DATA: Points Percentage vs. Shots Advantage, 1971 to present

[Doctor No]

Something I put together this morning.

Regular season NHL games only, 1971-72 to present (2015-16 includes games through February 22). Probably a few data errors over the course of all of these games, so keep that in mind.

Each row represents one regular season. Each column reflects the shots on goal advantage that a team had, and then the percentage of available game points that the same team earned.

For instance, if a team had a shot advantage of 20, and won the game, that column would get credited with two points for the win.

SEASON | -31 or less | -30 to -26 | -25 to -21 | -20 to -16 | -15 to -11 | -10 to -6 | -5 to -1 | 0 (even) | 1 to 5 | 6 to 10 | 11 to 15 | 16 to 20 | 21 to 25 | 26 to 30 | 31 or more
1971-72|0%|33%|26%|22%|29%|38%|45%|50%|55%|62%|71%|78%|74%|67%|100%
1972-73|0%|0%|10%|19%|33%|37%|45%|50%|55%|63%|67%|81%|90%|100%|100%
1973-74|30%|10%|16%|20%|32%|40%|45%|50%|55%|60%|68%|80%|84%|90%|70%
1974-75|0%|13%|11%|19%|32%|40%|37%|50%|63%|60%|68%|81%|89%|88%|100%
1975-76|5%|4%|15%|22%|30%|35%|48%|50%|52%|65%|70%|78%|85%|96%|95%
1976-77|0%|14%|22%|31%|37%|40%|45%|50%|55%|60%|63%|69%|78%|86%|100%
1977-78|25%|18%|30%|26%|32%|37%|45%|50%|55%|63%|68%|74%|70%|82%|75%
1978-79|0%|30%|19%|33%|36%|36%|46%|50%|54%|64%|64%|67%|81%|70%|100%
1979-80|33%|19%|31%|25%|34%|39%|50%|50%|50%|61%|66%|75%|69%|81%|67%
1980-81|29%|0%|32%|32%|40%|37%|49%|50%|51%|63%|60%|68%|68%|100%|71%
1981-82|8%|22%|24%|27%|33%|45%|43%|50%|57%|55%|67%|73%|76%|78%|92%
1982-83|40%|10%|15%|34%|34%|41%|45%|50%|55%|59%|66%|66%|85%|90%|60%
1983-84|0%|29%|37%|32%|38%|41%|52%|50%|48%|59%|62%|68%|63%|71%|100%
1984-85|33%|44%|26%|33%|37%|39%|45%|50%|55%|61%|63%|67%|74%|56%|67%
1985-86||21%|23%|42%|42%|41%|50%|50%|50%|59%|58%|58%|77%|79%|
1986-87|0%|40%|45%|52%|45%|45%|50%|50%|50%|55%|55%|48%|55%|60%|100%
