Corsi, shot quality, and the Toronto Maple Leafs

[Fourier]

The assertion that the role of binomial variance is invoked solely or primarily to excuse the imperfect correlation between predictive metrics and future results is laughable - to the point where no reasonable person would take the assertion - or the person making the assertion - seriously.

It would seem that you may have missed the point.

There is ample reason to be skeptical of the type of analysis that arises in this forum, in particular when it is applied to a specific circumstance such as this. That is not to say that there is no value to be found or that the discussion is not one of substance. When I read the posts that you have responded to for me the core issue has little to do with the specific mathematical manipulation being used but much more so this can be seen as debate about how valid the overall analysis might be when restricted to a circumstance that might well have rather unique characteristics.

The post I responded to included a list of individuals and their credentials with a clear implication that being "smart" made their opinions somehow beyond reproach. This is clearly not how things work in the academic world. And you suggested that these individuals, who by the way I respect to the degree that I have read their work, would have some absolute stance relative to the debate. Having read material from several of these individuals, I am less sure than you of their unwillingness to bend on the debate. But I also think that it is poor form to suggest of others an intellectual rigidity that they may not in fact appreciate.

 

[hatterson]

We know why though. The timing makes it very obvious. An incredibly tough stretch of games (hardest all year, and probably in recent memory), mixed with a ton of simultaneous injuries/suspensions/absences for personal reasons that affected every line. I'm sure the cameras invading their personal and work environments for a month+ didn't help much either.

I don't understand how people can look at the circumstances and not understand why goals dropped.

In fact, a lot of fluctuations in this league can probably be explained by the circumstances of the respective teams at those points.

Perhaps I wasn't clear enough in the point I was making regarding the graph as I was mentally referencing previous graphs, specifically the difference between conversion rates at 5v5 close and all even strength situations.

Here's the two charts (one from the previous page):

Not that those use different axis heights so they don't line up perfectly. However here's a direct comparison.

As you can see, from roughly game 12-13 until about game 30 or so, their goals for percentage in 5v5close situations was significantly higher (10-15%) than their goals for percentage in generic EV situations despite their shots for percentages being very close (never more than a couple percent off). The same thing repeats a bit, although not anywhere near as significant, from games ~45 until 55.

That implies one of four things:

1.) The Leafs were getting lucky in 5v5close situations
2.) The Leafs were getting unlucky in non 5v5close even strength situations.
3.) The Leafs decided to shot or goaltend worse in non 5v5close even strength situations.
4.) The Leafs somehow changed their system to limit their own quality chances or give up more quality chances to the opposition without measurably changing their shots for percentage.

#4 seems absurd to me, willingly playing a worse system (but only for a stretch of 15 games) during certain phases of the game would be crazy. #3 also seems absurd, albeit a little less so, since choosing to play worse is about as strange an idea you'll find.

#2 seems unlikely given that their performance was still significantly above expected (based on shots), which leaves #1.

Note, that I'm not saying the Leafs *can't* have a goals for rate above their shots for rate on a sustainable basis. It seems reasonable that skilled teams can do that. What I find crazy is to have a goals for rate 20% above shots for rate.

I'm willing to accept a team with elite shooters/goaltending turning an average amount of shots into an elite amount of goals, or a below average amount of shots into an above average amount of goals. I'm not willing to accept, without compelling evidence, that a team can sustainably turn a terrible amount of shots into an elite amount of goals.

 

[Delicious Dangles*]

Perhaps I wasn't clear enough in the point I was making regarding the graph as I was mentally referencing previous graphs, specifically the difference between conversion rates at 5v5 close and all even strength situations.

I'm not exactly sure why you think this is relevant. Are you telling me that for all other teams, 5v5 close GF and EV GF are exactly the same throughout all periods of the season?

We have stretches above and stretches below. Yes, perhaps the stretch above is a little longer, but why is this unusual to you?

What I take from this graph is that we tended to score more goals when the score was close than when we were out of it or when we had a safe lead.

One would expect these results, no? It seems logical that teams would more often score when they have a chance to win than when it is out of their reach or they are trying to hold a lead. It seems even more logical when you watch how the Leafs play.

