bruins projection stands out to me. I have them falling back this season. Is there a component of GAR for regresssion?

  3. redLitYogi says:

    Anyone surprised that the projections and the reality were so far apart? No amount of data crunching can account for fluctuations in emotion and psychology, for how the rhythms of individual performance (the cycles of highs and lows) will sync or not sync or for how well particular players will play with each other or any number of other variables in what is a literally immeasurably complex dynamic system. Still the piece succeeds: it may unintentionally mislead but it does entertain and I enjoyed it and the author’s acknowledgement of its failings. Alas, more and more, organizations in the real world attempt to use similarly clumsy metrics to assess the value of workers in their enterprises, and here they do grave harm to those workers and to those organizations. One would hope workers will find a way to push back because, applied this way, such clumsy evaluations create needless stress and — as this article illustrates — misunderstanding.

Leave a Reply

Your email address will not be published.