Image:
Invested in two new Leafs jerseys this offseason: One for my optimistic side, and another for just in case the wheels fall off again. 

Invested in two new Leafs jerseys this offseason: One for my optimistic side, and another for just in case the wheels fall off again. 

Rocking a slick new layout on both my Twitter and Tumblr Mobile pages!

Speaking of which — follow me on twitter!! twitter.com/LeafsRamblings

Because this is an issue that unfortunately still needs to be addressed.

The Philadelphia Flyers have made a really good stride forward in getting rid of their “Ice Girls” this year (Essentially, scantily-clad women who come onto the ice during breaks to shovel the ice, because I guess not enough people go to hockey games that some markets feel the need to sell sex during breaks). To the teams who continue to utilize this tactic: stop, now. You are cheapening the game with your blatant sexism and shedding a very undesirable light on this sport. You’re a disgrace to the sport as a whole. Enough is enough. 

An Introduction To Weighted Corsi Score (Updated 9/25)

6 days ago

By now, most hockey fans (even casual mainstream media-following fans) have probably at least heard of hockey’s basic Advanced Stats — the likes of Corsi, Fenwick, and PDO. If you are a fan of a historically good Corsi team like the LA Kings (I hate you) or a historically awful Corsi team like the Leafs (join the club), these words probably pop up more times than not while reading about your favorite team.

Last season was when advanced stats really began popping up everywhere within the sphere of hockey news that I frequently read / watch / listen to. The Toronto Maple Leafs were a historically terrible possession team, finishing second-worst to Buffalo in that department (along with many others) come the end of the season, despite many in the mainstream media claiming that the Leafs’ flukey hot patches during the season were a sustainable method of success that would lead to a berth in the playoffs. Heck, even I bought into it. I didn’t want to believe that they would come crashing down. I just liked seeing them win.

Of course, I was obviously in denial. It was clear that there was a lot wrong with this team that was being outshot on a nightly basis, relying almost solely on the presence of their red-hot run-and-gun first line and Vezina-calibre goaltending from Jonathan Bernier to stay afloat. And, even more obviously, they did ultimately come crashing down.

While they were still riding those ridiculous, odd-defying success streaks however, there were critics of advanced stats around every corner. While the fact that the Leafs were being outshot night in and night out was undeniable, they argued that the shot quality that the Leafs allowed was not great, and that they limiting the opposition’s shots to high up in the offensive zone near the blue line. This was true to a point — but only to a point. A tiny, microscopic point that neglects the fact that this team bled turnovers and could not defend themselves out of a wet paper bag. When their offense ultimately dried up, they faltered — permanently — and James Reimer arrived in the spotlight just in time to be Leafs Nation’s favorite scapegoat. 

Truthfully, I still think a lot about both sides of this argument. Don’t get me wrong; I subscribe to the advanced stats side of the argument wholeheartedly, but I do allow myself some room to attempt to build on these early developments in hockey’s analytical world. Taking the arguments of naysayers into consideration, as mentioned above, I wondered if there was a way to measure shot / shot attempt quality in an effort to gain a better understanding of not only which teams are the best teams in terms of Corsi, but also — and bear with me here — to measure the quality of a team’s Corsi numbers. What if different types of attempts were categorized and assigned a point-value in an effort to determine not only which teams have the puck most often and take the most attempts, but also to determine which teams make best use of their possession numbers. 

Before I go on, I would like to stress the fact that this is a statistic in the earliest of early stages. What I am trying to do here is to essentially establish the framework and some general means of applying a statistic that (to my knowledge) cannot even be put to use with the information that we currently have.

Weighted Corsi Score

What is it?

As explained above, Weighted Corsi Score is a means of measuring the quality of a team’s attempts on net. At its core, it is a method of determining which teams make the best use of their time with the puck and the attempts that they take on net. 

What about these “points” then? 

Once again, I’m going to stress how primitive this concept is, but as it stands now, I’ve outlined a concept for the point system in this diagram below 

(original diagram taken from war-on-ice.com

What the original diagram illustrates is the league-wide success rate of shots taken from various points within the offensive zone (all at even strength, 5v5). Unlike regular Corsi, which (correct me if I’m wrong) takes into account shots that come from outside the blue line, including dump-ins that happen to hit the net, Weighted Corsi Score does not take these into account. 

Based off of this diagram, there are three very prominent divisions of success areas: the dark green, which is the space between the hash marks and the front of the net, the light green, which is the area down the center from the blue line and past the face-off dots on both sides, and the gray area, consisting of essentially everything else in front of the goal-line in the offensive zone.