The only difference I could imagine you are alluding to would be that shots when close aren't equally higher, but we're talking about the Leafs here. Our strategy when down is not "get more shots on net", as it is for most teams. In fact, we tend to do the opposite in those situations, where we tend to just pass, pass, pass to open up good shots.

We also tend to collapse back when we have a lead near the end of the game, and we get essentially no shots.

This would naturally raise the SA for 5v5 close and lower the SF for 5v5 close, which may even out the expected difference from regular ES situations.

Note, that I'm not saying the Leafs *can't* have a goals for rate above their shots for rate on a sustainable basis. It seems reasonable that skilled teams can do that. What I find crazy is to have a goals for rate 20% above shots for rate.

But judging from the graphs, it would appear that we don't have an average GF rate 20% above SF rate during even strength situations. I'm not sure we even hit that mark at any one point.

Even if it were, I don't really understand why 20% would be the "threshold of crazy".

 

[hatterson]

I'm not exactly sure why you think this is relevant. Are you telling me that for all other teams, 5v5 close GF and EV GF are exactly the same throughout all periods of the season?

No, I never claimed that, the point isn't that they are different, it's that they were *so* different for an extended period of time before dropping back to being very close for the majority of the season.

We have stretches above and stretches below. Yes, perhaps the stretch above is a little longer, but why is this unusual to you?

It's not unusual. What's unusual is for a team to be able to maintain it (which the Leafs didn't) thus it appears to be good evidence that they regressed. Teams get hot and get lucky for stretches, but teams haven't sustained these numbers over a season long basis

What I take from this graph is that we tended to score more goals when the score was close than when we were out of it or when we had a safe lead.

One would expect these results, no? It seems logical that teams would more often score when they have a chance to win than when it is out of their reach or they are trying to hold a lead. It seems even more logical when you watch how the Leafs play.

Except that *both* teams have a chance to win in 5v5 close games so you'd expect the other team would be playing good as well. Close refers to games within 1 goal, so it's not as if the generic EV data is just adding in 5 goal blowout games. Either way, it seems strange for a team to willingly play significantly worse during *any* game, regardless of the score.

The only difference I could imagine you are alluding to would be that shots when close aren't equally higher, but we're talking about the Leafs here. Our strategy when down is not "get more shots on net", as it is for most teams. In fact, we tend to do the opposite in those situations, where we tend to just pass, pass, pass to open up good shots.

We also tend to collapse back when we have a lead near the end of the game, and we get essentially no shots.

This would naturally raise the SA for 5v5 close and lower the SF for 5v5 close, which may even out the expected difference from regular ES situations.

The point is that the shots for differential between 5v5close and generic EV isn't different by very much, which indicates that the style of play didn't change all that much. Unless your claim is that for most of the game in a 5v5close game the Leafs have a better shot differential but the game long numbers (even 5v5close) are hurt because we go into a shell at the end of games? In that case, I think you're dramatically over estimating the effect at the end of close games.

But judging from the graphs, it would appear that we don't have an average GF rate 20% above SF rate during even strength situations. I'm not sure we even hit that mark at any one point.

That's exactly the point. In generic EV situations, the goals for rate never gets above 20% by any significant amount. However in close situations, which end up having the biggest impact on points earned, it was frequently above 20 and always above 15 for a stretch of 20-25 games.

Even if it were, I don't really understand why 20% would be the "threshold of crazy".

20% isn't some magic number. The point is that generic EV scoring rate and 5v5close scoring rate should be fairly close. Even if you want to argue that 5v5close should be higher because for some reason the Leafs try harder and actually play when a game is close (but their opponents don't) having it so much higher for a 20 game stretch, but not for the other 50 games on record indicates that there's at least a portion of luck involved and luck is, by definition, not sustainable.

 

[Delicious Dangles*]

No, I never claimed that, the point isn't that they are different, it's that they were *so* different for an extended period of time before dropping back to being very close for the majority of the season.

The peak before the drop is 17%. And after the so-called "regression" there is a 13% peak.

