Talking about the Rangers, unless I decide to talk about something else

Archive for the ‘LGR’ Category

The Rangers vs. Regression

“I’m from New York/Concrete jungle where blogs are made of/There’s nothing you can’t do/Now you’re in New York” – Alicia Keys

Despite some serious inconsistency in the team’s performances, the Rangers started the year off 10-2-2. That’s really good! Henrik Lundqvist has been the best goalie in the league so far! Guys like Oscar Lindberg and JT Miller are wildly outperforming expectations! Sure, guys like Nash (who’s been good otherwise) and Kreider (who’s been real bad) haven’t produced, but the overall lineup has averaged 3 goals a game! These are all good things, yes?

batman well actually

Oh no. No, no, no. Regression awaits. Fact. Death, taxes, and regression to the mean.

Lots of sites writers I enjoy have written pieces (like here and here) about the Rangers unflattering advanced statistics – 46.5% Corsi at 5 on 5 as it currently stands, 28th in the league – as if the team’s record is merely a house of cards waiting to be blown over by the wind from a flap of a hummingbird’s wing. The stat has been tracked long enough to show that a high Corsi (meaning here for the uninitiated) is usually a reliable predictor for team wins and success. Every year there’s a notable exception where a team excels and makes the playoffs despite having poor possession numbers – last year was Calgary, 2 years ago was Colorado, 3 years ago was Toronto, etc. Each of those teams came out of nowhere to succeed for a season defying the numbers then had (barring a major turnaround for Calgary this year) epic collapses the next year.

3 things:

  • First: my major problem with Corsi is that all shots are counted equally. That is categorically insane. An Ovechkin one-timer from the left side of the zone counts the exact same as a player weakly flubbing a backhander at the goal in an attempt to get a whistle. Not exactly iron-clad!
  • Second: the teams used as examples of Corsi regression victims all had one year of success out of nowhere then crashed back to Earth, and the Rangers do NOT fit that bill. The last 4 seasons have ended as such: Conference Finals, 2nd Round, Cup Finals, Conference Finals with half the defense missing at least one leg. One of these teams is not like the others, in a good way.
  • Third: Corsi is measured over all 60 minutes (or at least all 5 on 5 minutes) of a game regardless of the score, which also presents issues. The Rangers have gotten up by a few goals on teams early in games a couple time this year, and that drastically changes the way the game gets played. Consciously or not, all teams tend to play a drastically more conservative game; that requires opponents to play more aggressively, which consequently leaves them more vulnerable to counter-attacks off giveaways and mistakes. By nature a team that sits back that way is going to have an unflattering shot differential because they’re begging the other team to push the play. 2 of the Rangers worst Corsi games of the year were last week against Washington (went up 4-1 less than 7 minutes into the 2nd, – 29 Corsi differential) and Saturday against Arizona (up 3-0 just over 5 minutes into the 2nd, -16 Corsi differential) reflect that.

Since it’s ridiculous to poke holes in an argument without making your own proposal, here goes mine:

Step 1 is to view all advanced stats using only Close Score numbers. Close Score numbers reflect the statistics recorded during 5 on 5 play when the game score is tied, and when either team has a 1 goal lead in the 1st or 2nd period. This is when teams are both attempting to drive play and, in my opinion, give a better reflection of a team’s true effectiveness.

Step 2 is to take into account both the possession numbers like Corsi and High-Danger Scoring chances. This helps negate the Corsi issue in which the Michael Del Zotto Memorial 70-MPH Wrist Shot From the Point With No Traffic In Front (trademarked 2010) gets counted the same as Stamkos taking a clear path one-timer from the slot, as it provides context for how dangerous the shots taken were.

When viewed through this prism, the numbers for the Rangers this year have been just fine. Corsi at close scores is 50.6%. High-danger scoring chances at close scores are even better at 54.2%. Their overall possession stats are still a cause for concern, although that’s more because it points to how poorly the team has been in the defensive zone. We have years of evidence that show we should expect improvement on that front, but it should inspire some fear until we actually see that improvement. On the bright side, that fear is always tempered somewhat when you look in your net and Henrik Lundqvist is standing there. Advanced stats, as anyone who subscribes to them will tell you, are meant to complement what you see on the ice to give you a better understanding of performance.

The team’s play has been imperfect. Dropping into TOO much of a shell when you take a big lead isn’t the smartest move. I still worry about the play of Girardi and Boyle in the D-zone, Staal has been up and down, and McDonagh’s skating still isn’t at the level we’re used to. A couple forwards (KREIDER) have been straight up lost out there. Soon, team shooting percentage will drop by a few percentage points and Hank/Raanta’s combined save percentage will drop to something semi-human. PDO doesn’t stay at 106+ for 82 or more games.

When you build your team from the net out, as the Rangers have, sometimes you can get through these periods of at-times-uninspired play. It won’t be a formula for success come playoff-time, but the next 68 games are here for the team to work those kinks out. At this point, though, it seems to me like people are mistaking smoke for fire.

Carolina tonight. Since I wrote all this the Rangers will lose by 8. #LGR.

Tag Cloud