1987-88|25%|20%|40%|32%|38%|44%|44%|50%|56%|56%|62%|68%|60%|80%|75%
1988-89|40%|27%|27%|40%|44%|49%|43%|50%|57%|51%|56%|60%|73%|73%|60%
1989-90|50%|64%|28%|39%|45%|47%|44%|50%|56%|53%|55%|61%|73%|36%|50%
1990-91|40%|32%|42%|35%|39%|45%|45%|50%|55%|55%|61%|65%|58%|68%|60%
1991-92||25%|27%|26%|40%|50%|50%|50%|50%|50%|60%|74%|73%|75%|
1992-93|0%|33%|39%|35%|34%|41%|49%|50%|51%|59%|66%|65%|61%|67%|100%
1993-94|17%|38%|53%|40%|50%|46%|47%|49%|53%|54%|50%|60%|47%|63%|83%
1994-95|50%|40%|40%|39%|41%|47%|52%|50%|48%|53%|59%|61%|60%|60%|50%
1995-96|50%|41%|44%|37%|46%|45%|44%|50%|56%|55%|54%|63%|56%|59%|50%
1996-97|30%|50%|37%|44%|47%|50%|53%|50%|47%|50%|53%|56%|63%|50%|70%
1997-98|83%|56%|46%|50%|49%|50%|48%|50%|52%|50%|51%|50%|54%|44%|17%
1998-99|58%|50%|38%|57%|45%|48%|47%|50%|53%|52%|55%|43%|63%|50%|42%
1999-00|50%|23%|43%|41%|51%|49%|49%|51%|55%|55%|56%|66%|63%|82%|50%
2000-01|0%|42%|40%|51%|47%|51%|50%|52%|55%|55%|58%|54%|67%|62%|100%
2001-02|57%|31%|38%|49%|50%|50%|55%|54%|50%|54%|54%|58%|68%|69%|64%
2002-03|75%|40%|45%|53%|55%|50%|53%|56%|53%|55%|52%|53%|60%|70%|38%
2003-04|42%|50%|55%|51%|51%|47%|50%|55%|54%|59%|55%|56%|48%|57%|58%
2005-06|50%|29%|54%|41%|51%|54%|54%|53%|58%|57%|63%|67%|60%|86%|50%
2006-07|67%|32%|64%|54%|51%|52%|54%|60%|57%|59%|59%|57%|51%|86%|63%
2007-08|33%|48%|48%|53%|54%|56%|54%|56%|57%|55%|57%|59%|65%|63%|67%
2008-09|50%|58%|46%|52%|52%|52%|56%|55%|56%|58%|59%|64%|60%|60%|60%
2009-10|70%|79%|56%|57%|55%|62%|56%|56%|55%|51%|56%|56%|64%|36%|55%
2010-11|25%|38%|69%|61%|61%|57%|59%|60%|52%|56%|50%|52%|40%|75%|75%
2011-12|50%|60%|65%|62%|55%|55%|60%|53%|54%|57%|58%|48%|49%|65%|50%
2012-13||63%|57%|49%|61%|56%|48%|60%|61%|57%|51%|63%|50%|44%|
2013-14|60%|57%|55%|56%|56%|56%|59%|58%|54%|56%|57%|56%|63%|59%|50%
2014-15|50%|47%|44%|51%|50%|59%|58%|57%|55%|53%|62%|61%|72%|62%|58%
2015-16||81%|61%|61%|60%|61%|57%|55%|56%|50%|53%|49%|54%|31%|

