Corsi, shot quality, and the Toronto Maple Leafs

[Master_Of_Districts]

They don't, and that is why it's flawed. It's the entire reason the Leafs look terrible in advanced statistics... they treat low quality shots as equal to high quality shots and with the Leafs system, they allow a lot of low quality shots against and generate most of their offence off the rush which means low quantity, but high quality of shots for them.

Differential shot quality may have some explanatory role, but the impact is less than you think.

For example, shot distance is, far and away, the most important component of shot quality.

In 2012-13, the average shot the Leafs took at even strength was taken from a distance of 35.6. Conversely, the average shot they conceded at even strength was taken from a distance of 37 feet.

So - in support of your point - it appears that the Leafs did, in fact, take higher quality shots than their opponents during even strength play last season. But the difference is quite small - only about 1.4 feet.

 

[TieClark]

Differential shot quality may have some explanatory role, but the impact is less than you think.

For example, shot distance is, far and away, the most important component of shot quality.

In 2012-13, the average shot the Leafs took at even strength was taken from a distance of 35.6. Conversely, the average shot they conceded at even strength was taken from a distance of 37 feet.

So - in support of your point - it appears that the Leafs did, in fact, take higher quality shots than their opponents during even strength play last season. But the difference is quite small - only about 1.4 feet.

Again, a Dion Phaneuf shot is a lot more dangerous than a Carl Gunnarsson shot. Similarly, the Leafs have several very dangerous offensive players in Kessel, JVR, Lupul, Kadri etc. There are just far too many factors to simply say a team that outshoots their opponent at even strength are generally better. It doesn't work that way.

If you watch the Leafs play, they play a system that completely contradicts the theory. They thoroughly outplayed Anaheim in period 2 and 3 and yet were outshot in the game. They sit back and completely collapse defensively, all 5 guys. Their offence comes almost exclusively off the rush other than a few sporadic shifts where they hem a team in. This results in very skewed 5 on 5 shot differentials and thus very skewed advanced statistics.

 

[TheDevilMadeMe]

Also, why not use Goal Differential as a proxy for Quality Shots For v.s. Quality Shots Against, then you might actually be talking about something that matters.

There are far more shots taken per game than goals, so you get a much better sample size with shots. Over the course of a single season, shot differential is a better predictor of future success than goal differential. However, as we all know, the goal of hockey is to score goals, not get shots, so over a large enough sample size (a period of multiple seasons), goal differential becomes superior to shot differential.

Basically, goal differential is better... but only so long as you have a big enough sample size (which is more than a single season).

 

[Rants Mulliniks]

Differential shot quality may have some explanatory role, but the impact is less than you think.

For example, shot distance is, far and away, the most important component of shot quality.

In 2012-13, the average shot the Leafs took at even strength was taken from a distance of 35.6. Conversely, the average shot they conceded at even strength was taken from a distance of 37 feet.

So - in support of your point - it appears that the Leafs did, in fact, take higher quality shots than their opponents during even strength play last season. But the difference is quite small - only about 1.4 feet.

I wouldn't call that true. You can look at where goals are scored from and see that two equal distances but different location traditionally yield wildly different results.

 

[Bedards Dad]

Forgive my ignorance (still learning the advance stats stuff) but how do the advanced stats guys differentiate Kessels snipe on the 2 on 1 vs, say, a random bad angle shot from someone like Bodie or something?

They don't and that is why many won't take it seriously. It is fine as a supportive stat with limited upside, but until shot quality is incorporated it will always be just that, a somewhat helpful secondary stat. Very little about it is advanced.

 

[Master_Of_Districts]

I wouldn't call that true. You can look at where goals are scored from and see that two equal distances but different location traditionally yield wildly different results.

I suppose shot angle is also important. Maybe even as important as distance itself.

The other components - deflections, rebounds and the like - are generally unimportant, mostly because they're rare and don't sustain at all at the team level.

But let's consider distance and angle. We can classify a shot as dangerous or non-dangerous based on the variables of distance and angle. With this approach, all shots taken from the "home plate" area in front of the goal would be classified as "dangerous." Did the Leafs take more shots from this area of the ice than their opponents last season at even strength?

This link provides an answer to that very question.

The chart discloses that the Leafs took 345 even strength shots from "dangerous" areas, whereas their opponents took 391 even strength shots from "dangerous" areas.