Which brings me to my next point: much like shot attempts from outside the offensive zone, shots taken from behind the goal line (regardless of whether or not they find their way in) should not count towards Weighted Corsi Score. My reasoning for this is that goals resulting from these plays are simply too lucky to accurately quantify or predict. 

Under Weighted Corsi Score, shot attempts (Shots on goal, shots that hit the post, shots that miss the net and shots that are blocked) aside from goals that come from the dark green area come in at .75 points each. Getting into this area is difficult and your shots are almost guaranteed to be of the best quality in this area. Shots that match the same criteria as previously mentioned that come from the light green region count for .5 points each, as they are represent a region where shooting percentages are lower than directly in front of the net. 

While these regions follow similar criteria, I envision that the gray areas should work somewhat differently. Given that shooting percentages are guaranteed to be lowest in these areas (as the shots are more than often going to be of lower quality and from father away and thus more easily stopped) the point value for these attempts vary based on the success of the shot. A shot on goal (saved) or a shot that hits the post is worth .25 points, while a shot attempt from the gray area that misses the net or is blocked is worth .15 points. Still something, but not as much. 

Now we get to goals. Goals within the gray area (where, once again, league shooting percentages are at their lowest) are worth 100 points, divided by the scoring team’s PDO, and multiplied by the opposing goaltender’s save percentage. 

For Example: 

Let’s imagine that Phil Kessel takes a shot from above the circles that beats Ondrej Pavelec for a goal (not too hard to imagine, really). The goal’s initial value would be 100 points. 

Next up, we need Toronto’s PDO (save percentage + shooting percentage), essentially to determine whether this is another “lucky’” thing happening to an equally “lucky” team. 

Let’s use the Leafs’ final PDO rating from last year — 101.20. 

And now we divide: 100 / 101.20 = .98 

So far, we have the Leafs raking in a .98 Corsi Score for this goal. However, there is still one more step to go. As we all know, Ondrej Pavelec is not necessarily known for being good at what he does. His .901 save percentage would back up that claim. So, let’s put it to use. 

The next step is to multiply our Corsi Score on the play so far (.98) by Pavelec’s save percentage (.901) 

.98 x .901 = .88

Suddenly, the Leafs aren’t raking in much more than they would get for a missed shot attempt in front of the net for a goal. Why? Because they probably got pretty lucky, similar to what happened for a pretty big chunk of last season.

It’s a little easier to assign a Corsi Score to the goals that are scored from within the green areas. Goals that occur in the light green area, for example, should not factor in the PDO aspect that is utilized when analyzing the gray areas. Given the high league-wide shooting percentage in this area as it is, it wouldn’t make sense. You are simply more likely to score in this area regardless of who you are, but it’s the act of finding your way into this space that counts. Goals coming from the light green region have an initial Corsi Score of 1 point, which is then subsequently multiplied by the opposing goaltender’s save percentage. What does this mean? To put it frankly, it means that the Corsi Score of a goal occurring in this area is entirely based on the goaltender’s save percentage, and the odds that were overcome to result in this goal. If John Scott scored a goal on Tuuka Rask from within this area, for example, his initial Corsi Score of 1 would be divided by .930, resulting in — you guessed it — a .93 Corsi Score for the goal. 

***Note: When it comes to utilizing team stats like PDO and Save Percentage in these instances, the question of what time period we should pull these stats from definitely comes into question. If, for example, it’s the second game of the NHL season and Rask was coming off of a first game in which he allowed 5 goals on 20 shots (.750 save percentage), that would certainly skew the data. This is where this concept needs to be explored further to come to a definitive conclusion. 

Finally, we come to the dark green area, the easiest of them all. Due to the extremely high likelihood of a goal in this area, all goals are worth 1 point. Plain and simple. I can’t see why any outside factors should play a significant role in the Corsi Score of a play like this.

So, once it’s all said and done, here’s what we have. Let’s create a really easy example to work with here. 

Suppose that in a given game, the Leafs attempt 50 even-strength shots.

20 of those attempts are shots on goal

The other 30 are attempts are missed attempts

Out of the 30 missed attempts, 18 come from the gray area, 8 come from the light green area, and 4 more come from within the dark green. 

Corsi Score for missed attempts: (18[missed attempts from gray area] x .15[points for each missed attempt) + (8 x .5) + (4 x .75) = 9.7 Corsi Score from missed attempts.

3 of the 20 shots on goal result in goals (we’ll come back to this).

Out of the 17 remaining shots on goal, 10 come from the gray area, 5 come from within the light green, and 2 come from the dark green area: (10 x .25) + (5 x .5) + (2 x .75) = 6.5 Corsi Score from shot on goal

Out of the three shots on goal that resulted in goals, one goal comes from each of the gray, light green and dark green regions of the offensive zone. The team PDO is 101.2, and the opposing goaltender’s save percentage is around league average at .912. 