It does appear that having a shots advantage is becoming less and less of a real advantage.

Next post will have this grouped up into fewer columns (perhaps more useable).

Click to read more...

 

[Doctor No]

Grouped up the end columns a bit here (admittedly there are sample size concerns):

SEASON | -16 or less | -15 to -11 | -10 to -6 | -5 to -1 | 0 to 0 | 1 to 5 | 6 to 10 | 11 to 15 | 16 or more
1971-72|21%|29%|38%|45%|50%|55%|62%|71%|79%
1972-73|14%|33%|37%|45%|50%|55%|63%|67%|86%
1973-74|19%|32%|40%|45%|50%|55%|60%|68%|81%
1974-75|14%|32%|40%|37%|50%|63%|60%|68%|86%
1975-76|16%|30%|35%|48%|50%|52%|65%|70%|84%
1976-77|26%|37%|40%|45%|50%|55%|60%|63%|74%
1977-78|26%|32%|37%|45%|50%|55%|63%|68%|74%
1978-79|29%|36%|36%|46%|50%|54%|64%|64%|71%
1979-80|27%|34%|39%|50%|50%|50%|61%|66%|73%
1980-81|31%|40%|37%|49%|50%|51%|63%|60%|69%
1981-82|25%|33%|45%|43%|50%|57%|55%|67%|75%
1982-83|28%|34%|41%|45%|50%|55%|59%|66%|72%
1983-84|32%|38%|41%|52%|50%|48%|59%|62%|68%
1984-85|32%|37%|39%|45%|50%|55%|61%|63%|68%
1985-86|36%|42%|41%|50%|50%|50%|59%|58%|64%
1986-87|48%|45%|45%|50%|50%|50%|55%|55%|52%
1987-88|32%|38%|44%|44%|50%|56%|56%|62%|68%
1988-89|36%|44%|49%|43%|50%|57%|51%|56%|64%
1989-90|39%|45%|47%|44%|50%|56%|53%|55%|61%
1990-91|37%|39%|45%|45%|50%|55%|55%|61%|63%
1991-92|27%|40%|50%|50%|50%|50%|50%|60%|73%
1992-93|36%|34%|41%|49%|50%|51%|59%|66%|64%
1993-94|43%|50%|46%|47%|50%|53%|54%|50%|57%
1994-95|40%|41%|47%|52%|50%|48%|53%|59%|60%
1995-96|40%|46%|45%|44%|50%|56%|55%|54%|60%
1996-97|42%|47%|50%|53%|50%|47%|50%|53%|58%
1997-98|50%|49%|50%|48%|50%|52%|50%|51%|50%
1998-99|52%|45%|48%|47%|50%|53%|52%|55%|48%
1999-00|40%|51%|49%|49%|51%|55%|55%|56%|66%
2000-01|46%|47%|51%|50%|52%|55%|55%|58%|59%
2001-02|44%|50%|50%|55%|54%|50%|54%|54%|62%
2002-03|51%|55%|50%|53%|56%|53%|55%|52%|56%
2003-04|51%|51%|47%|50%|55%|54%|59%|55%|55%
2005-06|44%|51%|54%|54%|53%|58%|57%|63%|67%
2006-07|55%|51%|52%|54%|60%|57%|59%|59%|58%
2007-08|50%|54%|56%|54%|56%|57%|55%|57%|61%
2008-09|51%|52%|52%|56%|55%|56%|58%|59%|62%
2009-10|60%|55%|62%|56%|56%|55%|51%|56%|56%
2010-11|62%|61%|57%|59%|60%|52%|56%|50%|50%
2011-12|62%|55%|55%|60%|53%|54%|57%|58%|49%
2012-13|52%|61%|56%|48%|60%|61%|57%|51%|58%
2013-14|56%|56%|56%|59%|58%|54%|56%|57%|58%
2014-15|49%|50%|59%|58%|57%|55%|53%|62%|63%
2015-16|62%|60%|61%|57%|55%|56%|50%|53%|49%

 

[Doctor No]

I'll be adding some other details about shot metrics as I get the desire.

As it stands, the common maxim of "shoot the puck to win games" appears to be fading.

 

[Michael Farkas]

I'll be adding some other details about shot metrics as I get the desire.

As it stands, the common maxim of "shoot the puck to win games" appears to be fading.

As I stated in a recent PM to you, thank goodness. The thing that was immediately recognizable about Corsi is just how easy it can be coached. A shot is not necessarily a random event that happens...to a degree, sure, but not too enough of a degree to make it gospel, in my opinion.

I'm confident that at the level I coach at, if - above all else - I wanted to coach best the Corsi team in the league, I absolutely could. It's not hard.

Now, naturally, no one is gunning for that crown...that's not the secret formula (Dallas Eakins)...but that doesn't make the data useless...it just needed time to be contextualized. If you started measuring the stock market in the summer of 1933 and had made definitive predictions by that winter, I'd be remiss to not mention what had been happening over the previous four years - more accurately, I suppose, we went from none (well, we had SOG) to some...you need time for data to settle without jumping at the first thing you can get your hands on...

 

[Doctor No]

I definitely agree - when shooting more is a consequence of good coaching, then we see the advantages (as we do, say, pre-2000).

When shooting more is a deliberate decision ("we're going to shoot more, and that's our goal"), at the expense of other things that are good, then the results aren't as good (and appear to be even detrimental to a point).

