Then, just months ago, a wins above replacement page popped up on war-on-ice. After an 11-part series on the war-on-ice blog and the personal website of war-on-ice creator A.C. Thomas explaining the rationale behind their brand new statistic, they had found WAR* for hockey - and they called it GAR, or Goals Above Replacement**. Of course, there have been other catch-all stats available to the public - Goals Versus Threshold**, or GVT, is a good example - but GAR is the first catch-all stat where I have found proof that the stat works.
Where? In money puck's article. In the article, he finds that team GAR has a moderate to high correlation with both goal differential and points and, in another article, found that 16 of 20 Conference Finalists over the past six years have been in the top quartile in team GAR.
So, now that I know that GAR works, I can go on to the 'predictions' part of the title. In money puck's article about GAR and success, he predicts player GAR using a regression model (shown in the article). Then, he adds it up and puts the sum of the team's predicted GAR and the sum of the team's three-year average GAR through another regression model to predict how many points a team will get.
So how will the Bruins do?
Accounting for roster differences (Dougie Hamilton is gone!), this year's Bruins team should look like this:
Name | Three-Year Average GAR | Predicted GAR |
Matt Beleskey | 5.72 | 3.60 |
Patrice Bergeron | 18.24 | 9.22 |
Brett Connolly | 1.82 | 1.85 |
Loui Eriksson | 8.02 | 4.63 |
Jimmy Hayes | 2.41 | 2.11 |
Chris Kelly | 2.93 | 2.35 |
David Krejci | 4.80 | 3.19 |
Brad Marchand | 14.83 | 7.68 |
David Pastrnak | 9.92 | 5.48 |
Zac Rinaldo | -4.31 | -0.90 |
Ryan Spooner | 0.45 | 1.23 |
Max Talbot | -4.14 | -0.83 |
Name | Three-Year Average GAR | Predicted GAR |
Zdeno Chara | 7.20 | 4.26 |
Matt Irwin | 2.11 | 1.98 |
Torey Krug | 6.20 | 3.81 |
Adam McQuaid | -0.14 | 0.97 |
Kevan Miller*** | 1.80 | 1.84 |
Dennis Seidenberg | -7.95 | -2.54 |
Zach Trotman | 0.66 | 1.33 |
Name | Three-Year Average GAR | Predicted GAR |
Tuukka Rask | 16.33 | 8.36 |
But using predicted GAR (again excluding Kevan Miller in the calculation), the Bruins are projected to have a much less surprising - and much more pedestrian - projection of 94 points, otherwise known as justthisclose to a playoff berth.
A projection of 94 points isn't good since 95 points is traditionally the benchmark a team must surpass in order to make the playoffs, and that benchmark could be rising.
Last season the Bruins became the 4th team since the lockout to miss the playoffs with 95 or more points and became the only team since the lockout to miss the playoffs with 96 or more points. They were joined by the Los Angeles Kings in the Over-95-Points-But-Out-of-the-Playoffs Club (O95PBOPC for short).
The projection of 94 points, I think, could even be an optimistic projection given our defense is shredded. But even 94 points probably isn't enough.
* WAR, or Wins Above Replacement, was originally a baseball statistic (or "sabermetric") that is an all-in-one metric that captures precisely how much a player is contributing to his team. This stat has often been used as a metaphor for a catch-all statistic: WAR for hockey.
** Replacement or replacement level just means how well your typical injury call-up would perform. A replacement level player will have a GAR of 0.00. The "threshold" in goals versus threshold is just another word for replacement level.
*** Kevan Miller is currently on injured reserve and may not be available for the entire season.