In fact, them being at similar points seems to be the rarer occurrence. There is a sizeable gap for a long stretch both before and after the supposed regression.

It's not unusual. What's unusual is for a team to be able to maintain it (which the Leafs didn't) thus it appears to be good evidence that they regressed. Teams get hot and get lucky for stretches, but teams haven't sustained these numbers over a season long basis

Well of course nobody is going to sustain their peak over an entire season. No Leaf fan was saying that we were going to win 75% of our games this season. That is why it is a peak. It is the very definition. You said yourself that it is a season of highs and lows, and that is what the graph shows.

Sustainability in hockey is not keeping everything exactly the same for every game all season. That is impossible. There are too many ever-changing variables. It is about having a combination of highs and lows that together consistently result in a good record over the course of a season.

Leafs had elite stretches of play near the beginning of the season and around the halfway mark of the season. Our worst stretch came in between, when a lot of other impeding variables were present. Our position in the Eastern conference has held relatively stable as well. That is not evidence of regression.

Except that *both* teams have a chance to win in 5v5 close games so you'd expect the other team would be playing good as well. Close refers to games within 1 goal, so it's not as if the generic EV data is just adding in 5 goal blowout games. Either way, it seems strange for a team to willingly play significantly worse during *any* game, regardless of the score.

Both teams have a chance to win, but if one already has the lead, they don`t need to do as much. And when the Leafs hold the lead in the 3rd, even in 1-goal games, they DON'T do as much. The system completely changes, and nearly all offensive pressure aside from counter-attacks stops.

On the other side of the coin, when we are trying to gain hold of a lead, we can be downright dominating.

It may seem strange that such gigantic swings can take place based on the score, but it is in fact very common to see, and even more common than average for the Leafs. That is what happens when you have a league full of parity that relies heavily on momentum, and a team that changes strategy based on situation.

And actually yes, the generic EV data WOULD be adding in a lot of blowout games for the Leafs.

The point is that the shots for differential between 5v5close and generic EV isn't different by very much, which indicates that the style of play didn't change all that much. Unless your claim is that for most of the game in a 5v5close game the Leafs have a better shot differential but the game long numbers (even 5v5close) are hurt because we go into a shell at the end of games? In that case, I think you're dramatically over estimating the effect at the end of close games.

That is a pretty huge assumption to say that because the shots for differential remains similar, that means that style of play doesn't change. Style of play changes A LOT. It is also quite a stretch to assume that teams play the same in all scenarios of close games, regardless of down by 1, tied, or up by one.

I think you are dramatically underestimating the shell that the Leafs go in. It is not uncommon for the entire final shot differential for the game to be caused by a single period of play.

That's exactly the point. In generic EV situations, the goals for rate never gets above 20% by any significant amount. However in close situations, which end up having the biggest impact on points earned, it was frequently above 20 and always above 15 for a stretch of 20-25 games.

Both show extended stretches above the shot lines, with the close situations line going below or above the All EV situations line at different times.

20% isn't some magic number. The point is that generic EV scoring rate and 5v5close scoring rate should be fairly close. Even if you want to argue that 5v5close should be higher because for some reason the Leafs try harder and actually play when a game is close (but their opponents don't) having it so much higher for a 20 game stretch, but not for the other 50 games on record indicates that there's at least a portion of luck involved and luck is, by definition, not sustainable.

Except it is NOT so much different for the other 50 games.

I could argue just as easily that it is so much lower for the 20 game stretch in the middle, which is not consistent with the other 50 games.

You have given no proof to support the assertion that EV scoring rate and 5v5 close scoring rate should be fairly close, is fairly close for all other teams in the league. And this would be going under the incorrect assumption that we play like the rest of the league anyway.

 

[Master_Of_Districts]

It would seem that you may have missed the point.

There is ample reason to be skeptical of the type of analysis that arises in this forum, in particular when it is applied to a specific circumstance such as this. That is not to say that there is no value to be found or that the discussion is not one of substance. When I read the posts that you have responded to for me the core issue has little to do with the specific mathematical manipulation being used but much more so this can be seen as debate about how valid the overall analysis might be when restricted to a circumstance that might well have rather unique characteristics.