Gray goal: 100 / 101.2 = .98 x .912 = .89 Corsi Score

Light Green goal: .912 Corsi Score

Dark Green goal: 1 Corsi Score

Goals total: 2.8 Corsi Score

Adding together the Corsi Scores from missed attempts, shots, and goals:

9.7 + 6.5 + 2.8 = 19 Raw Corsi Score

And, as our final step to finding a Weighted Corsi Score, we take the number of attempted 5v5 shots and divide this number by the Raw Corsi Score above

50 / 19 = 2.63 Weighted Corsi Score

Now, to compare these numbers to traditional Corsi numbers, the Corsi differential (Corsi For minus Corsi Against) from the game is taken into consideration. This measure acts to ensure that a team does not take, for example, 10 shot attempts in a game, get lucky and score on 3 of them and have the rest come from inside the slot). Imagine, in this scenario, that Toronto’s Corsi For was 51%, meaning the opposition’s would be 49%. 

51 - 49 = 2

Final Numbers: 2.63 Weighted Corsi Score with +2%CF

How do we interpret these numbers? 

While we have no tangible data to work with (As nobody tracks shot attempts based on location as far as I know), the goal here appears simple: to amass a Weighted Corsi Score as close to 1 as possible. Each attempt is guaranteed to garner you points, and better and smarter attempts garner you more points per attempt. The math behind it may seem tedious, but the concept is as simple as that. The closer the point value of each of your attempts is to 1, the closer your overall Weighted Corsi Score should be to 1 in the end. In addition to this, a team should of course always strive for a positive Corsi score. 

In Conclusion

This, of course, is all just a concept as it currently stands. However, giving weight to different types of shot attempts would help to provide a definitive basis not for shot quality, but for quality of shot attempts. It’s almost as if the whining and lamenting of the mainstream media met up with the world of basement-blogging fancy stats, and the two had a baby. On top of this, that baby might be pretty damn useful in the future. 

An Introduction to Weighted Corsi Score (Updated 9/24)

1 week ago

By now, most hockey fans (even casual mainstream media-following fans) have probably at least heard of hockey’s basic Advanced Stats — the likes of Corsi, Fenwick, and PDO. If you are a fan of a historically good Corsi team like the LA Kings (I hate you) or a historically awful Corsi team like the Leafs (join the club), these words probably pop up more times than not while reading about your favorite team.

Last season was when advanced stats really began popping up everywhere within the sphere of hockey news that I frequently read / watch / listen to. The Toronto Maple Leafs were a historically terrible possession team, finishing second-worst to Buffalo in that department (along with many others) come the end of the season, despite many in the mainstream media claiming that the Leafs’ flukey hot patches during the season were a sustainable method of success that would lead to a berth in the playoffs. Heck, even I bought into it. I didn’t want to believe that they would come crashing down. I just liked seeing them win.

Of course, I was obviously in denial. It was clear that there was a lot wrong with this team that was being outshot on a nightly basis, relying almost solely on the presence of their red-hot run-and-gun first line and Vezina-calibre goaltending from Jonathan Bernier to stay afloat. And, even more obviously, they did ultimately come crashing down.

While they were still riding those ridiculous, odd-defying success streaks however, there were critics of advanced stats around every corner. While the fact that the Leafs were being outshot night in and night out was undeniable, they argued that the shot quality that the Leafs allowed was not great, and that they limiting the opposition’s shots to high up in the offensive zone near the blue line. This was true to a point — but only to a point. A tiny, microscopic point that neglects the fact that this team bled turnovers and could not defend themselves out of a wet paper bag. When their offense ultimately dried up, they faltered — permanently — and James Reimer arrived in the spotlight just in time to be Leafs Nation’s favorite scapegoat. 

Truthfully, I still think a lot about both sides of this argument. Don’t get me wrong; I subscribe to the advanced stats side of the argument wholeheartedly, but I do allow myself some room to attempt to build on these early developments in hockey’s analytical world. Taking the arguments of naysayers into consideration, as mentioned above, I wondered if there was a way to measure shot / shot attempt quality in an effort to gain a better understanding of not only which teams are the best teams in terms of Corsi, but also — and bear with me here — to measure the quality of a team’s Corsi numbers. What if different types of attempts were categorized and assigned a point-value in an effort to determine not only which teams have the puck most often and take the most attempts, but also to determine which teams make best use of their possession numbers. 

