Corsi, shot quality, and the Toronto Maple Leafs

[Master_Of_Districts]

http://www.somekindofninja.com/nhl/index.php


That link will provide all the information you really need on why Corsi is misleading for the Leafs. If you look around the league, the Leafs are getting just as many prime scoring shots (within 10 ft) but are getting far fewer shots from outside (20+ ft).

The Corsi sucks because the Leafs allow a ton of shots from the outside but don't take many themselves. In the actual scoring positions they aren't bad at all.

It's interesting that you would characterize shots taken from a distance greater than 20 ft as being from the "outside."

How far, exactly, do you think the blueline is from the goal line?

(Hint: it's a lot more than 20 ft).

 

[Feed Me A Stray Cat]

1. Because teams win by outscoring, not by out-shooting/out-chancing. Also, because once again you are ignoring the other points I mentioned (and didn't) which influence the game.
2. Not a good data point. Just an irrelevant one on its own.
3. Not a good data point. Just an irrelevant one on its own.
4. Timing is everything. These are humans, not machines/simulations. A shorthanded goal, an end-of-period goal, etc. All goals do not have the same effect on a team. Also, if a team tends to take leads early, you will likely see more shots against them as they sit back more.
5. Just because you don't believe it doesn't make it false, or something you can just ignore.
6. I can say random things without proving it too.
7. No, it doesn't. And almost always =/= always.
8. Actually the Leafs have been one of the best teams for 65 games now, while being outshot. The small sample size referred to you thinking we would be outshot by 10 over the entire season, despite playing a small sample of teams (including some of the best at outshooting opponents, and a lot of Western teams), and a small sample of games.

1. How do teams score goals? They must take shots, and then they must convert on them. We've established that the Leafs are bad at the first part. In regards to the latter, we know that teams simply don't maintain 11%+ shooting percentages over the long haul, so that will most likely regress. As a result, the Leafs' offense will suffer unless they start generating more shots.
2. It's not irrelevant, it's part of the picture, even if it doesn't fully explain the picture, but if you want to cast it aside I can't stop you.
3. " "
4. In regards to early leads - the Leafs corsi/fenwick is terrible in close game/tied situations as well. And it seems like you think the Leafs have a special ability to score clutch goals - I personally don't believe in clutch, as it's been more or less disproven in sports such as baseball where the data is good enough to get an objective picture.
5. Nothing about Bernier's and Reimer's history leads me to believe they're elite. Perhaps they will show that they are over the next couple seasons, but as it stands that's an extraordinary assumption.
6. http://objectivenhl.blogspot.com/2010/04/repeatability-of-special-teams.html
Powerplay shooting percentage has almost no season to season correlation. Guess what does? Shot generation!
7. No use arguing this point, because despite the fact that we now have data that shows the Leafs getting outshot from 10 feet, you will simply ignore it because we don't know the specifics of each shot.
8. 65 games is not a very large sample. Plenty of teams lucked their way in S% and/or Sv% over a full season only to regress. You're correct that I simply shouldn't assume that the 10 delta will persist the full season on shot differential, but by the same token you shouldn't assume the Leafs shooting and save ability will either, especially when we know teams have more control over the former. The Leafs were also -7 on shot differential last year, so they have 65 games of history in being bad in that department.

 

[Feed Me A Stray Cat]

By 3... The kings have a worse ratio and the bruins have only 3 more shots for than against... That's really what you took from that?

Yes. I guess the ratio is interesting because it shows the Leafs system creates a higher proportion of quality shots than other teams, but in regards to whether the Leafs will maintain their pace the ratio doesn't matter that much.

What matters is that the Leafs are still getting outshot from inclose, while getting trumped in shots from the outside.

That site is very cool though.

 

[Delicious Dangles*]

The Leafs are still getting out shot in close. If you're getting outshot by 130 from a distance (because some of those pucks will go in), you better be outshooting your opponents from in close, and the Leafs are not.

Yet with superior goaltending, high-end offensive players, and good special teams, which the Leafs have all of, you can more than overcome the negligible difference in in-close shots.

This was to show that this "massive outshooting" that is taking place is a lot less relevant than it is being made out to be, because the majority of the "outshooting" is taking place in the low-quality areas.