The post I responded to included a list of individuals and their credentials with a clear implication that being "smart" made their opinions somehow beyond reproach. This is clearly not how things work in the academic world. And you suggested that these individuals, who by the way I respect to the degree that I have read their work, would have some absolute stance relative to the debate. Having read material from several of these individuals, I am less sure than you of their unwillingness to bend on the debate.

Which material or work of the individuals I cited render you unsure of their "unwillingness to bend on the debate"?

Because unless you're able to cite specific examples, you're got going to persuade me.

But I also think that it is poor form to suggest of others an intellectual rigidity that they may not in fact appreciate.

Look - binomial variance is a real factor that must be accounted for in any serious analysis. That much is clear to anyone who's actually taken the time to run the numbers.

 

[Fourier]

Which material or work of the individuals I cited render you unsure of their "unwillingness to bend on the debate"?

Because unless you're able to cite specific examples, you're got going to persuade me.



Look - binomial variance is a real factor that must be accounted for in any serious analysis. That much is clear to anyone who's actually taken the time to run the numbers.

You may want to go back and reread the specific post I quoted since you continue to miss my point. DD's use of the word ridiculous is an over statement. There has been reasoned rebuttal to many of his objections. But the premise of his argument remains intact and as I see it this has nothing to do with the validity of binomial variance.

The core problem I have with much of the debate on advanced statistics is the degree to which some people feel their analysis reveals absolute truth. This most often manifests itself when a broad base analysis is applied to try and say something definitive about a specific situation or individual.

Hockey is an extremely difficult sport to model accurately. And while I am often very impressed by the work of many of those who try to do so, including your own, I remain a skeptic about the true scope of what we have learned. Would you expect that to be different if I held a doctorate in mathematics as you claimed in the original post I quoted? Or is it my own ignorance that prevents me from understanding this complex stuff?

 

[hatterson]

The peak before the drop is 17%. And after the so-called "regression" there is a 13% peak.

In fact, them being at similar points seems to be the rarer occurrence. There is a sizeable gap for a long stretch both before and after the supposed regression.


Well of course nobody is going to sustain their peak over an entire season. No Leaf fan was saying that we were going to win 75% of our games this season. That is why it is a peak. It is the very definition. You said yourself that it is a season of highs and lows, and that is what the graph shows.

Sustainability in hockey is not keeping everything exactly the same for every game all season. That is impossible. There are too many ever-changing variables. It is about having a combination of highs and lows that together consistently result in a good record over the course of a season.

Leafs had elite stretches of play near the beginning of the season and around the halfway mark of the season. Our worst stretch came in between, when a lot of other impeding variables were present. Our position in the Eastern conference has held relatively stable as well. That is not evidence of regression.

It sounds like you're agreeing with me and then suddenly you're at "That is not evidence of regression" which confuses me. Not being able to maintain your peak through an entire season *is* regression, that's exactly what people were talking about. The Leafs would not and have not maintained the same level of goal conversion, especially in close games, as they did for the ~1/4 of a season near the beginning.

The Leafs *did* regress. The fact that they stayed above a 10% difference for 15 straight overlapping 10 game stretches (25 games overall) but have only exceeded 10% for another 2 overlapping 10 game stretches shows that it's more than likely the difference during that first ~25 game stretch was influenced by some luck which is, by definition, not sustainable.

Yes, I'm well aware that their place in the standings has held pretty solid (albeit unfortunately slipped a tad recently), but during the 26 game stretch from Nov 21st until Jan 12th the Leafs record in "corsi situations" was an abysmal 3-14-9 (w-l-t). The reason they maintained their position in the standings was that of those 9 shootout games, they managed to win 6 of them.

Those shootout wins don't prove the generic corsi/fenwick close model wrong (given that it's a probabilistic model a single outlier doesn't prove it wrong anyway, but that's another issue), all they do is prove that people who blindly apply the corsi/fenwick close model to overall records, without properly clarifying their statements are being sloppy or taking shortcuts.