That's a slightly better percentage (0.468) than their overall EV fenwick/corsi percentage, but not by a lot.

So, once again, while it seems that differential shot quality had some explanatory role, it's effect seems to be overstated - and probably by a lot.

Some will surely criticize this approach on the basis that it ignores the context under which the shots were taken. But once again - we're already limiting our sample to shots taken close to the net from the middle of the offensive zone. In other words, the most dangerous subset of all shots taken. Contextual factors can only be so important.

In any event, no one has presented any evidence that such contextual factors, if accounted for, would even favor the Leafs.

 

[GuineaPig]

Forgive my ignorance (still learning the advance stats stuff) but how do the advanced stats guys differentiate Kessels snipe on the 2 on 1 vs, say, a random bad angle shot from someone like Bodie or something?

They don't. But at the same time, the inverse is true: there's no differentiation between a low-quality chance and a high-quality one.

Shot quality absolutely exists and plays a major part over small sample sizes, but every study has suggested that in the modern NHL things more or less even out over the long haul (at even strength). So as long as you're not dealing with small sample sizes, shot quality can more or less be ignored. It's a hindrance, but only a minor one (especially because one should be dealing with larger samples regardless).

 

[Znith]

The other components - deflections, rebounds and the like - are generally unimportant, mostly because they're rare and don't sustain at all at the team level.

This is a pretty ridiculous statement. Some goaltenders allow many more rebounds than others, creating many more scoring chances for the opposite teams. In fact, most scoring chances are created by rebounds and deflections!

Some teams focus on getting guys in front of the net more and getting deflections. Johan Franzen made a career of doing this. Side to side passing and one timers are far harder to stop than straight up shots.

To say that every team plays more or less the same system and has the same level of talent on its shooters is pretty ridiculous. I disagree with much of what you're saying, and challenge you to prove it.

 

[Rants Mulliniks]

I suppose shot angle is also important. Maybe even as important as distance itself.

The other components - deflections, rebounds and the like - are generally unimportant, mostly because they're rare and don't sustain at all at the team level.

But let's consider distance and angle. We can classify a shot as dangerous or non-dangerous based on the variables of distance and angle. With this approach, all shots taken from the "home plate" area in front of the goal would be classified as "dangerous." Did the Leafs take more shots from this area of the ice than their opponents last season at even strength?

This link provides an answer to that very question.

The chart discloses that the Leafs took 345 even strength shots from "dangerous" areas, whereas their opponents took 391 even strength shots from "dangerous" areas.

That's a slightly better percentage (0.468) than their overall EV fenwick/corsi percentage, but not by a lot.

So, once again, while it seems that differential shot quality had some explanatory role, it's effect seems to be overstated - and probably by a lot.

Some will surely criticize this approach on the basis that it ignores the context under which the shots were taken. But once again - we're already limiting our sample to shots taken close to the net from the middle of the offensive zone. In other words, the most dangerous subset of all shots taken. Contextual factors can only be so important.

In any event, no one has presented any evidence that such contextual factors, if accounted for, would even favor the Leafs.

If you look at the "heat map" produced in another thread compared to the location of shots in their games this year, on many occasions a far greater percentage of Leaf shots (and even totals) are in the higher quality areas. Yet people remain obsessed by shot total. One would probably question why it is that the Leafs seem to frequently be better at getting shots in those high quality areas. It would seem logical that they probably have possession in the high quality area more often. It's nice that other teams have possession in the offensive zone and all but where would you rather have posession? There or in the high quality areas?

Here's a sample of game where I was told the Leafs were "lucky". Look at the discrepancy in shots from high quality areas in the Leafs favour:

 

[Rants Mulliniks]

They don't. But at the same time, the inverse is true: there's no differentiation between a low-quality chance and a high-quality one.

Shot quality absolutely exists and plays a major part over small sample sizes, but every study has suggested that in the modern NHL things more or less even out over the long haul (at even strength). So as long as you're not dealing with small sample sizes, shot quality can more or less be ignored. It's a hindrance, but only a minor one (especially because one should be dealing with larger samples regardless).

So what's the magic number? It will soon be a full season that it's been holding true.

 

[Brock Radunske]

I'm sure it's been mentioned already, but shots from say....Kessel, JVR, Lupul, Kadri likely hold a lot more weight than shots from say...any forwards on Nashville. They may get more but it doesn't mean they'll score.