 

[Delicious Dangles*]

1. How do teams score goals? They must take shots, and then they must convert on them. We've established that the Leafs are bad at the first part. In regards to the latter, we know that teams simply don't maintain 11%+ shooting percentages over the long haul, so that will most likely regress. As a result, the Leafs' offense will suffer unless they start generating more shots.
2. It's not irrelevant, it's part of the picture, even if it doesn't fully explain the picture, but if you want to cast it aside I can't stop you.
3. " "
4. In regards to early leads - the Leafs corsi/fenwick is terrible in close game/tied situations as well. And it seems like you think the Leafs have a special ability to score clutch goals - I personally don't believe in clutch, as it's been more or less disproven in sports such as baseball where the data is good enough to get an objective picture.
5. Nothing about Bernier's and Reimer's history leads me to believe they're elite. Perhaps they will show that they are over the next couple seasons, but as it stands that's an extraordinary assumption.
6. http://objectivenhl.blogspot.com/2010/04/repeatability-of-special-teams.html
Powerplay shooting percentage has almost no season to season correlation. Guess what does? Shot generation!
7. No use arguing this point, because despite the fact that we now have data that shows the Leafs getting outshot from 10 feet, you will simply ignore it because we don't know the specifics of each shot.
8. 65 games is not a very large sample. Plenty of teams lucked their way in S% and/or Sv% over a full season only to regress. You're correct that I simply shouldn't assume that the 10 delta will persist the full season on shot differential, but by the same token you shouldn't assume the Leafs shooting and save ability will either, especially when we know teams have more control over the former. The Leafs were also -7 on shot differential last year, so they have 65 games of history in being bad in that department.

1. We have not established that the Leafs are bad at the first part. Being more selective in shooting does not mean you are worse at shooting. Also, we do not know that teams cannot sustain an 11% shooting percentage. It is entirely within the realm of possibility. There is also no reason to think that the shot specifics will stay the same if the goal-scoring specifics change.
2. It is part of the picture, but pretty much irrelevant unless the whole picture is being looked at.
3. It is part of the picture, but pretty much irrelevant unless the whole picture is being looked at.
4. Baseball statistics is a terrible way to prove the existence of clutch in hockey. Whether or not clutchness exists, this was to show that there are an endless number of factors that can affect these statistics in hockey.
5. Is it the fact that they've put up top numbers since entering the league, or the fact that they have dominated lower leagues, that makes you skeptical?
6. This shows nothing that suggests that teams are unable to consistently post good or bad percentages from year to year.
7. The difference is negligible and shots from 10 feet or closer are not automatically scoring chances, or of equal quality.
8. I am not assuming anything. I am simply watching the games, analyzing the results, and analyzing WHY the results are what they are. That last part is what a lot of people forget to do.

 

[The Mighty Duck Man]

Argument against "style of play" is that history tells us it's very rare and usually a one-off if a team does as well as the Leafs appear to and so drastically get out-shot. That's why you hear analysts saying "Have the Leafs found a way to defy Corsi?", it's because there is no historical precedence for this being a thing that generally happens. If there was a "style of play" where you simply get less shots, let the other team have lots of shots, and still win consistently, then shot differential would be all over the board from multiple teams employing those systems, not so consistent.

Also, IIRC, Carlyle's winning Ducks' teams all had good shot differential as well.

MrVisser covered most of it, but I'd like to add that the pro Leafs argument hinges on one fact: that the Leafs create better/more scoring opportunities than their opposition. If the Leafs aren't doing this, there is no way that their team performance could be construed as good.

Based on the research presented above, the Leafs are giving up more shots close to the net than they take, which doesn't bode well for the pro Leafs argument (although it doesn't negate it, since we don't know the true quality of these close shots).

Overall, it will be very hard to outchance the opposition when you're getting outshot 36-26 on a nightly basis. While 70% of those extra shots might be trash, a couple will be good, and even some of the trash shots will find their way in over the long run.

I'm not really defending the Leafs a whole lot, just making an observation. CORSI does reward a certain style of play is all I'm saying. Are the Leafs better than CORSI makes them out to be? I do think so. Their collapsing style doesn't allow for a lot of great chances and they have enough talent to make it work offensively. However, I can't help but agree with you guys that they aren't a great hockey team. Even considering style of play, the shot differential is insane, and IMO they simply aren't playing great hockey. So while they're better than CORSI indicates, I'd also say they're definitely worse than their record indicates, and coming back down to earth is inevitable.

Also, advanced data doesn't go back far enough, but since they won the cup, Carlyle's Ducks had pretty even shot differentials and were middle of the pack in regards to other advanced stats during the good years, and they fell off the map after that. A lot of that had to do with goaltending(their CORSI wasn't bad with JSG in net because he hated his players blocking shots, whereas Hiller likes it and they needed something to replace Pronger) and obviously losing two HHoF defensemen, but even with those talented teams in 2008 and 2009, they weren't much of a puck possession squad, and a lot of that is Carlyle.