Sure, it might be fun to point that out, but it doesn't actually help anyone. We're more interested in advancing the understanding (or lack of understanding) of the game and its statistical analysis, not in seeing who was right or wrong.

Both teams have a chance to win, but if one already has the lead, they don`t need to do as much. And when the Leafs hold the lead in the 3rd, even in 1-goal games, they DON'T do as much. The system completely changes, and nearly all offensive pressure aside from counter-attacks stops.

On the other side of the coin, when we are trying to gain hold of a lead, we can be downright dominating.

It may seem strange that such gigantic swings can take place based on the score, but it is in fact very common to see, and even more common than average for the Leafs. That is what happens when you have a league full of parity that relies heavily on momentum, and a team that changes strategy based on situation.

And actually yes, the generic EV data WOULD be adding in a lot of blowout games for the Leafs.


That is a pretty huge assumption to say that because the shots for differential remains similar, that means that style of play doesn't change. Style of play changes A LOT. It is also quite a stretch to assume that teams play the same in all scenarios of close games, regardless of down by 1, tied, or up by one.

I think you are dramatically underestimating the shell that the Leafs go in. It is not uncommon for the entire final shot differential for the game to be caused by a single period of play.


Both show extended stretches above the shot lines, with the close situations line going below or above the All EV situations line at different times.

I am well aware of the system the Leafs play, I watch the team as well. I'm also aware that the so called system of limiting chances to the outside doesn't seem to jive with the numbers regarding shot location and type in any significant way. Yes the Leafs allow a slightly smaller percentage of shots against from dangerous areas of the ice compared to other teams, however they still allow more shots against from dangerous areas than they take.

I've yet to see a true detailed analysis of the system that actually shows it's strengths. All claims I've seen rely on "just watch the team and you'll see it" or "Carlyle did it before" which in the case of the first is rife with confirmation bias and in the second is simply false.

If the Leafs are indeed able to employ this system it should shot up in the overall shot data. We should see significantly more shots from the point, and significantly fewer from the home plate area. Unless new data has popped up recently, any studies haven't shown significant differences in that manner.

Conversation rates, although sometimes hinting of underlying skill, are also plagued by sample size issues. As we can see the goals for conversion rate varies widely compared to the shots rate, as you would expect when goals involve some form of luck/variance.

Except it is NOT so much different for the other 50 games.

I could argue just as easily that it is so much lower for the 20 game stretch in the middle, which is not consistent with the other 50 games.

You have given no proof to support the assertion that EV scoring rate and 5v5 close scoring rate should be fairly close, is fairly close for all other teams in the league. And this would be going under the incorrect assumption that we play like the rest of the league anyway.

Excluding the 15 data points that we're debating, there are 45 remaining currently (I'm a game or two behind). Of those 45, 24 (53%) are within +/-5 points and 38 (84%) are within +/-7.5 points. So yes, it seems that 5v5close and generic EV track pretty close.

 

[MastuhNinks]

Fun fact: 17 of the league's 30 teams have a better win percentage when being outshot than their win percentage when outshooting opponents.

Shot differential isn't a completely useless statistic, but it is way, WAY overvalued by the 'advanced stats community'.

 

[Bear of Bad News]

Fun fact: 17 of the league's 30 teams have a better win percentage when being outshot than their win percentage when outshooting opponents.

Shot differential isn't a completely useless statistic, but it is way, WAY overvalued by the 'advanced stats community'.

Fun fact: the "advanced stats community" has moved well beyond this. See some of the responses here:

http://hfboards.mandatory.com/showthread.php?t=1624957

 

[svat]

Fun fact: 17 of the league's 30 teams have a better win percentage when being outshot than their win percentage when outshooting opponents.

