During last night's Islanders-Rangers game, there was general gnashing of teeth, shaking of fists, and wailing of women in the game thread at Coach Capuano's decision to ice the fourth line after the Rangers had pulled their goalie and were seeking a late equalizer. The anger only intensified as the line played a long shift in which they struggled to clear the zone. (Although, admittedly, the story of that line's struggles have grown in the telling, as Extra Skater shows that Cizikas, McDonald, and Martin only gave up 5, 5, and 4 Corsi 5v5 events, respectively, for the entire game, a far cry from the 8 chances some commentators were suggesting the line had surrendered on that shift alone.)
Although the line -- and the Isles -- did not concede a goal, Capuano's decision to ice the fourth line in that situation inspired renewed claims that he should be axed.
In the comments after the game, I suggested that Cappy's decision to send out the fourth line, while certainly not his only option, was not unreasonable under the circumstances. In case you missed it, I noted that over the month of January, generally corresponding with the Islanders' return to successful hockey, the fourth line had only conceded 2 EV goals.
That compared favorably, I noted, with the 10 EV goals conceded by the first line, the 6 EV goals conceded by the second line, and the 5 EV goals conceded by the third line over the same period. Indeed, I noted that the Islanders had allowed 18 goals since the fourth line last conceded a goal, in the first period against Dallas on January 12. Accordingly, I concluded Cappy's decision to continue to rely on a line that had yielded consistent good results of late was a reasonable decision, and certainly not a firable offense.
To be brief, some disagreed.
Anyone who wishes to review (or re-live?) the debate may do so in the appropriate threads; that's not the point of this post. Rather, a variety of other metrics were mooted by various commentators (and not always in the friendliest terms, it must be noted) as superior alternatives to the outcome-orientated data I had presented. Since these commentators declined to actually post the results of their possession-metric based suggestions, however, I have endeavored to do so here. At the expense of spoiling your anticipation, dear reader, I believe that these data suggest Coach Capuano's decision last night was defensible.
The brunt of the criticism was that goals against are essentially random; thus, a better criterion of which forward line to play while protecting a lead in the final seconds of a game would look at the on-ice chances allowed by the forwards, regardless of whether these chances were ultimately successful by the opposing team.
The first four charts that follow present this season's data for each of the 12 forwards in yesterday's game, as taken from ExtraSkater. (Unfortunately, I could not figure out how to separate out Vanek's Islanders statistics from his Sabres statistics, so his data includes his data for the entire season thus far.) They all present 5 v 5 play, and are sorted respectively by Corsi Against/60, Fenwick Against/60, Shots Against/60, and Goals Against/60. As goals are a subset of shots, which are a subset of Fenwick, which are a subset of Corsi, we are essentially drilling down:
Table 1. Corsi Against/60 (5 v 5)
Table 2. Fenwick Against/60 (5 v 5)
As one might well expect, the Corsi and Fenwick rankings are essentially identical. Nelson gives up the fewest Corsi/Fenwick events, followed by Bailey, Regin, and Grabner. In both tables, the first line is rooted soundly at the bottom of the rankings, and this is particularly the case in the Fenwick table where there is a difference of 4.5 Fenwick events between Cizikas in 9th place and Tavares in 10th. The significant breaks in the Corsi table occur between McDonald (54.76) and Clutterbuck (57.41), in 6th and 7th place, respectively, and between Martin (59.37) and Tavares (61.80) in 9th and 10th.
Fenwick, of course, is simply Corsi with blocked shots removed. Comparison between the tables shows that Clutterbuck, Martin, and Cizikas are blocking several more shots than the other skaters, but not significantly more to change the overall rankings, which are mostly identical between the two charts.
Broadly speaking, the differential between Corsi/Fenwick For and Corsi/Fenwick Against can suggest whether a team is driving possession and, thus, creating more scoring opportunities than the opposition. But for the kind of end-of-game situation in which Coach Capuano found himself last night, shot attempts that are blocked or miss the net are not particularly salient; the more important question involves which players are allowing shots on goal in the first place. Shots on goal allowed per 90 addresses this issue while controlling for on-ice save percentage (as vociferously requested by AP77, amongst others.). Indeed, Shots/60 simply means "goals allowed per 60 when save percentage is kept constant." Notably, these rankings diverge substantially from the Corsi/Fenwick data:
Table 3. Shots Against/60 (5 v 5)
When it comes to which players force the goalie to make the fewest saves, irrespective of save percentage, number one is every LHHer's favorite Islander, Josh Bailey. (This must have been the reason AP77 was pushing so hard for this stat, as we all know how much he adores JB Smoove.) But look who's right there at numbers 2 and 3: McDonald and Martin, the fourth line wings.