 

[Feed Me A Stray Cat]

1. We have not established that the Leafs are bad at the first part. Being more selective in shooting does not mean you are worse at shooting. Also, we do not know that teams cannot sustain an 11% shooting percentage. It is entirely within the realm of possibility. There is also no reason to think that the shot specifics will stay the same if the goal-scoring specifics change.
2. It is part of the picture, but pretty much irrelevant unless the whole picture is being looked at.
3. It is part of the picture, but pretty much irrelevant unless the whole picture is being looked at.
4. Baseball statistics is a terrible way to prove the existence of clutch in hockey. Whether or not clutchness exists, this was to show that there are an endless number of factors that can affect these statistics in hockey.
5. Is it the fact that they've put up top numbers since entering the league, or the fact that they have dominated lower leagues, that makes you skeptical?
6. This shows nothing that suggests that teams are unable to consistently post good or bad percentages from year to year.
7. The difference is negligible and shots from 10 feet or closer are not automatically scoring chances, or of equal quality.
8. I am not assuming anything. I am simply watching the games, analyzing the results, and analyzing WHY the results are what they are. That last part is what a lot of people forget to do.

1. No team has been able to sustain a 10%+ ES shooting percentage season to season for a very long time. If you think the Leafs have cracked the code of doing so based on a 65 game sample, that's your prerogative.
4. It's my own personal belief that clutch doesn't exist, which is confirmed by the fact that it's been refuted every time in other sports. The general notion of clutch ability is the same across all sports - that certain players/teams are better at the game in certain high pressure situations.
5. Jonathan Bernier and James Reimer's save percentages since entering the league are .917, which is only a couple percentage points above what the median save percentage has been over that time. Hardly "top", and certainly not elite. If you want to talk in those terms, look at what guys like Rask and Schneider have done over similar GP. If their save percentages revert to their career average, the Leafs are in big trouble.
6. Yes it does. If teams had any type of skill in powerplay shooting percentage, it would be shown by a correlation coefficient greater than .06.
7. It doesn't matter that the difference is negligible - the point is that the Leafs should be outshooting their opponents in the <10ft category by a decent margin to justify the argument that their scoring chance centric style of play is what explains their 11-6 record, because they're heavily in the red on shots outside 10ft.
8. You're assuming that the Leafs have engineered a style of play that leads to results that no team in the NHL has seen in a decade, possibly longer. I'm going to err on the side of recent history, and assume that the Leafs are more similar to the 09-10 Colorado Avalanche than an anomaly we've yet to see.

 

[hatterson]

Quick note regarding Reimers save percentage, the last 2 articles in the "Shot Quality Project" series on sportsnet actually address that and offer some interesting viewpoints.

http://www.sportsnet.ca/hockey/maple-leafs-reimer-showing-unique-ability/
http://www.sportsnet.ca/hockey/reimers-rebounds-no-big-deal-for-maple-leafs/

 

[TieClark]

Yes. I guess the ratio is interesting because it shows the Leafs system creates a higher proportion of quality shots than other teams, but in regards to whether the Leafs will maintain their pace the ratio doesn't matter that much.

What matters is that the Leafs are still getting outshot from inclose, while getting trumped in shots from the outside.

That site is very cool though.

What it shows is they're right with the big boys when it comes to in close shots, but the Corsi is effected drastically due to the large amounts of long range shots they both allow and don't take themselves

 

[sharks9]

What it shows is they're right with the big boys when it comes to in close shots, but the Corsi is effected drastically due to the large amounts of long range shots they both allow and don't take themselves

So would you say Reimer and Bernier have inflated numbers because they face more low percentage shots than the average goalie?