 

[overpass]

Differential shot quality may have some explanatory role, but the impact is less than you think.

For example, shot distance is, far and away, the most important component of shot quality.

In 2012-13, the average shot the Leafs took at even strength was taken from a distance of 35.6. Conversely, the average shot they conceded at even strength was taken from a distance of 37 feet.

So - in support of your point - it appears that the Leafs did, in fact, take higher quality shots than their opponents during even strength play last season. But the difference is quite small - only about 1.4 feet.

I think you mean that shot distance is the most important component of shot quality that we have data for and can analyze.

It's very likely that lateral movement of the puck just before a shot has a significant effect in shot quality. Teams are always trying to get the opposing goalie moving across the net when they shoot, while preventing cross-ice passes in their own end and allowing their goalie to be square to the shot. But we don't have freely available data on the motion of the puck during play so it hasn't been quantified (unless teams are doing this internally.)

 

[TieClark]

They don't. But at the same time, the inverse is true: there's no differentiation between a low-quality chance and a high-quality one.

Shot quality absolutely exists and plays a major part over small sample sizes, but every study has suggested that in the modern NHL things more or less even out over the long haul (at even strength). So as long as you're not dealing with small sample sizes, shot quality can more or less be ignored. It's a hindrance, but only a minor one (especially because one should be dealing with larger samples regardless).

This just isn't true and common sense should tell you that. If Alex Ovechkin is coming down on your goalie in the playoffs on a 2 on 1 you probably tense right up. If Frazer McLaren is coming down on your goalie in the same scenario, I doubt you're nearly as nervous. Yet McLaren had a shooting % of 15 while Ovie's was 14.5...

 

[Stats01]

I haven't read through this thread but I just wanted to say **** Corsi. Don't understand the recent obsession with "advanced" stats. Like Damien Cox said in one of his recent articles. The people coming out and complaining about the Leafs seemingly "lack of puck possession" seem to forget that "puck possession" has been around forever. This is nothing new folks, call it what you want whether the Leafs are some how lucky or whatever, what I see is a one hell of a talented team that can score at will. If you make a mistake you will pay for it.

 

[Stats01]

I'm really starting to see a trend where people are starting to complain even after their team wins a game because the corsi stats don't make it seem like a "pretty" win or a "well deserved" win. It's total B.S. I'm not going to let someone on here rain on my mood when the Leafs win. Bring up Corsi all you want you'll just get ignored.

 

[Bedards Dad]

I'm really starting to see a trend where people are starting to complain even after their team wins a game because the corsi stats don't make it seem like a "pretty" win or a "well deserved" win. It's total B.S. I'm not going to let someone on here rain on my mood when the Leafs win. Bring up Corsi all you want you'll just get ignored.

Exactly!

There is a thread on the Canucks board where people are head over heels about their Corsi and completely ignoring the fact they are just slightly above .500 (5-4-1). They can have all the corsi victories they want, but I'll stick with the 2 points.

 

[Bear of Bad News]

Exactly!

There is a thread on the Canucks board where people are head over heels about their Corsi and completely ignoring the fact they are just slightly above .500 (5-4-1). They can have all the corsi victories they want, but I'll stick with the 2 points.

I'm not going to read this thread to you post-by-post, but you're missing the point.

 

[Bedards Dad]

I'm not going to read this thread to you post-by-post, but you're missing the point.

Corsi is trying to find a way to simplify the game and predict what will happen. The issue? The game is far to dynamic for that. This isn't baseball, and shouldn't be treated like such, there's no money ball in hockey.

 

[Bear of Bad News]

there's no money ball in hockey.

Why? Because you choose not to look for it?

Of course there is. The fact that CORSI isn't a perfect predictor doesn't mean that "Moneyball" techniques don't exist in hockey. "Moneyball" techniques refer to finding skills or abilities that are undervalued (and underpriced) in the sport. To say that it doesn't exist in hockey is to say that hockey economics are in perfect equilibrium right now. Do you really believe that? (If you say "yes", I have a follow-up question: Really?)

The smart teams are using methods already (and they're also smart enough not to publish a book about them).

Claiming that there aren't arbitrage situations in hockey certainly makes the game simpler for those who won't (or can't) look for those situations. But the situations are certainly there.

