BEER LEAGUE HEROES
  • Home
  • NHL Draft Talk
  • Hockey Analytics with Walt!

Have the Oilers Improved Under Todd Nelson?

2/19/2015

2 Comments

 
The analytics suggest mostly no.
I'm as happy as any fan that the Oilers are winning more, but let's not give Nelson too much credit. Correlation is not causation. Since Todd Nelson has been coaching, the Oilers winning record has improved, but that doesn't mean they are winning because of Nelson. There may be another ("3rd") variable involved in causing the improved win/loss record, which I explain below.

First Reason for Losing: Poor Save%
What are some the reasons the Oilers were losing before Nelson and during Dallas Eakins time? The main problem was horrible goaltending. And before the reader blames the poor goals against on defense, this season's defensive stats--Corsi-Against and scoring chances against--do not show that the Oilers were allowing more shot attempts or scoring chances than previous seasons. In fact, under Eakins, the Oilers were better in these metrics. Thus, the terrible goals against can be squarely blamed on the goaltenders.

Has the Oilers' goal tending improved since Nelson? Scrivens' adjusted save% (5v5) before Nelson was 90.9% and Fasth's was 90.2%. After Nelson (excluding the shared coaching time with MacTavish), Scrivens is 91.7% and Fasth is 91.2%. With an improvement of 0.9%, this translates into 5 fewer goals against over 23 games (547 shots against) . In 1-goal games, which the Oilers have had many, that's the difference between a loss and a tie, or a win. The reason, then, for improvement in goals against is most simply explained as regression to the mean. Most goalies regress toward a 92% save% (except for the few elite goaltenders like Lundqvist). In other words, due to variance or "back luck," the goalies were under-performing under Eakins, but under Nelson they are playing closer to their "true" ability. Is it logical to attribute an improved save% to Nelson's coaching? I prefer the simpler explanation: regression to the mean.

Second Reason for Losing: Poor Shooting%

The other problem the Oilers were having under Eakins was scoring. Have they generated more even-strength (5v5) shot attempts and scoring chances with Nelson compared to Eakins? With Eakins, the Oilers 5v5 Corsi differential was 50% and the scoring chance differential was 49.4%. Under Nelson, the 5v5 Corsi differential is 46.6% & scoring chance differential is 41.8%. Thus, at 5v5, the Oilers offense generation has worsened under Nelson. No doubt the permanent loss of Perron, in addition to the injury-loss games of Hall and Pouilot have contributed to this drop in offense.

But what about shooting%? The Oilers shooting% has been poor, but this is a tricky metric.It doesn't necessarily reflect team ability to score. It tends to regress toward 8% regardless of team. For that reason, it's considered a "luck" or "variance" variable much like save%. Before Nelson, shooting% was 6.7%. After Nelson, it's 7.6%. This translates into 5 more goals over 23 games. Again, similar to improved save%, this is a case of regression to the mean.

Now one could say the Oilers shooting% has improved because they are crashing the net more, deflecting more shots, and screening the goalie more. If this is true, we would see an increase in scoring chances. But as we saw above, scoring chance differential has declined under Nelson, not improved.

It's All About Regression to the Mean
To summarize, the Oilers goaltending has regressed toward the mean as has their shooting%. Both these statistical factors have boosted the goal differential by 10 goals (5 for & 5 against) compared to these metrics during Eakins' period. These statistics happen on their own as more games are played, specifically, short-term unusually high or low numbers in shooting and save percentage gradually level out to their respective averages over the long-term. 

All that being said, one sets of factors has improved under Nelson and that is special teams. Before Eakins was fired, the Oilers' special teams ranked 14th to 18th, which dipped even further under MacTavish and Nelson. (The power-play is reliably measured with Corsi-For per 60 min & the penalty-kill is reliably measured with Corsi-Against per 60 min.) Based on my possession-based ranking system, under Nelson, their PP moved up to 7th and the PK improved to 12th. When separating Eakins' from Nelson's period, the differences are even more pronounced. As noted by Jonathan Willis, under Nelson's time only, the Oilers' PP ranks 4th and the PK ranks 3rd. I will credit Nelson for improving the special teams. 

Win/loss records and goal differentials can be deceiving over short periods of time. They are not reliable indicators of what's actually happening due to the randomness in goals in smaller samples. Thus, what we're seeing as improvement under Nelson at even-strength is somewhat illusory. What's occurring is regression to the mean. Nonetheless, I will acknowledge Nelson has made improvements to the team--based on measurable evidence--with respect to special teams, which is good. But the drop in even-strength performance is a concern and shows that the team really isn't all that better overall.

I'll also acknowledge, though, that losing Perron (traded to Pittsburgh) and losing Hall to injury (the Oilers' best possession and offensive player) is contributing to the team's weakened Corsi and scoring chance metrics. Thus, I am not implying Nelson is making the team worse. The loss of top-6 forwards like Perron, Pouliot, and Hall has undoubtedly figured strongly into the Oilers reduced offensive generation.

Thanks for reading! I welcome your feedback and criticism. If you don't agree, let me know (with your reasons, of course).