Before I go on, I would like to stress the fact that this is a statistic in the earliest of early stages. What I am trying to do here is to essentially establish the framework and some general means of applying a statistic that (to my knowledge) cannot even be put to use with the information that we currently have.

Weighted Corsi Score

What is it?

As explained above, Weighted Corsi Score is a means of measuring the quality of a team’s attempts on net. At its core, it is a method of determining which teams make the best use of their time with the puck and the attempts that they take on net. 

What about these “points” then? 

Once again, I’m going to stress how primitive this concept is, but as it stands now, I’ve outlined a concept for the point system in this diagram below 

(original diagram taken from war-on-ice.com)

image

What the original diagram illustrates is the league-wide success rate of shots taken from various points within the offensive zone (all at even strength, 5v5). Unlike regular Corsi, which (correct me if I’m wrong) takes into account shots that come from outside the blue line, including dump-ins that happen to hit the net, Weighted Corsi Score does not take these into account. 

Based off of this diagram, there are three very prominent divisions of success areas: the dark green, which is the space between the hash marks and the front of the net, the light green, which is the area down the center from the blue line and past the face-off dots on both sides, and the gray area, consisting of essentially everything else in front of the goal-line in the offensive zone.

Which brings me to my next point: much like shot attempts from outside the offensive zone, shots taken from behind the goal line (regardless of whether or not they find their way in) should not count towards Weighted Corsi Score. My reasoning for this is that goals resulting from these plays are simply too lucky to accurately quantify or predict. 

Under Weighted Corsi Score, shot attempts (Shots on goal, shots that hit the post, shots that miss the net and shots that are blocked) aside from goals that come from the dark green area come in at .75 points each. Getting into this area is difficult and your shots are almost guaranteed to be of the best quality in this area. Shots that match the same criteria as previously mentioned that come from the light green region count for .5 points each, as they are represent a region where shooting percentages are lower than directly in front of the net. 

While these regions follow similar criteria, I envision that the gray areas should work somewhat differently. Given that shooting percentages are guaranteed to be lowest in these areas (as the shots are more than often going to be of lower quality and from father away and thus more easily stopped) the point value for these attempts vary based on the success of the shot. A shot on goal (saved) or a shot that hits the post is worth .25 points, while a shot attempt from the gray area that misses the net or is blocked is worth .15 points. Still something, but not as much. 

Now we get to goals. Goals within the gray area (where, once again, league shooting percentages are at their lowest) are worth 100 points, divided by the scoring team’s PDO, and multiplied by the opposing goaltender’s save percentage. 

For Example: 

Let’s imagine that Phil Kessel takes a shot from above the circles that beats Ondrej Pavelec for a goal (not too hard to imagine, really). The goal’s initial value would be 100 points. 

Next up, we need Toronto’s PDO (save percentage + shooting percentage), essentially to determine whether this is another “lucky’” thing happening to an equally “lucky” team, or perhaps even a lucky thing happening to a statistically unlucky team (which would inflate the value of the goal). Let’s use the Leafs’ final PDO rating from last year — 101.20. 

And now we divide: 100 / 101.20 = .98 

So far, we have the Leafs raking in a .98 Corsi Score for this goal. However, there is still one more step to go. As we all know, Andrej Pavelec is not necessarily known for being good at what he does. His .901 save percentage would back up that claim. So, let’s put it to use. 

The next step is to multiply our Corsi Score on the play so far (.98) by Pavelec’s save percentage (.901) 

.98 x .901 = .88

Suddenly, the Leafs aren’t raking in much more than they would get for a missed shot attempt in front of the net for a goal. Why? Because they probably got pretty lucky, similar to what happened for a pretty big chunk of last season.

It’s a little easier to assign a Corsi Score to the goals that are scored from within the green areas. Goals that occur in the light green area, for example, should not factor in the PDO aspect that is utilized when analyzing the gray areas. Given the high league-wide shooting percentage in this area as it is, it wouldn’t make sense. You are simply more likely to score in this area regardless of who you are, but it’s the act of finding your way into this space that counts. Goals coming from the light green region have an initial Corsi Score of 1 point, which is then subsequently multiplied by the opposing goaltender’s save percentage. What does this mean? To put it frankly, it means that the Corsi Score of a goal occurring in this area is entirely based on the goaltender’s save percentage, and the odds that were overcome to result in this goal. If John Scott scored a goal on Tuuka Rask from within this area, for example, his initial Corsi Score of 1 would be divided by .930, resulting in — you guessed it — a .93 Corsi Score for the goal. 

