So, about those goalies: Defense matters edition

Hockey fans love narratives and writers love creating them. Narratives give the human mind explanations, explanations we then generalize and apply to other situations. A favorite narrative among fans and writers alike is "The Savior," in which a single player comes onto a team in the middle of a season of struggle and by his very presence turns everything around. We've heard this one a lot around here over the past year, and it goes like this:

The Tampa Bay Lightning struggled throughout the first part of the 2010-11 season, with shoddy defense and worse goaltending. On New Years' Day, the Jedi Master (Lightning General Manager Steve Yzerman) took a risk on an aging samurai goaltender, whose heroic performance over the next three months saved the season and led the team to within one goal of the Stanley Cup final.

This story's almost always followed by, "Yzerman should do that again."

The problem is that, while that's a great story, reality is a lot more complex than the narrative allows for. This is still a team game, and it takes more than just good goaltending to make the playoffs. The case for "The Goalie Savior" may be thinner than some want to believe. For one thing, it is nearly impossible, especially given the measurements we currently have, to separate individual goaltender performance from team defense. We shouldn't forget that while Yzerman did bring Dwayne Roloson in on January 1, he also brought in Eric Brewer on February 18. And that trade mattered, too.

For the sake of trying to understand what lessons we really ought to be learning from the Roloson trade, it's critical that we try to determine what difference, if any, the Brewer trade made in the team's fortunes. To do that, I divided the season into four sections, related to the dates of the trades. Roloson's first game was game 40 (January 4). Dan Ellis's last game was game 59 (Feb. 19), which was also Eric Brewer's first game.

[These are all 5-on-5 numbers.]

To Dec 31

Jan 1 -Feb 18

Feb 19 - Apr 9

Jan 1 - Apr 9

Record (Wins-Losses):





Win %:










Pt %















ES SA/60:





ES SF/60:





No of 5+ GA games:





What does all of this show? Well for one thing, the team actually performed slightly worse in more than one area after the first of the year than they did before. They still took more points than they left on the table, but that percentage dropped. Between the time Roloson started in Tampa Bay and the time Ellis left and Brewer came on board, goaltending numbers got significantly better, but puck possession numbers started to get worse. There were still plenty of "howlers" defensively, with 7 games (out of 19) in which more than 5 goals against were given up.

However, while the Lightning possessed and shot the puck less than before, team defense was getting better over the second half of the season (January 1 to April 9). Moreover, those numbers got even better after Ellis departed and Brewer joined the team. Save percentage jumped another 0.033, GAA dropped by more than half a goal per 60 minutes at even strength, and only once did the team give up more than 5 goals. The important question, though, is: How much of that change can be attributed to goaltending and how much to upgraded defense? Let's take a look at Brewer's numbers.

Brewer's defensive stats for the year include his time in St Louis. His CorsiRelQoC was a team-high 0.806, meaning he faced the toughest competition, while his personal CorsiRel was -4.9. He started in the defensive zone 56% of the time and his teams had a .917 save percentage while he was on the ice. In other words, he had tough assignments and did fairly well at them, but not extraordinarily well. So, while he was important, his presence alone can't explain the drop in goals against or the increase in save percentage.

What about Roloson? Did he "save the season?" Not by himself, but his performance, too, was critical. If the goaltending had continued at the 0.888 rate that Ellis and Smith were putting up (while puck possession continued to deteriorate at the same rate), the team would have given up some 34 more even-strength goals than they actually did over the course of the season. That change alone would have given them a -27 goal differential on the season, good for the 10th in the conference (if everything else were held the same, which of course it couldn't be, since some of those goals would have to have been given up to teams within the conference.)

However, figuring Roloson & Ellis's 0.910 save percentage in the same situations--changing the save percentage but assuming the same shots against and goals for--we can estimate that the team would have given up 13 more even-strength goals than they actually did (that is, than with both Roloson and Brewer on board.) This would leave the Bolts with a -6 goal differential on the season, if they had relied on the change Roloson alone brought to the team.

Save Percentage

Est. Goals Against

Goal Differential

Ellis & Smith




Roloson & Ellis

0.910 (after Jan 1)



Roloson & Smith + Brewer

0.923 (after Feb 19) (actual)

74 (actual)

+7 (actual)

In reality, the Lightning were tied for the lowest goal differential among playoff teams in the Eastern Conference. Both the Montreal Canadiens and the Lightning had a +7 goal differential. No team made the playoffs giving up more goals than they scored. The Carolina Hurricanes were closest with a -3, and they sat in 9th place at the end of the regular season. Those 13 goals could have made the difference between getting into the playoffs and having a long summer. This doesn't even consider the difference these trades made on the power play or penalty kill.

Of course it's impossible to know what would have happened had the team not gotten both better goaltending and better defense, but with the sinking shot metrics the Lightning put up over the second half of the season, both of those things were critical to the team's success. And Roloson's contribution, while necessary, was likely not sufficient to propel the team into the playoffs. Brewer's contribution was apparently just as necessary. Defense matters.