Sportsnet: Maple Leafs must add proven defenders, move on from all-offence identity

[TOGuy14]

For an all offence identity we sure do seem to get shut out and play without much of an attack in a lot of games...

 

[GirardSpinorama]

They are accurate.

Models don't need to confirm a bias to be right.

I haven't seen you address any actual problems.

Theres a problem in the actual data (check on the seth jones is underrated trend, where people have pointed out that NHL.com shot locations are not always correct), the assumptions made, and the accuracy of only being 80% reliable in predicting a goal. 80% is TERRIBLE; when you consider the marginal difference between good and bad goal scoring/prevention teams.

Analytics is supposed to be and will likely be (with improvements) a disruptive way to look at things. The oneus is on you to convince people with traditional view points on why this new hot product is useful. So far, I don't see its value. Maybe one day when they track goaltender positioning or goaltender line of sight, this stat will be more useful.

 

[Shanwhatplan]

But you don't get a + for every type of point, and he played on some horrible rebuilding Colorado teams. There are better ways to say you don't like Barrie defensively than the worst stat in the history of stats.

He's not a defensive defenseman, but his weakness are overblown. Whether you personally like him or the style of defenseman he is is irrelevant. The point is that he was a top 4 defenseman when we acquired him, so that's not "ignoring the defense" like you claimed. It's also not his only move to address the defense.

Yes, I do realize that, but you mentioned the Capitals and said that Trotzs teams play a certain way. Washington was weak defensively in the year they won.

But spending extended periods of time in the defensive zone isn't good defensive structure, so the quote you referenced doesn't make sense. Also, the Islanders are not very good at preventing high quality chances. There are other aspects of defense they are good at, but that's not really one of them.

No, signing 4 goalies for 5m each is not a good move. It has nothing to do with who does it.

You just don’t get it, and I don’t think you ever will. The Barrie trade was horrible. You get a + if you’re on the ice when your team scores, and a - when you’re scored on (oh, I realize you’re going to say that you knew that). The point is, over time the +/- will help show what a player is really like. As I’ve pointed out to you many times, Barrie MAY have been a top 4 OFFENSIVE defenceman (again, that’s very debatable), but we needed a top 4 DEFENSIVE defenceman! So, yes, Dubas ignored the problem.

I said Trotz’ team plays a certain way, and his team is the Islanders. And if you’d listen to his quote, you’d realize his point was that his team is very solid defensively, and will not give up very many quality chances. Not that they do spend a lot of time in their zone. So, yes, my quote does make sense. Get your facts straight. It’s amazing how you like to twist other people’s words around to suit your agenda.

Sorry, but it’s so obvious you feel Dubas can never do any wrong.

I am so looking forward to the next few years. If Dubas keeps this core together, and they take us deep into the playoffs (assuming we even make the playoffs) a few times, then I may start to believe he knows what he’s doing. However, I still feel Marner or Nylander needs to be traded in order to address our other needs.

 

[Dekes For Days]

Theres a problem in the actual data, the assumptions made, and the accuracy of only being 80% reliable in predicting a goal.

That's not addressing actual problems. That's just making vague, unsupported claims. I don't see where you got 80% from, and 80% is pretty good anyway considering it's not supposed to consider the quality of the shooting talent or opposing goalie in any given instance. And over significant samples, these things even out to be relatively insignificant.

The oneus is on you to convince people with traditional view points on why this new hot product is useful.

It's not really that new. It's been refined and improved over many years, and the evidence for it's usefulness is widely available. It's not my job to pull outdated viewpoints into the future kicking and screaming. I've explained the benefits of utilizing these statistics quite a bit, and regardless of how detailed you believe the stats to be, they are undeniably better than solely utilizing the eye test, or drawing whatever arbitrary conclusions you want about defense/goaltending from goals against.

 

[Stephen]

Theres a problem in the actual data (check on the seth jones is underrated trend, where people have pointed out that NHL.com shot locations are not always correct), the assumptions made, and the accuracy of only being 80% reliable in predicting a goal. 80% is TERRIBLE; when you consider the marginal difference between good and bad goal scoring/prevention teams.

Analytics is supposed to be and will likely be (with improvements) a disruptive way to look at things. The oneus is on you to convince people with traditional view points on why this new hot product is useful. So far, I don't see its value. Maybe one day when they track goaltender positioning or goaltender line of sight, this stat will be more useful.