Note: When it comes to utilizing team stats like PDO and Save Percentage in these instances, the question of what time period we should pull these stats from definitely comes into question. If, for example, it’s the second game of the NHL season and Rask was coming off of a first game in which he allowed 5 goals on 20 shots (.750 save percentage), that would certainly skew the data. This is where this concept needs to be explored further to come to a definitive conclusion on which stats should be used.

Finally, we come to the dark green area, the easiest of them all. Due to the extremely high likelihood of a goal in this area, all goals are worth 1 point. Plain and simple. I can’t see why any outside factors should play a significant role in the Corsi Score of a play like this.

So, once it’s all said and done, here’s what we have. Let’s create a really easy example to work with here. 

Suppose that in a given game, the Leafs attempt 50 even-strength shots.

-20 of those attempts are shots on goal

-The other 30 are attempts are missed attempts

-Out of the 30 missed attempts, 18 come from the gray area, 8 come from the light green area, and 4 more come from within the dark green. 

Corsi Score for missed attempts: (18[missed attempts from gray area] x .15[points for each missed attempt) + (8 x .5) + (4 x .75) = 9.7 Corsi Score from missed attempts.

-3 of the 20 shots on goal result in goals (we’ll come back to this).

-Out of the 17 remaining shots on goal, 10 come from the gray area, 5 come from within the light green, and 2 come from the dark green area: (10 x .25) + (5 x .5) + (2 x .75) = 6.5 Corsi Score from shot on goal

-Out of the three shots on goal that resulted in goals, one goal comes from each of the gray, light green and dark green regions of the offensive zone. The team PDO is 101.2, and the opposing goaltender’s save percentage is around league average at .912. 

Gray goal: 100 / 101.2 = .98 x .912 = .89 Corsi Score

Light Green goal: .912 Corsi Score

Dark Green goal: 1 Corsi Score

Goals total: 2.8 Corsi Score

Adding together the Corsi Scores from missed attempts, shots, and goals:

9.7 + 6.5 + 2.8 = 19 Raw Corsi Score

And, as our final step, we take the number of attempted 5v5 shots and divide this number by the Raw Corsi Score above

50 / 19 = 2.63 Weighted Corsi Score

Now, to compare these numbers to traditional Corsi numbers, the Corsi differential (Corsi For minus Corsi Against) from the game is taken into consideration. This measure acts to ensure that a team does not take, for example, 10 shot attempts in a game, get lucky and score on 3 of them and have the rest come from inside the slot). Imagine, in this scenario, that Toronto’s Corsi For was 51%, meaning the opposition’s would be 49%. 

51 - 49 = 2

Final Numbers: 2.63 Weighted Corsi Score with +2%CF

 

How do we interpret these numbers? 

While we still have no measurable data to work with (As nobody tracks shot attempts based on location as far as I know), the theoretical goal here would appear to be simple: to amass a Weighted Corsi Score as close to 1 as possible. Each attempt is guaranteed to garner you points, and better and smarter attempts garner you more points per attempt. The math behind it may seem tedious, but the concept is as simple as that. The closer the point value of each of your attempts is to 1, the closer your overall Weighted Corsi Score should be to 1 in the end. In addition to this, a team should of course always strive for a positive Corsi score. 

In Conclusion

This, of course, is all just a concept as it currently stands. However, giving weight to different types of shot attempts would help to provide a definitive basis not for shot quality, but for quality of shot attempts. It’s almost as if the whining and lamenting of the mainstream media met up with the world of basement-blogging fancy stats, and the two had a baby. On top of this, that baby might be pretty damn useful in the future. 

Image:
About to watch the baes for the first time this (pre)season!

About to watch the baes for the first time this (pre)season!

Image:
So ready for hockey season to kick off

So ready for hockey season to kick off

Image:
Leafs Close fenwick% since the beginning of the 2010-11 season. Red line shows when Randy Carlyle was hired. Leafs were actually decent possession team under Wilson but have steadily declined since Carlyle’s hiring. Can new analytics guys reverse this trend with Carlyle still the head coach? It can’t get much worse. 

Leafs Close fenwick% since the beginning of the 2010-11 season. Red line shows when Randy Carlyle was hired. Leafs were actually decent possession team under Wilson but have steadily declined since Carlyle’s hiring. Can new analytics guys reverse this trend with Carlyle still the head coach? It can’t get much worse. 

Follow Me On Twitter! @LeafsRamblings

3 weeks ago

Hey All,

I’ve begun actively using twitter a lot more than before. Give me a follow at https://twitter.com/LeafsRamblings! 

I’ll be back to posting regular pre-game and opinion pieces once the season gets underway, but for now the best way to follow me is definitely through the short-and-sweetness that is twitter. 

#IsItOctoberYet?