Shot differential isn't a completely useless statistic, but it is way, WAY overvalued by the 'advanced stats community'.

score effects. teams up in games are more likely to sit on leads.

this is why stats like 5 on 5 corsi and fenwick close (just means the percentage of corsi or fenwick events in your favor when the game is within 1 goal) is important. Every single team, even elite teams, tend to let up more corsi events and shots on their net when up in games. Once we know more about shot quality it will likely prove (it already sort of does) with scoring chance data that teams down by a couple that are out-shooting typically don't outchance their opponent

 

[MVP of West Hollywd]

Well the Leafs were definitely due for a regression and are now paying the piper. Considering their negative goal differential did not indicate contending for 3rd in the East and how hot they were before the Olympic break - and neither did their roster talent level.

 

[Bear of Bad News]

Nice Scott Cullen article from Wednesday afternoon:
http://www.tsn.ca/nhl/story/?id=447435

Focuses on the Leafs' recent struggles (which is why I put it here), but also covers the Penguins, Red Wings, and Lightning.

In short, it addresses how the Leafs have succeeded despite poor puck possession numbers (high save percentage plus shootout success), and where that's gone wrong recently.

 

[Cap'n Flavour]

lol.

 

[EichHart]

Lots of LOLs in this thread. Math doesn't lie.

 

[TheBrew]

Personally i love that the Leafs are giving a big ol' middle finger to advanced stats.

Well i guess advanced stats are giving a big ol' middle finger back to the Leafs

 

[Bear of Bad News]

More from the Regina Leader-Post on what's going on:
http://www.leaderpost.com/sports/Leafs+losing+streak+astonishing+collapse/9679974/story.html

 

[Baemon Severson*]

If the collapse of the 13-14 Leafs doesn't convince you that "advanced" stats are meaningful then nothing will. People have been predicting their demise since last season and were laughed out of threads. In fact, due to the Leafs' shootout record and high save percentage, they have largely overachieved this season.

 

[hatterson]

If the collapse of the 13-14 Leafs doesn't convince you that "advanced" stats are meaningful then nothing will. People have been predicting their demise since last season and were laughed out of threads. In fact, due to the Leafs' shootout record and high save percentage, they have largely overachieved this season.

It shouldn't convince you of anything since the Leafs aren't losing because they're being shelled all game. They're losing because of poor goaltending and poor special teams. Advanced stats didn't predict the starting goalie being injured and the backup playing terrible. Advanced stats didn't predict the special teams losing games as opposed to winning them.

In this 8 game losing streak there's only 1 game where they got significantly outplayed. The 5-3 loss against St. Louis. In all other games they either outshot their opponents or were within a couple shots.

No one is denying that the Leafs have had a good run, but their collapse the last 8 games doesn't magically prove the "Corsi is everything" crowd correct. The underlying questions of shot quality still remain unaddressed.

Even if you believe it's proof, advanced stats doesn't predict a team goes on a 0-8 skid when they're competitive from a shot standpoint in 7 of the games. Even a poor record of 2-4-1 has the Leafs inside the playoffs, possibly with as much as a 5 point cushion on 9th depending on who those wins came against.

And to clarify the goaltending point. The Leafs haven't been losing because we're no longer having a goaltender stand on his head. We're losing (or at least not getting any points) because we're receiving downright poor goaltending. In 5 of the 8 games it's been below a .900 save percentage and 2 of the ones above that were at .906 and .903.

 

[svat]

It shouldn't convince you of anything since the Leafs aren't losing because they're being shelled all game. They're losing because of poor goaltending and poor special teams. Advanced stats didn't predict the starting goalie being injured and the backup playing terrible. Advanced stats didn't predict the special teams losing games as opposed to winning them.

In this 8 game losing streak there's only 1 game where they got significantly outplayed. The 5-3 loss against St. Louis. In all other games they either outshot their opponents or were within a couple shots.

No one is denying that the Leafs have had a good run, but their collapse the last 8 games doesn't magically prove the "Corsi is everything" crowd correct. The underlying questions of shot quality still remain unaddressed.

Even if you believe it's proof, advanced stats doesn't predict a team goes on a 0-8 skid when they're competitive from a shot standpoint in 7 of the games. Even a poor record of 2-4-1 has the Leafs inside the playoffs, possibly with as much as a 5 point cushion on 9th depending on who those wins came against.