 

[Delicious Dangles*]

1. No team has been able to sustain a 10%+ ES shooting percentage season to season for a very long time. If you think the Leafs have cracked the code of doing so based on a 65 game sample, that's your prerogative.
4. It's my own personal belief that clutch doesn't exist, which is confirmed by the fact that it's been refuted every time in other sports. The general notion of clutch ability is the same across all sports - that certain players/teams are better at the game in certain high pressure situations.
5. Jonathan Bernier and James Reimer's save percentages since entering the league are .917, which is only a couple percentage points above what the median save percentage has been over that time. Hardly "top", and certainly not elite. If you want to talk in those terms, look at what guys like Rask and Schneider have done over similar GP. If their save percentages revert to their career average, the Leafs are in big trouble.
6. Yes it does. If teams had any type of skill in powerplay shooting percentage, it would be shown by a correlation coefficient greater than .06.
7. It doesn't matter that the difference is negligible - the point is that the Leafs should be outshooting their opponents in the <10ft category by a decent margin to justify the argument that their scoring chance centric style of play is what explains their 11-6 record, because they're heavily in the red on shots outside 10ft.
8. You're assuming that the Leafs have engineered a style of play that leads to results that no team in the NHL has seen in a decade, possibly longer. I'm going to err on the side of recent history, and assume that the Leafs are more similar to the 09-10 Colorado Avalanche than an anomaly we've yet to see.

1. That's because most teams employ a shoot-shoot-shoot! strategy, which lowers their shooting percentages. Just because something has not happened in a while, it does not mean it will never happen. I do believe this is the 2nd year this is happening.
4. Using baseball is a terrible way to draw that conclusion. There is pretty much no comparison to be made. They are entirely different types of games. However, whether or not clutchness or momentum exists is irrelevant. This is about how there are a multitude of factors that influence the data sets that are not accounted for.
5. Both are among the youngest goaltenders in the league, and one has been a backup for his whole career so far. Reimer's average is also pushed down by his injury-filled season where he had lingering effects.

Rask and Schneider played for defensive powerhouses, and are pretty good goalies as well.

There is nothing to suggest that Reimer and Bernier cannot keep up the play they have shown so far. Nothing they have shown this year is something we haven't seen previously from them. It's not like this was unexpected of Bernier either. Lord knows we heard enough of how awesome he was before he became a Leaf.
6. Looking at PPs, since the lockout, let's look at a few teams:

Montreal: 5th, 1st, 1st, 2nd, 7th, 5th, 2nd
Washington: 8th, 2nd, 1st, 1st, 1st
Philly: 2nd, 6th, 3rd, 6th, 3rd
Pittsburgh: 6th, 5th, 4th, 5th, 7th, 2nd
Anaheim: 3rd, 5th, 5th, 3rd, 4th
San Jose: 2nd, 3rd, 4th, 2nd, 2nd, 7th, 8th
Detroit: 1st, 3rd, 1st, 9th, 5th

Yeah, looks real random.

Also, looking over the data, teams tend to have 2-3 year stretches (at least) in similar territories/percentages.
7. We don't have to be outshooting anybody from in close either, if we have good goaltending, special teams, offensive players who can convert, and higher quality shots from inside. A comparison with supposed "top teams" was given, and many are in similar situations in regards to up-close shots. Why are the Leafs being singled out?

When you take out the easily-saveable perimeter shots, we are pretty much the same as everybody else.
8. You do that.

 

[Delicious Dangles*]

So would you say Reimer and Bernier have inflated numbers because they face more low percentage shots than the average goalie?

Yeah, probably a little bit. That still doesn't diminish what they've done for the team or them as players, and they still have to make all the saves, but they've had more "easy saves" than your average team.

 

[BoredBrandonPridham]

I'm not really defending the Leafs a whole lot, just making an observation. CORSI does reward a certain style of play is all I'm saying. Are the Leafs better than CORSI makes them out to be? I do think so. Their collapsing style doesn't allow for a lot of great chances and they have enough talent to make it work offensively. However, I can't help but agree with you guys that they aren't a great hockey team. Even considering style of play, the shot differential is insane, and IMO they simply aren't playing great hockey. So while they're better than CORSI indicates, I'd also say they're definitely worse than their record indicates, and coming back down to earth is inevitable.

Also, advanced data doesn't go back far enough, but since they won the cup, Carlyle's Ducks had pretty even shot differentials and were middle of the pack in regards to other advanced stats during the good years, and they fell off the map after that. A lot of that had to do with goaltending(their CORSI wasn't bad with JSG in net because he hated his players blocking shots, whereas Hiller likes it and they needed something to replace Pronger) and obviously losing two HHoF defensemen, but even with those talented teams in 2008 and 2009, they weren't much of a puck possession squad, and a lot of that is Carlyle.

Whoa, hey, I was just explaining the "style of play" argument because it wasn't acknowledged in the post I quoted. I'm not dissing the Leafs!

 

[TieClark]

So would you say Reimer and Bernier have inflated numbers because they face more low percentage shots than the average goalie?

Yes actually and I said it last rust as well. That doesn't mean they both haven't been very good, but I don't think they're better than Lundqvist despite their stats being better either.