Walter Foddis
You can follow me on Twitter @Waltlaw69

Notes. Data courtesy of war-on-ice.com.

Addendum: I didn't realize Jonathan Willis had provided a similar analysis about 5 days ago. We pretty much come to the same conclusions. One thing Willis did was break down who was scoring more to improve the team's shooting%. According to Willis, the improvement can be completely attributed to Eberle's scoring. Over the last 5 games, I think you can also add Yakupov's scoring, whose goal-scoring is long overdue. Thanks to Derek Roy's self-appointed mentoring, Yakupov's scoring is finally "regressing to the mean."
2 Comments
Walter Foddis
2/20/2015 06:39:05 am

One way to assess the impact a player has on team possession is to look at the Relative Corsi differential. This metric tells us a team's Corsi differential when the player is on the ice vs. off the ice. Thus, we can see how much better, or worse, is a team's shot attempt differential with the player on the ice.

I noted that the loss of Perron (trade), Hall (injury), and Pouliot (injury) has undoubtedly hurt the Oilers ability to generate 5v5 offense under Nelson. Here are the Relative Corsi differentials for each at 5v5:

Perron - 3.3% (Corsi% = 52.3%)
Hall - 5.2% (Corsi%=52.3%)
Pouliot - 2.6% (Corsi%=50.9%)

Taking Hall's Relative Corsi figure, 5.2% means that when Hall is on the ice, the team's Corsi differential is 52.3%. When he is off the ice, the team's Corsi% is 47.1%. For Perron, when on the ice, the team's Corsi% is 52.3% and when he is off, it is 49%.

As I noted in my blog, the team's Corsi differential under Nelson, compared to Eakins, has dropped from 50% to 46.6%. Hard not to conclude that when you lose strong possession players like Hall & Perron, the team's overall possession numbers will falter. Much of what we're seeing in the team's reduced offensive generation is probably due to losing these top-6 forwards.

I did not mean to suggest the Oilers are playing worse under Nelson, although that could be assumed by the title. My intent was to point out what the analytics are telling us. Namely, shot attempt and scoring chance generation has dropped, while shooting and save percentages have regressed toward their means, which are the main measurable reasons why the Oilers are winning more games under Nelson.

Reply
Walter Foddis
2/20/2015 06:39:15 am

One way to assess the impact a player has on team possession is to look at the Relative Corsi differential. This metric tells us a team's Corsi differential when the player is on the ice vs. off the ice. Thus, we can see how much better, or worse, is a team's shot attempt differential with the player on the ice.

I noted that the loss of Perron (trade), Hall (injury), and Pouliot (injury) has undoubtedly hurt the Oilers ability to generate 5v5 offense under Nelson. Here are the Relative Corsi differentials for each at 5v5:

Perron - 3.3% (Corsi% = 52.3%)
Hall - 5.2% (Corsi%=52.3%)
Pouliot - 2.6% (Corsi%=50.9%)

Taking Hall's Relative Corsi figure, 5.2% means that when Hall is on the ice, the team's Corsi differential is 52.3%. When he is off the ice, the team's Corsi% is 47.1%. For Perron, when on the ice, the team's Corsi% is 52.3% and when he is off, it is 49%.

As I noted in my blog, the team's Corsi differential under Nelson, compared to Eakins, has dropped from 50% to 46.6%. Hard not to conclude that when you lose strong possession players like Hall & Perron, the team's overall possession numbers will falter. Much of what we're seeing in the team's reduced offensive generation is probably due to losing these top-6 forwards.

I did not mean to suggest the Oilers are playing worse under Nelson, although that could be assumed by the title. My intent was to point out what the analytics are telling us. Namely, shot attempt and scoring chance generation has dropped, while shooting and save percentages have regressed toward their means, which are the main measurable reasons why the Oilers are winning more games under Nelson.

Reply



Leave a Reply.

    Edmonton Oilers Corsi Contribution & Shot Attempt Generation Efficiency - 2014/15

    Walt's NHL Power Rankings - 2014/15

    Archives

    April 2015
    March 2015
    February 2015
    January 2015
    December 2014
    November 2014
    August 2014

    Categories

    All
    Advanced Stats
    Analytics Of A Trade
    Anze Kopitar
    Arizone Coyotes
    Beer League Heroes
    Buffalo Sabres
    Connor McDavid
    Corsi
    Dallas Eakins
    David Backes
    David Perron
    Derek Roy
    Devan Dubnyk
    Edmonton Oilers
    Fancy Stats
    Fancy Stats Rankings
    Fancy Stats Summary
    Game Summary
    Joe Thornton
    John Tavares
    Jonathan Toews
    Marc Arcobello
    Mark Fraser
    Minnesota Wild
    Nail Yakupov
    New York Islanders
    NHL
    NHL Power Rankings
    Patrice Bergeron
    PDO
    Pittsburgh Penguins
    Playoff Series Breakdown
    Rob Klinkhammer
    Selke Trophy
    St.Louis Blues
    Todd Nelson
    Trades
    Travis Zajac
    Walter Foddis

    RSS Feed

Powered by Create your own unique website with customizable templates.