And to clarify the goaltending point. The Leafs haven't been losing because we're no longer having a goaltender stand on his head. We're losing (or at least not getting any points) because we're receiving downright poor goaltending. In 5 of the 8 games it's been below a .900 save percentage and 2 of the ones above that were at .906 and .903.

like a bunch of people said on twitter, the shot metrics didn't predict an 8 game losing streak (why would they), but they did predict the leafs 26-28-8 record after their hot start. the leafs went from getting insane luck to now getting some pretty lousy luck (even the worst teams aren't going to lose 8 straight without getting unlucky).

also fwiw, the leafs corsi close % over the 8 game streak is just as bad as it has been all year.

 

[Micklebot]

It shouldn't convince you of anything since the Leafs aren't losing because they're being shelled all game. They're losing because of poor goaltending and poor special teams. Advanced stats didn't predict the starting goalie being injured and the backup playing terrible. Advanced stats didn't predict the special teams losing games as opposed to winning them.

In this 8 game losing streak there's only 1 game where they got significantly outplayed. The 5-3 loss against St. Louis. In all other games they either outshot their opponents or were within a couple shots.

No one is denying that the Leafs have had a good run, but their collapse the last 8 games doesn't magically prove the "Corsi is everything" crowd correct. The underlying questions of shot quality still remain unaddressed.

Even if you believe it's proof, advanced stats doesn't predict a team goes on a 0-8 skid when they're competitive from a shot standpoint in 7 of the games. Even a poor record of 2-4-1 has the Leafs inside the playoffs, possibly with as much as a 5 point cushion on 9th depending on who those wins came against.

And to clarify the goaltending point. The Leafs haven't been losing because we're no longer having a goaltender stand on his head. We're losing (or at least not getting any points) because we're receiving downright poor goaltending. In 5 of the 8 games it's been below a .900 save percentage and 2 of the ones above that were at .906 and .903.

Out of curiosity, what do you define as competitive from a shots standpoint? Of the last 8 games, Toronto has had the follow SF% at 5vsclose: 42.90%, 54.30%, 41.20%44.20%, 35.70%, 27.30%, 60.70%, 54.20%. I think I'd only consider 3 of those competitive, though a case could be made for 4.

They didn't fair any better at shot attempts either 43.50%, 48.40%, 28.60%, 51.00%, 40.00%, 17.60%, 49.10%, 45.80%. Again, 3 or 4, maybe 5 are competitve depending on your baseline.

I'd agree that Goaltending has been the larger issue, but if you're not getting outplayed you can overcome shoddy goaltending from time to time. This is a case of everything going wrong at the same time, which to be honest, you'd expect when you get an 8 game losing streak.

 

[Minor Boarding]

It shouldn't convince you of anything since the Leafs aren't losing because they're being shelled all game. They're losing because of poor goaltending and poor special teams. Advanced stats didn't predict the starting goalie being injured and the backup playing terrible. Advanced stats didn't predict the special teams losing games as opposed to winning them.

In this 8 game losing streak there's only 1 game where they got significantly outplayed. The 5-3 loss against St. Louis. In all other games they either outshot their opponents or were within a couple shots.

No one is denying that the Leafs have had a good run, but their collapse the last 8 games doesn't magically prove the "Corsi is everything" crowd correct. The underlying questions of shot quality still remain unaddressed.

Even if you believe it's proof, advanced stats doesn't predict a team goes on a 0-8 skid when they're competitive from a shot standpoint in 7 of the games. Even a poor record of 2-4-1 has the Leafs inside the playoffs, possibly with as much as a 5 point cushion on 9th depending on who those wins came against.

And to clarify the goaltending point. The Leafs haven't been losing because we're no longer having a goaltender stand on his head. We're losing (or at least not getting any points) because we're receiving downright poor goaltending. In 5 of the 8 games it's been below a .900 save percentage and 2 of the ones above that were at .906 and .903.

The reason Leafs are losing is in the general stats page: 3.09 GA/G == 4th worse in the league.
Live by the sword, die by the sword.