 

[Michael Farkas]

Follow evolution of defensive systems as well, removal of the red line removes the emphasis on defending lines (see: mid 90's Devils)...plus the lack of "efficiency" goaltending pre-1995 or so...so shots became the enemy...the way to prevent goals was to prevent shots...

Now, goalies stop, what, 104% of all unscreened/untipped shots from 40 feet or more or whatever...? Shots aren't the enemy. Goals are the enemy. Shots don't matter nearly as much. You don't need to protect the lines so much as you do need to protect the net...protect against second chance opportunities and high-quality chances. That's where goalies get beat up. Letting guys take wrist shots from outside the dots on the right wing boards (aka Claude Julien's defensive tactics) will not beat goalies today...you can have those shots. It's as good as a dump in when you play a collapse style and you have even an iota of transition game in your repertoire...

 

[Doctor No]

Some more fun - as many (here) know, I calculate a goaltender's actual vs. expected performance in each game, based upon the season and the opponent faced (so if the 1974-75 Canucks shoot 10%, we'd expect a goaltender facing 30 shots to stop about 27).

For a few different seasons in my data, these are the distribution of performances (weighted by minutes played), for the number of shots faced per 60 minutes.

1971-72:

|BAVG|AVG|AAVG
[0,15]|47%|30%|23%
(15,20]|30%|46%|24%
(20,25]|32%|35%|33%
(25,30]|24%|41%|35%
(30,35]|28%|36%|36%
(35,40]|28%|34%|38%
(40,45]|26%|38%|36%
(45,50]|31%|35%|34%
(50|62%|21%|18%

1981-82:

|BAVG|AVG|AAVG
[0,15]|67%|13%|20%
(15,20]|43%|31%|26%
(20,25]|30%|39%|30%
(25,30]|28%|39%|33%
(30,35]|30%|36%|33%
(35,40]|25%|35%|40%
(40,45]|21%|37%|43%
(45,50]|21%|39%|40%
(50|31%|28%|41%

1991-92:

|BAVG|AVG|AAVG
[0,15]|20%|33%|47%
(15,20]|37%|36%|27%
(20,25]|36%|35%|29%
(25,30]|30%|38%|32%
(30,35]|24%|36%|40%
(35,40]|22%|39%|39%
(40,45]|22%|46%|32%
(45,50]|31%|37%|32%
(50|43%|29%|29%

2001-02:

|BAVG|AVG|AAVG
[0,15]|43%|38%|19%
(15,20]|35%|37%|28%
(20,25]|31%|37%|32%
(25,30]|27%|40%|32%
(30,35]|25%|35%|41%
(35,40]|18%|41%|41%
(40,45]|20%|29%|50%
(45,50]|37%|17%|45%
(50|25%|35%|40%

2011-12:

|BAVG|AVG|AAVG
[0,15]|54%|24%|22%
(15,20]|43%|30%|26%
(20,25]|36%|32%|32%
(25,30]|28%|42%|30%
(30,35]|23%|38%|39%
(35,40]|21%|37%|42%
(40,45]|19%|32%|48%
(45,50]|22%|19%|59%
(50|57%|25%|18%

2015-16 (to date):

|BAVG|AVG|AAVG
[0,15]|25%|35%|40%
(15,20]|38%|41%|22%
(20,25]|34%|32%|34%
(25,30]|31%|39%|30%
(30,35]|25%|39%|36%
(35,40]|19%|32%|49%
(40,45]|21%|27%|53%
(45,50]|28%|20%|52%
(50|51%|21%|28%

 

[Doctor No]

Broadly speaking: the more shots faced, the likelihood of a below-average game (*) goes down, and the likelihood of an above-average game (*) goes up.

(*) Measured by save percentage.

 

[Dertell]

Would be great to have some form of r2 value.

As I stated in a recent PM to you, thank goodness. The thing that was immediately recognizable about Corsi is just how easy it can be coached. A shot is not necessarily a random event that happens...to a degree, sure, but not too enough of a degree to make it gospel, in my opinion.