About the only thing constant between this table and the Corsi/Fenwick rankings is that Tavares, Okposo, and Vanek continue to be dead last, with Tavares in 10th place giving up approximately 3 more shots on goal per 60 than Nielsen in 9th. In fact, although Cizikas is only 7th in these rankings, the difference between Bailey in first place and Cizikas in 7th in less than one shot per 60. About an equivalent difference of about one shot in 60 separates Cizikas in 7th from Nielsen in 9th. (Indeed, especially because these data do not distinguish between quality of competition, it's probably safe to assume that any of the non-first line skaters is about equally equivalent at preventing opposing shots on goal.) The difference between Bailey in 1st and Nielsen in 9th is well less than the difference between Nielsen and Tavares in 10th.
There continues to be a passionate (and quite unresolved) debate about whether the play of skaters -- and, particularly, forwards -- can consistently influence shot quality. If you've read my posts, you know where I stand on this matter: I think back-checking is a hockey skill, and lines that can back-check successfully help their team win.
Unfortunately, the effectiveness of one's back-check is notoriously difficult to measure with the individual player data currently collected because, moreso than any other hockey skill, back-checking (and team defense, generally) is a team skill that involves all five skaters and the goalie working together. I do believe we'll eventually develop some passable metric to measure a skater's ability to influence shot quality, however. (One place to start may be the length of time an opposing player is allowed between receiving and playing the puck, for example.) In any event, Table 4 presents the players who, by luck or skill or some combination thereof, have allowed the fewest 5 v 5 goals per 60 minutes:
Table 4. Goals Against/60 (5 v 5)
Notably, Martin -- who is first on this list -- was also third in the shots rankings, suggesting he is one of the team's better defensive players whether you believe a forward can influence the quality of shots or only the quantity of shots. Clutterbuck and Cizkas round out the top 3, and there is a fairly substantial gulf between these three players and Regin in 4th place. I and several other LHHers have argued that Clutterbuck should play on a defensive line with Cizikas and Martin, and this chart certainly bears that out.
McDonald, despite giving up the second fewest shots, is on the ice for a ludicrous 3.3 EV goals against/60. For those of you who only care about quantity of shots allowed, I suppose McDonald is your man. In contrast, if you think a player can affect shot-quality, you don't want Colin McDonald anywhere near the ice in a close game. (I suppose we can now refer to this debate as the McDonald Conundrum, which sounds like a lost Robert Ludlum book.)
Nielson and Nelson -- whom many LHHers would have preferred on the ice in place of the fourth line last night -- each concede more than 3 EV goals per 60, and sit in 8th and 11th place, respectively. As the fifth and final chart shows, opposing players manage a significantly better shooting percentage against these players than the other Islanders -- although only Martin, Clutterbuck, Cizikas, and Vanek hold opposing players below the mean (or, if you prefer, benefit from a favorable save percentage).
Table 5. Opposing Shot Percentage (5 v 5)
In conclusion, even the FancyStats(tm) suggest that Coach Capuano's decision to play the fourth line in last night's high leverage situation was reasonable. It was not the only reasonable option available to him, surely, but nor was it unreasonable. None of the nine forwards on lines 2 through 4 have particularly distinguished themselves as markedly superior in terms of Corsi and Fenwick events allowed, and the first line has distinguished itself merely in allowing many more such chances.
In terms of actual shots allowed, Martin and McDonald have been two of the three best Islanders in terms of allowing the fewest shots, and Martin and Cizikas have been enjoyed particular success (whether due to luck, skill, or a combination of the two) in not conceding goals -- especially recently. That the Islanders succeeded in not allowing the Rangers to score -- and, in fact, winning the big game -- is also not a fact I believe should be overlooked, as results do (after all) matter.