Using Danger Zones to End the Reimer-Bernier Debate
October in Toronto always features a recurrent ritual drama; sadly, I’m not talking about the start of a new hockey season.
The surest sign of summer’s death is the annual performance of Reimer versus Bernier. We all know how and when it ends already (poorly, in early April), but we still manage to get caught up in its unlikely plot twists, heroic acts, and tragic defeats.
Fans will take sides, and vehemently argue for their chosen goaltender’s superior ability and quality, as surely as the seasons turn.
To stop this eternal cycle, I dug as deeply as I can to find the most objective answer to that perpetual question: is James Reimer better then Jonathan Bernier, or vice versa?
The data is clear. Let me tell you who the numbers favour.
The Traditional Case
The oft-cited official NHL career stats for Bernier and Reimer place them on a similar footing. The equal number of games played makes the comparison especially symmetrical:
It’s not hard to see why Leafs supporters are divided on the duo. Reimer has a fractionally better win-loss percentage (even discounting overtime losses), and more shutouts. Bernier has a marginally greater save percentage, and a significantly better goals-against average.
In total, Bernier looks stronger, but Reimer apologists are quick to point out that Bernier’s numbers are a reflection of his time behind the Stanley Cup-winning Los Angeles Kings’ defense. They also argue that Reimer has had to face more (and more difficult) shots, making his statistics deceptively less impressive. Finally, they argue that he’s a much nicer guy.
I can’t speak to the last of these objections, but I can address the first two using more advanced statistics than the NHL provides.
Applying the Danger Zones: Parallel Goaltender Comparison
It’s possible to compare Reimer and Bernier directly using a danger zone method I developed this summer. An example of what I now call the Goalie On Another Team (GOAT) variety of this method (featuring Bernier) is available here. For our present purposes, the directly comparative version of the method (now known as Parallel Goaltender Comparison) is most useful.
To quickly summarize, War on Ice provides a breakdown of a goalie’s shots against into danger zones. Shots from far out are low danger, shots from close in are high danger, and shots in between are medium danger. This makes it possible to use a given goalie’s save percentages in each of these zones to determine how he might have done facing the same number of low, medium, and high-danger shots as another goalie. By accounting for shot volume and distance, such a parallel comparison helps to even the playing field when evaluating two goaltenders.
The argument that Reimer has faced more (and more difficult) shots over his career is true, but not powerfully so: Reimer has faced an average of 33.54 shots per 60 minutes at even strength, compared to Bernier’s 30.35. Similarly, Reimer’s high-danger shots account for 28.5% of his total shots faced, while Bernier’s account for 27.7%.
Even accounting for the volume and danger-zone differences, however, Bernier maintains his advantage. Using even-strength data only to minimize the team effects special-teams systems create, this is how our heroes would look if they faced each other’s career shots:
|Additional Goals||SV%||SV% Change||GAA||GAA Change|
|Bernier as Reimer||-4||92.35||0.1||2.57||-0.03|
|Reimer as Bernier||6||92.25||-0.15||2.35||0.05|
The table tells us that, over the course of their careers, there is some small difference between the two goaltenders. If Bernier had faced Reimer’s career shots, he would have a slightly higher overall save percentage than Reimer, and a slightly lower goals-against average. Reimer would see his save percentage dip somewhat lower than Bernier’s, and see his goals-against average rise. Bernier comes out ahead, but by even less than his traditional stats showed.
As Sean Tierney demonstrated in a recent article, despite Bernier’s relatively poor 2014-15 season, he still maintained a better even-strength save percentage than Reimer in every danger zone. A Parallel Goaltender Comparison using the two seasons Reimer and Bernier went head-to-head is even more revealing:
|Additional Goals||SV%||SV% Change||GAA||GAA Change|
|Bernier as Reimer||-15||92.64||0.88||2.60||-0.31|
|Reimer as Bernier||24||91.72||-0.90||2.72||0.29|
Bernier would have allowed 15 fewer goals than Reimer over the last two seasons if he had faced his shots. His save percentage would exceed Reimer’s significantly, and his goals-against average would be much lower. If Reimer had faced all Bernier’s shots, his save percentage would be much lower than Bernier’s, and his goals-against average much higher. Since Bernier’s arrival in Toronto, he has significantly outperformed Reimer.
Interestingly, Reimer still faces more shots against than Bernier, even considering only the seasons they have played behind the same defense. His 35.38 average shots against per 60 minutes is clearly greater than Bernier’s 32.79, leading me to speculate that the goaltenders may be responsible for at least part of that discrepancy: rebound control, or its lack, will affect shots against. Despite the volume difference, both goalies saw about the same percentage of their total shots against coming from the high-danger zone: 28.3% for Reimer, and 28.2% for Bernier. Nothing here indicates that Reimer faces more difficult shots.
Reimer and Bernier’s traditional career statistics, advanced career statistics, and advanced head-to-head-seasons statistics all point to the same number: 45. While Bernier’s career numbers put him only slightly ahead of Reimer, his superior play behind the same team makes his victory clear. No measure favours poor James.
So that’s that. I expect this article to end the debate once and for all.