 

[HarrisonFord]

1. How do teams score goals? They must take shots, and then they must convert on them. We've established that the Leafs are bad at the first part. In regards to the latter, we know that teams simply don't maintain 11%+ shooting percentages over the long haul, so that will most likely regress. As a result, the Leafs' offense will suffer unless they start generating more shots.
2. It's not irrelevant, it's part of the picture, even if it doesn't fully explain the picture, but if you want to cast it aside I can't stop you.
3. " "
4. In regards to early leads - the Leafs corsi/fenwick is terrible in close game/tied situations as well. And it seems like you think the Leafs have a special ability to score clutch goals - I personally don't believe in clutch, as it's been more or less disproven in sports such as baseball where the data is good enough to get an objective picture.
5. Nothing about Bernier's and Reimer's history leads me to believe they're elite. Perhaps they will show that they are over the next couple seasons, but as it stands that's an extraordinary assumption.
6. http://objectivenhl.blogspot.com/2010/04/repeatability-of-special-teams.html
Powerplay shooting percentage has almost no season to season correlation. Guess what does? Shot generation!
7. No use arguing this point, because despite the fact that we now have data that shows the Leafs getting outshot from 10 feet, you will simply ignore it because we don't know the specifics of each shot.
8. 65 games is not a very large sample. Plenty of teams lucked their way in S% and/or Sv% over a full season only to regress. You're correct that I simply shouldn't assume that the 10 delta will persist the full season on shot differential, but by the same token you shouldn't assume the Leafs shooting and save ability will either, especially when we know teams have more control over the former. The Leafs were also -7 on shot differential last year, so they have 65 games of history in being bad in that department.

As an economics major, I LOVE this argument. Here's a few questions for you, tell me if you can answer them.

1. How likely are they to regress? 80%? 85%? The league wide shooting percentage last season was 9.11%. The standard deviation of this is 1.11% If we are using the flawed assumptions that are used, then there should have been 5% of teams (1 team is 3.33% of the league; so one team at least) with a shooting percentage above 11.33%. And in fact there was only one. The Leafs. So the fact that there is a team there above 11% shouldn't be surprising. And as for the sustainability of it, if you look at the highest shooting percentage teams over the last 4 seasons, you start seeing the same teams up there. For example

Tampa Bay Lightning:

2011 shooting % - 10.42% (Ranked 1st)
2012 shooting % - 11.11% (Ranked 3rd)
2013 shooting % - 10.47% (Ranked 4th)

If we are treating the reliability of corsi, and regression of all teams to the same mean, using randomization as being one of the primary underlying assumptions, then either they somehow managed to be lucky enough to shoot which statistically shows them being (3.129% likely in 2011, 3.593% likely in 2012, and 9.342% likely this season), or collectively using league wide averages, there was a 0.01% chance of them shooting that well. Or do you think there's possibly some sort of underlying correlation between the three seasons?

Two other examples of teams with high shooting percentages over the last 4 or 5 seasons are the Capitals and Penguins.

That brings me to question/point number 2.

2) While I understand that every team has a mean to regress to, not every team has the same mean shooting percentage to regress to. Due to certain factors (player skill, team offensive systems, etc.), some teams (Pittsburgh, Washington, TB, maybe the Leafs?) have higher mean shooting percentages than other teams. Therefore, you cannot expect that the Leafs "natural shooting percentage" is the league average, because that's simply not the case.

 

[Ho Borvat]

I think the problem with Corsi Debate is that people forget that its a stat, no different from (lets say) save %.

Your team can get sub .900 goaltending and still win hockey games the same way you can have a worse corsi than the other team and still win games.

Your a lot better off if you can get .930 goaltending, the same way your better off if you outshoot your opponents.

 

[Master_Of_Districts]

As an economics major, I LOVE this argument. Here's a few questions for you, tell me if you can answer them.

1. How likely are they to regress? 80%? 85%? The league wide shooting percentage last season was 9.11%. The standard deviation of this is 1.11% If we are using the flawed assumptions that are used, then there should have been 5% of teams (1 team is 3.33% of the league; so one team at least) with a shooting percentage above 11.33%. And in fact there was only one. The Leafs. So the fact that there is a team there above 11% shouldn't be surprising. And as for the sustainability of it, if you look at the highest shooting percentage teams over the last 4 seasons, you start seeing the same teams up there. For example

Tampa Bay Lightning:

2011 shooting % - 10.42% (Ranked 1st)
2012 shooting % - 11.11% (Ranked 3rd)
2013 shooting % - 10.47% (Ranked 4th)

If we are treating the reliability of corsi, and regression of all teams to the same mean, using randomization as being one of the primary underlying assumptions, then either they somehow managed to be lucky enough to shoot which statistically shows them being (3.129% likely in 2011, 3.593% likely in 2012, and 9.342% likely this season), or collectively using league wide averages, there was a 0.01% chance of them shooting that well. Or do you think there's possibly some sort of underlying correlation between the three seasons?