 

[Master_Of_Districts]

If you look at the "heat map" produced in another thread compared to the location of shots in their games this year, on many occasions a far greater percentage of Leaf shots (and even totals) are in the higher quality areas. Yet people remain obsessed by shot total. One would probably question why it is that the Leafs seem to frequently be better at getting shots in those high quality areas. It would seem logical that they probably have possession in the high quality area more often. It's nice that other teams have possession in the offensive zone and all but where would you rather have posession? There or in the high quality areas?

Here's a sample of game where I was told the Leafs were "lucky". Look at the discrepancy in shots from high quality areas in the Leafs favour:

The link I posted allows one to view the locations of *all* shots taken during the 2012-13 season - and other seasons on top of that.

And yet you're presenting - and relying on - a screen shot from a single game.

Why?

 

[Master_Of_Districts]

And for all the talk about how the Leafs are apparently "defying advanced stats," they're only outscored the opposition by one goal during 5-on-5 play thus far.

One goal!

http://stats.hockeyanalysis.com/teamstats.php?db=201314&sit=5v5&disp=1&sortdir=DESC&sort=GFPCT

And yet that's enough to call the utility of possession based metrics into question.

Based on some of the commentary in this thread, you would think that the Leafs were the incarnation of the 1980s Oilers. You know, a team that actually *could* score at will and *did* defy the percentages and *was* an elite team for an extended period of time.

 

[TieClark]

The link I posted allows one to view the locations of *all* shots taken during the 2012-13 season - and other seasons on top of that.

And yet you're presenting - and relying on - a screen shot from a single game.

Why?

Because it's a perfect example of why it's flawed. You can't take the Leafs vs the Average and claim it makes sense because the Leafs aren't a super team. There are teams better than the leafs and teams worse than the leafs. The argument is that the Leafs aren't nearly as bad as Corsi indicates because of the shot quality vs quantity. Nobody is saying Toronto>every team in the league, but rather Toronto>>>What their Corsi indicates.

Also to add... it also ignores WHO is taking the shots and from where which is just about as significant as it gets.

 

[TieClark]

And for all the talk about how the Leafs are apparently "defying advanced stats," they're only outscored the opposition by one goal during 5-on-5 play thus far.

One goal!

http://stats.hockeyanalysis.com/teamstats.php?db=201314&sit=5v5&disp=1&sortdir=DESC&sort=GFPCT

But why does that matter? An average team at 5 on 5 with great special teams is still going to finish the season with above average results.

 

[Master_Of_Districts]

But why does that matter? An average team at 5 on 5 with great special teams is still going to finish the season with above average results.

TOR had an average powerplay (5-on-4) last year with respect to scoring rate.

They were one of the worst teams with respect to shot generation, which is the best predictor of future success on the powerplay.

(Just like at even strength! It's amazing the way the world works sometimes).

This year, their shot generation has improved somewhat, but it's still early, and they're only middling in this regard in any event.

Their success has been buoyed by a high shooting percentage. Of course, as smart people have known for a while, the correlation between a team's shooting percentage on the powerplay over a ten game sample and its underlying shooting talent is virtually zero.

So that doesn't really bode well either.

As for the penalty kill, the results were great last year, both in terms of shots against and save percentage. Full marks.

But this year they've just been getting flat out bombed in terms of shots against - 5th worst in the league, in fact.

Given how good they were at shot prevention last year, I suspect that they'll improve in this regard.

And they'll have to - because the logician in me tells me that their goalies may not continue to save over 94% of the shots they face on the PK.

Unless, of course, they're both four standard deviations more talented than Dominik Hasek in his prime. (Not likely!).

Anyway, all of that's to say that your characterization of the Leafs' special teams as "great" doesn't really hold any water. At all.

 

[Delicious Dangles*]

I don't know how it's so hard for people to understand that 4 potential PPG players, some of the best forward depth in the league, and an offensively minded defence, mixed with a system that emphasizes shot quality over quantity in the offensive zone, and boxing out players for higher quantity of low quality shots on defence, will result in an outlier for a statistic that only looks at quantity and uses metrics that assume all teams and players are the same.

Just because good teams tend to have good corsi does not mean that teams are good because they have good corsi. And even if it did, that does not mean that having a good team without good corsi is impossible.

Lots of statistics majors in here improperly using statistics.

A single statistic should never be used to make projections, let alone one with such glaring problems and omissions.