I'm really interested in seeing how GPS biometric tracking data will evolve the analytics conversation. I want to see things like:

-Skating stats: total distance traveled, average speed, stops and starts, heart rate etc.
-Collisions.
-Movement vectors, like whether a player attacks the net in straight lines, or avoids heavy traffic, or whatever.
-Time spent in each zone.

 

[Dekes For Days]

You get a + if you’re on the ice when your team scores, and a - when you’re scored on

No, that's not true. It gives you a plus when you score in certain situations, and a - when you get scored on in certain situations, and it's not very well tied to an individual player or their abilities.

we needed a top 4 DEFENSIVE defenceman!

We needed top 4 defensemen. Period. Regardless of what you think about Barrie, it was undeniably an improvement on our other options going into this year. Defensemen also aren't just "offense" or "defense". Barrie provided elements other than production that we lacked, and had been exploited, like an ability to exit the zone from the right side instead of just endlessly icing it.

I said Trotz’ team plays a certain way, and his team is the Islanders.

But it used to be the Capitals, which you mentioned, and they didn't play that way when they won.

And if you’d listen to his quote, you’d realize his point was that his team is very solid defensively, and will not give up very many quality chances.

His team is alright defensively, but they do give up a decent amount of quality chances. No team wants to spend endless amounts of time in their own zone. That's a horrible strategy.

 

[Menzinger]

I'm really interested in seeing how GPS biometric tracking data will evolve the analytics conversation. I want to see things like:

-Skating stats: total distance traveled, average speed, stops and starts, heart rate etc.
-Collisions.
-Movement vectors, like whether a player attacks the net in straight lines, or avoids heavy traffic, or whatever.
-Time spent in each zone.

Agreed. But it'll take a while for teams to figure out how to properly use it. The teams. Who already have the folks in those analysis roles will have a step ahead.

Early on though there's going to be a lot of inane debates that player X is twice as good as player Y because they skated through the neutral zone. 005 seconds quicker

 

[Stephen]

Agreed. But it'll take a while for teams to figure out how to properly use it. The teams. Who already have the folks in those analysis roles will have a step ahead.

Early on though there's going to be a lot of inane debates that player X is twice as good as player Y because they skated through the neutral zone. 005 seconds quicker

It'll definitely open up a huge can of worms, but I'm ready to dig into the new conversation!

 

[Dekes For Days]

Early on though there's going to be a lot of inane debates that player X is twice as good as player Y because they skated through the neutral zone. 005 seconds quicker

Lol. Yeah, I'm definitely expecting to see a lot of this.

 

[Stephen]

Lol. Yeah, I'm definitely expecting to see a lot of this.

Sounds like an inherent bias to pure information.

 

[Trapper]

Like two superstars on ELCs.

Too bad the Leafs couldn’t do it when they had “three”.

 

[Shanwhatplan]

No, that's not true. It gives you a plus when you score in certain situations, and a - when you get scored on in certain situations, and it's not very well tied to an individual player or their abilities.

We needed top 4 defensemen. Period. Regardless of what you think about Barrie, it was undeniably an improvement on our other options going into this year. Defensemen also aren't just "offense" or "defense". Barrie provided elements other than production that we lacked, and had been exploited, like an ability to exit the zone from the right side instead of just endlessly icing it.

But it used to be the Capitals, which you mentioned, and they didn't play that way when they won.

His team is alright defensively, but they do give up a decent amount of quality chances. No team wants to spend endless amounts of time in their own zone. That's a horrible strategy.

I really don’t know why I’m wasting my time with you. +/- is determined as follows, you get a + if you’re on the ice when your team scores at even strength or while short handed, and a - if it’s the other way. So what does -66 say about Barrie??? And that’s great if he can skate the puck out of our zone, but if he’s on the ice FAR MORE OFTEN when the other team scores, that kind of defeats the purpose, or am I missing something?

As far as Trotz and the Capitals and Islanders, and his quote, I believe I referred to him as a Cup winning coach (with the Capitals),but in terms of his quote regarding hemmed up in his own zone, I was referring to the Islanders. I will gladly admit I’m wrong if you show me where I said otherwise (I’m not going to go searching).
And please, TRY to get the GIST of what Trotz was saying. It’s really the basis of why I quoted him. smh

 

[Dekes For Days]

Sounds like an inherent bias to pure information.