Two other examples of teams with high shooting percentages over the last 4 or 5 seasons are the Capitals and Penguins.

That brings me to question/point number 2.

2) While I understand that every team has a mean to regress to, not every team has the same mean shooting percentage to regress to. Due to certain factors (player skill, team offensive systems, etc.), some teams (Pittsburgh, Washington, TB, maybe the Leafs?) have higher mean shooting percentages than other teams. Therefore, you cannot expect that the Leafs "natural shooting percentage" is the league average, because that's simply not the case.

Overall shooting percentage is confounded by empty net goals, team differences in powerplay shooting percentage, and the fact that some teams take a higher proportion of shots on the powerplay than others.

If we look at even strength shooting percentage specifically, it's clear that, while some teams have more shooting talent than others, the talent differences between teams are far, far smaller than what's commonly assumed.

The interyear correlation at the team level is about ~0.10. Offseason roster turnover is, of course, a confounding factor, but the split-half reliability - from one half of the season to the other - is scarcely higher (about ~0.15). By comparison, the split-half reliability for Corsi is on the order of 0.80 to 0.90.

In fact, the skill component to even strength shooting percentage is calculable. The skill standard deviation is about 0.45%. Given that the distribution is approximately normal, a team at the 98th percentile would be expected to shoot 9.0% over the long run.

That, of course, is not trivial, but to argue that even strength shooting percentages of 10% or 11% are sustainable just doesn't hold any water.

 

[Feed Me A Stray Cat]

As an economics major, I LOVE this argument. Here's a few questions for you, tell me if you can answer them.

1. How likely are they to regress? 80%? 85%? The league wide shooting percentage last season was 9.11%. The standard deviation of this is 1.11% If we are using the flawed assumptions that are used, then there should have been 5% of teams (1 team is 3.33% of the league; so one team at least) with a shooting percentage above 11.33%. And in fact there was only one. The Leafs. So the fact that there is a team there above 11% shouldn't be surprising. And as for the sustainability of it, if you look at the highest shooting percentage teams over the last 4 seasons, you start seeing the same teams up there. For example

Tampa Bay Lightning:

2011 shooting % - 10.42% (Ranked 1st)
2012 shooting % - 11.11% (Ranked 3rd)
2013 shooting % - 10.47% (Ranked 4th)

If we are treating the reliability of corsi, and regression of all teams to the same mean, using randomization as being one of the primary underlying assumptions, then either they somehow managed to be lucky enough to shoot which statistically shows them being (3.129% likely in 2011, 3.593% likely in 2012, and 9.342% likely this season), or collectively using league wide averages, there was a 0.01% chance of them shooting that well. Or do you think there's possibly some sort of underlying correlation between the three seasons?

Two other examples of teams with high shooting percentages over the last 4 or 5 seasons are the Capitals and Penguins.

That brings me to question/point number 2.

2) While I understand that every team has a mean to regress to, not every team has the same mean shooting percentage to regress to. Due to certain factors (player skill, team offensive systems, etc.), some teams (Pittsburgh, Washington, TB, maybe the Leafs?) have higher mean shooting percentages than other teams. Therefore, you cannot expect that the Leafs "natural shooting percentage" is the league average, because that's simply not the case.

I was also an economics major and appreciate your nuanced view of this topic.

Master of Districts basically has it covered.

I agree with your last point - that the Leafs might have a higher baseline shooting percentage than the NHL average. However, their baseline shooting percentage won't be 10-11% at even strength, which is what they've managed over the last 65 games.

 

[Feed Me A Stray Cat]

1. That's because most teams employ a shoot-shoot-shoot! strategy, which lowers their shooting percentages. Just because something has not happened in a while, it does not mean it will never happen. I do believe this is the 2nd year this is happening.
4. Using baseball is a terrible way to draw that conclusion. There is pretty much no comparison to be made. They are entirely different types of games. However, whether or not clutchness or momentum exists is irrelevant. This is about how there are a multitude of factors that influence the data sets that are not accounted for.
5. Both are among the youngest goaltenders in the league, and one has been a backup for his whole career so far. Reimer's average is also pushed down by his injury-filled season where he had lingering effects.