I'm confident that at the level I coach at, if - above all else - I wanted to coach best the Corsi team in the league, I absolutely could. It's not hard.

Now, naturally, no one is gunning for that crown...that's not the secret formula (Dallas Eakins)...but that doesn't make the data useless...it just needed time to be contextualized. If you started measuring the stock market in the summer of 1933 and had made definitive predictions by that winter, I'd be remiss to not mention what had been happening over the previous four years - more accurately, I suppose, we went from none (well, we had SOG) to some...you need time for data to settle without jumping at the first thing you can get your hands on...

Go watch the kings.

 

[Doctor No]

Go watch the kings.

Any chance you'd be willing to explain what you meant by this, or are you just suggesting that Mike would enjoy watching the Kings?

 

[Dertell]

Any chance you'd be willing to explain what you meant by this, or are you just suggesting that Mike would enjoy watching the Kings?

It's pretty clear just by watching them that it is coached to shoot just about everything to the net when it gets a chance over positioning for a scoring chance.

But obviously, "watch X play" isn't a good argument in this sub-forum and for good reasons. The kings, since the year they hired Darryl Sutter (2011-12), has by a fair amount the highest CF60 of the league at 59.4 (2nd highest are the sharks at 58.7). They also have the lowest Shooting% in the league at (6.9) in that timespam.

Also I'm not sure where this "Eakin tried to win the corsi cup" comes from, considering the oilers have had extremely poor corsi while he was coaching them.

 

[Doctor No]

Thanks! That makes more sense.

And I think that "watch the game" works fine here, as a part of an argument - it's usually watching the game that causes me to look for something in the data to begin with.

 

[decma]

Also I'm not sure where this "Eakin tried to win the corsi cup" comes from, considering the oilers have had extremely poor corsi while he was coaching them.

I think it may have come from this statement by Eberle (which I believe was referencing the Eakins era). (there may have been other articles making the same point). From the Edmonton Journal:

“I think maybe at the start of the year we were really focused on shooting the puck and Corsi numbers,” Eberle told the Sun’s Derek Van Diest. “When that starts creeping in your head, you’re just shooting the puck just to shoot it and you’re not trying to create the best opportunity you can.”

The issue here seems to be that if a coach tells a player that he will be judged on his Corsi, his shots-at-net rates, it might be counterproductive. Instead of a player trying to make the best play, he might just fire it at the net from a weak scoring position. This would be a particular concern for Eberle. Of all the Oilers shooters, he’s the most picky.


http://edmontonjournal.com/sports/h...-corsi-fail-of-the-edmonton-oilers-in-2014-15

 

[Kevin Forbes]

Do goalies actually play worse when facing less work?

During the intermission of the SJ/StL game, I heard the panel regurgitate the old adage that when goalies aren't busy (not facing a lot of action) they are prone to losing focus and then being susceptible to "weak" goals.

I'm wondering if anyone has done analysis on this to see if that thought actually holds water?

Looking at the 2015-16 season, goalies faced an average of 29.58 shots/game, but when factoring in overtime periods, we're really looking at an average of about 9.5 shots per period.

So I guess, we would be looking at save percentages when there are fewer than 9.5 shots per period and then comparing to those that are right around the average and those that are above average.

Is there anywhere that would have data broken apart in such a fashion?

I suppose another factor to consider would be that if there are significantly fewer shots (assuming this means play is focused on the other end of the ice), the circumstances that do lead to a shot on goal. Does the play being primarily at one end of the ice then lead to the opposite goalie facing more breakaways and odd-man rushes?

Thoughts?

 

[JetsFan815]

Thanks for the work on this. Am I reading this data right? A team that got outshot by 16 shots or more in 2015-2016 season earned 62% points in those games? That seems a bit odd but perhaps score effects at play?

Goalie Sv% has gone up dramatically in the NHL so I wouldn't assume that the decreasing point% when outshooting the opponents by a large number of shots is a function of "coaches teaching corsi"