Translation -> Leafs suck defensively and their lucky start might not be enough to overcome the harsh reality that they ice an average team that has a couple of huge holes in their roster as far as the lack of a true #1 center and #1 defenseman...
Fact is the Leafs are cooling off as some of those stats predicted.

And quite frankly, most of the recent Stanley Cup winners have been also some of the better advanced stats teams...

 

[hatterson]

Out of curiosity, what do you define as competitive from a shots standpoint? Of the last 8 games, Toronto has had the follow SF% at 5vsclose: 42.90%, 54.30%, 41.20%44.20%, 35.70%, 27.30%, 60.70%, 54.20%. I think I'd only consider 3 of those competitive, though a case could be made for 4.

They didn't fair any better at shot attempts either 43.50%, 48.40%, 28.60%, 51.00%, 40.00%, 17.60%, 49.10%, 45.80%. Again, 3 or 4, maybe 5 are competitve depending on your baseline.

I'd agree that Goaltending has been the larger issue, but if you're not getting outplayed you can overcome shoddy goaltending from time to time. This is a case of everything going wrong at the same time, which to be honest, you'd expect when you get an 8 game losing streak.

Based on numbers, not percentages the Leafs have had the following corsi differentials 5v5 close:

-3, -2, -18, +2, -5, -33, -1, -4

I'd say all of the shot differentials except -18 and -33 are competitive.

When dealing with single games I think sometimes we can forget how much influence a single flurry can have on overall percentages.

 

[svat]

The reason Leafs are losing is in the general stats page: 3.09 GA/G == 4th worse in the league.
Live by the sword, die by the sword.

Translation -> Leafs suck defensively and their lucky start might not be enough to overcome the harsh reality that they ice an average team that has a couple of huge holes in their roster as far as the lack of a true #1 center and #1 defenseman...
Fact is the Leafs are cooling off as some of those stats predicted.

And quite frankly, most of the recent Stanley Cup winners have been also some of the better advanced stats teams...

yep. outliers being the bruins who rode insanely hot goaltending, were privy to a pretty big goalie breakdown in a few games and also were not a bad possession team but just rather mediocre.

the penguins also during their cup season weren't world beaters but improved a ton under bylsma, and also happen to have crosby and malkin.

this infographic really just tells you how important possession is.

http://www.habseyesontheprize.com/2013/4/4/4178716/why-possession-matters-a-visual-guide-to-fenwick

 

[hatterson]

The reason Leafs are losing is in the general stats page: 3.09 GA/G == 4th worse in the league.
Live by the sword, die by the sword.

Translation -> Leafs suck defensively and their lucky start might not be enough to overcome the harsh reality that they ice an average team that has a couple of huge holes in their roster as far as the lack of a true #1 center and #1 defenseman...
Fact is the Leafs are cooling off as some of those stats predicted.

And quite frankly, most of the recent Stanley Cup winners have been also some of the better advanced stats teams...

They're actually cooling off far *more* than the stats predicted. Which is why I said that people using it as proof of the validity of the stats are in error.

I'm all for the use of advanced stats and I believe in their power. But the Leafs luck in the last 10-15 games has been anomalous from a corsi/fenwick standpoint so them missing the playoffs doesn't prove anything in that regard.

The fact remains that for a large percentage of the year the Leafs were able to maintain a goals for percentage higher than their corsi/fenwick/shots for percentage. Yes, as I said earlier in the thread, their luck from the beginning of the season (in which they had goals for percentages in the 60+% range despite CF/FF/SF% in the low 40s) cooled off, but they still maintained a non-trivial increase for a large portion of the season.

If that's luck, fine, call it luck (although I think that's lazy from a statistical standpoint), but you can't argue that the Leafs collapsing and missing the playoffs proves it was luck when the collapse and playoff miss requires something just as lucky (or unlucky)

Edit: To be clear, I'm not arguing that the Leafs aren't deeply flawed, or that we're just unlucky to miss the playoffs. Trust me, I'm well aware of the flaws in this team. I'm commenting only on the fact that using this 8 game losing streak to "prove" corsi is just as flawed as using their record 15 games ago to "disprove" it.