Rask and Schneider played for defensive powerhouses, and are pretty good goalies as well.

There is nothing to suggest that Reimer and Bernier cannot keep up the play they have shown so far. Nothing they have shown this year is something we haven't seen previously from them. It's not like this was unexpected of Bernier either. Lord knows we heard enough of how awesome he was before he became a Leaf.
6. Looking at PPs, since the lockout, let's look at a few teams:

Montreal: 5th, 1st, 1st, 2nd, 7th, 5th, 2nd
Washington: 8th, 2nd, 1st, 1st, 1st
Philly: 2nd, 6th, 3rd, 6th, 3rd
Pittsburgh: 6th, 5th, 4th, 5th, 7th, 2nd
Anaheim: 3rd, 5th, 5th, 3rd, 4th
San Jose: 2nd, 3rd, 4th, 2nd, 2nd, 7th, 8th
Detroit: 1st, 3rd, 1st, 9th, 5th

Yeah, looks real random.

Also, looking over the data, teams tend to have 2-3 year stretches (at least) in similar territories/percentages.
7. We don't have to be outshooting anybody from in close either, if we have good goaltending, special teams, offensive players who can convert, and higher quality shots from inside. A comparison with supposed "top teams" was given, and many are in similar situations in regards to up-close shots. Why are the Leafs being singled out?

When you take out the easily-saveable perimeter shots, we are pretty much the same as everybody else.
8. You do that.

"Nothing they've shown this year is something we haven't seen before." Get real dude. They are posting .940 save percentages. They have never showed that level of ability, nor has any other goaltender in NHL history over an adequate sample.

I didn't say the totality of a powerplay performance was random, I said powerplay shooting percentage was. The real skill is shot generation on the PP.

I'd be interested in seeing where those teams finished in PP shooting percentage each season.

 

[Delicious Dangles*]

"Nothing they've shown this year is something we haven't seen before." Get real dude. They are posting .940 save percentages. They have never showed that level of ability, nor has any other goaltender in NHL history over an adequate sample.

I didn't say the totality of a powerplay performance was random, I said powerplay shooting percentage was. The real skill is shot generation on the PP.

I'd be interested in seeing where those teams finished in PP shooting percentage each season.

You're looking solely at save percentages to determine if Reimer and Bernier can keep up their play. I am looking at play to determine if they can keep up their play.

Bernier has already put like 3 goals in his own net (which I doubt will be sustained every 18 games), and Reimer was actually probably a little better last year (maybe because he actually got consistent starts?). Their save percentage is inflated because they've had a whole bunch of easy saves to go along with it.

If you want to look at PP shooting percentage, be my guest. I can definitely see PP shot counts from year to year being random, but that really has no bearing on how good the actual powerplay is. Thus, good special teams CAN be sustained.

 

[HarrisonFord]

Overall shooting percentage is confounded by empty net goals, team differences in powerplay shooting percentage, and the fact that some teams take a higher proportion of shots on the powerplay than others.

If we look at even strength shooting percentage specifically, it's clear that, while some teams have more shooting talent than others, the talent differences between teams are far, far smaller than what's commonly assumed.

The interyear correlation at the team level is about ~0.10. Offseason roster turnover is, of course, a confounding factor, but the split-half reliability - from one half of the season to the other - is scarcely higher (about ~0.15). By comparison, the split-half reliability for Corsi is on the order of 0.80 to 0.90.

In fact, the skill component to even strength shooting percentage is calculable. The skill standard deviation is about 0.45%. Given that the distribution is approximately normal, a team at the 98th percentile would be expected to shoot 9.0% over the long run.

That, of course, is not trivial, but to argue that even strength shooting percentages of 10% or 11% are sustainable just doesn't hold any water.

How can you ignore the stats from the PP though? And when we are talking about 10% and 11% shooting percentages, we ARE talking about percentages that include PP shooting percentages. I mean, if you're trying to compare teams and the relativity of shooting percentages as a measure of probable success, you HAVE to include PP stats because the power play is still a big part of this game.

And I think that if you're talking about differences in talent between teams being negligible then you're making a mistake in judgement. I haven't gone and looked (but will tomorrow), but looking at the spread of shot attempts by team (how many players have taken shots), then you can sort of "inflate your shooting % by design as a coach. Keeping with the TB Lightning as the example, when you have Steven Stamkos (career SH% of 17.5%) taking 300+ shots a year, and a guy with a great transition chance creator in Martin St. Louis, then you can expect that TBL is going to have a higher natural shooting percentage mean than, say, a team like Buffalo this season.

I was also an economics major and appreciate your nuanced view of this topic.

Master of Districts basically has it covered.

I agree with your last point - that the Leafs might have a higher baseline shooting percentage than the NHL average. However, their baseline shooting percentage won't be 10-11% at even strength, which is what they've managed over the last 65 games.

Economics ruins the way you look at stuff eh?

And okay, we've seen that it is 65 games with the Leafs, but the baseline shooting percentage of the TBL that I brought up earlier is over 200 games. While the 10%-11% shooting percentage may not be sustainable for the Leafs for 200 games, after looking at the TBL, Pens, Caps, etc. I don't really think we can strongly argue that the Leafs cannot sustain this shooting percentage for another 80 games; as it seems that every year there are 3-5 teams who can commonly be found in the top of the league shooting percentages for a few seasons in a row.

 

[Master_Of_Districts]

How can you ignore the stats from the PP though? And when we are talking about 10% and 11% shooting percentages, we ARE talking about percentages that include PP shooting percentages. I mean, if you're trying to compare teams and the relativity of shooting percentages as a measure of probable success, you HAVE to include PP stats because the power play is still a big part of this game.

It's not ignored.

It's simply treated separately.

Powerplay shooting percentage has been analyzed as well. A similar conclusion follows - the distribution in talent is far narrower than the distribution in results. The talent component is somewhat greater than at even strength but not by much. The standard deviation in talent is well under 1%.

And I think that if you're talking about differences in talent between teams being negligible then you're making a mistake in judgement. I haven't gone and looked (but will tomorrow), but looking at the spread of shot attempts by team (how many players have taken shots), then you can sort of "inflate your shooting % by design as a coach. Keeping with the TB Lightning as the example, when you have Steven Stamkos (career SH% of 17.5%) taking 300+ shots a year, and a guy with a great transition chance creator in Martin St. Louis, then you can expect that TBL is going to have a higher natural shooting percentage mean than, say, a team like Buffalo this season.

I didn't say it was negligible. In fact, I believe I said the standard deviation in talent was "not trivial." And it isn't trivial.

But once again - the observed differences far exceed the differences in underlying talent. The spread in observed results is four times as wide as the talent distribution after 41 games. Three times as wide after 82 games. People tend to take the observed results at face value when they absolutely should not.

And another factor is that, because the impact of randomness is over twice as important as the impact in skill with respect to observed results in EV SH% over the course of the regular season, it's very difficult to distinguish the more talented teams from the less talented ones.

We know that some teams have above average shooting talent in any given season, but it's difficult to identify them with much certainty or confidence.

 

[Master_Of_Districts]

As for Tampa Bay, I'm not sure if they're the shooting percentage juggernaut that you're characterizing them as.

They're 1st in 5-on-5 SH% over the last two years (2011-12 & 2012-13 combined sample), but shot below the league average in 2010-11.

They're shooting 8.42% (barely above average) at 5-on-5 so far this year, although that'll surely increase after tonight's game.

On the PP (5-on-4), they're unremarkable - 14th in the league from 2010-13 (combined sample); 11th from 2011-13 (again, combined sample).

 

[Feed Me A Stray Cat]

And okay, we've seen that it is 65 games with the Leafs, but the baseline shooting percentage of the TBL that I brought up earlier is over 200 games. While the 10%-11% shooting percentage may not be sustainable for the Leafs for 200 games, after looking at the TBL, Pens, Caps, etc. I don't really think we can strongly argue that the Leafs cannot sustain this shooting percentage for another 80 games; as it seems that every year there are 3-5 teams who can commonly be found in the top of the league shooting percentages for a few seasons in a row.

The shooting percentage of Tampa that you referenced was their overall shooting percentage, not their ES shooting percentage.

 

[Auston Marlander]

"Nothing they've shown this year is something we haven't seen before." Get real dude. They are posting .940 save percentages. They have never showed that level of ability, nor has any other goaltender in NHL history over an adequate sample.

I didn't say the totality of a powerplay performance was random, I said powerplay shooting percentage was. The real skill is shot generation on the PP.

I'd be interested in seeing where those teams finished in PP shooting percentage each season.

And yet they are not leading the league in SV%. Will they have their SV% drop, yes probably a bit, but so will the others up top. Stop acting like they are the only 2 who are playing well.