..What?

Too bad the Leafs couldn’t do it when they had “three”.

I wouldn't exactly call Nylander a superstar when he was on his ELC, at least not on the level of the other 4.

 

[Stephen]

Too bad the Leafs couldn’t do it when they had “three”.

Well, not like the Canucks were up against an opponent as strong as Boston or Washington... lol

 

[Dekes For Days]

you get a + if you’re on the ice when your team scores at even strength or while short handed, and a - if it’s the other way.

That's not what you originally said.

So what does -66 say about Barrie???

Not much more than he played on some bad teams.

And that’s great if he can skate the puck out of our zone, but if he’s on the ice FAR MORE OFTEN when the other team scores, that kind of defeats the purpose, or am I missing something?

He wasn't on the ice "far more often" when the other team scored. Regardless, Barrie is not a defensive wizard, but he brings useful elements and his addition was not "ignoring the defense".

 

[ToneDog]

For an all offence identity we sure do seem to get shut out and play without much of an attack in a lot of games...

Seriously who are the snipers on the team (especially when a goal is needed) ? #34, #91 and maybe #88. The rest are either plugs or 20 goal scorers at best. And our D provides virtually nothing wrt goals.

Little offence when it counts + terrible defence + questionable goaltending in critical situations = no playoffs or early playoff exit.

#BlameItOnShannyAndDubas

 

[shortfuze]

-He played all 7 in 2012-2013 vs Boston
-He played all 6 in 2016-2017 vs Washington
-He played 4 games against Boston in 2017-2018

The only one where your statement holds true is last season's playoffs. New fan are you?

Nah. Just lazy and hungover. But thanks for the info.

 

[GirardSpinorama]

That's not addressing actual problems. That's just making vague, unsupported claims. I don't see where you got 80% from, and 80% is pretty good anyway considering it's not supposed to consider the quality of the shooting talent or opposing goalie in any given instance. And over significant samples, these things even out to be relatively insignificant.

It's not really that new. It's been refined and improved over many years, and the evidence for it's usefulness is widely available. It's not my job to pull outdated viewpoints into the future kicking and screaming. I've explained the benefits of utilizing these statistics quite a bit, and regardless of how detailed you believe the stats to be, they are undeniably better than solely utilizing the eye test, or drawing whatever arbitrary conclusions you want about defense/goaltending from goals against.

Hockey analytics has gone from pure garbage (basically shots but with shot attempts) to utilizing basic regressions and/or some simple decision tree models. It is no where NEAR good enough to convince people who actually watch hockey AND have basic knowledge of analytics in other fields (banking, finance etc). 80% is not good enough as I have already explained for it to replace the eye test. Maybe you can use it as a filter and then validate with actually watching the games. But for now, the large gap is likely due to all the missing factors.

If you want to replace the eye test, the data quality better be good, it MUST include all important variables. One day, someone will improve the model enough to show how bad the leafs are this year, but at that point you will start disregarding it.

 

[Throw More Waffles]

Well, not like the Canucks were up against an opponent as strong as Boston or Washington... lol

They just beat last years cup winner...

What in the world?

Edit: I think maybe you were being sarcastic and I didn't catch it.

 

[Shanwhatplan]

That's not what you originally said.

Not much more than he played on some bad teams.

He wasn't on the ice "far more often" when the other team scored. Regardless, Barrie is not a defensive wizard, but he brings useful elements and his addition was not "ignoring the defense".

So please tell me what you said in response to my +/- statement.

And please tell me what all these other “useful” elements are that Barrie brings. Please!

 

[Dekes For Days]

80% is not good enough as I have already explained for it to replace the eye test.

You haven't supported your 80% claim, you haven't explained why that wouldn't be good enough, especially over significant samples that minimize the discrepancies, you haven't addressed any of my responses explaining why it is good enough, and you haven't supported ignoring all data and solely using the eye test being better, so I'm going to continue to utilize the wide array of extremely valuable information that is available to me to best evaluate my team, so that I can more effectively target where the weakness are.

 

[GirardSpinorama]

I'm really interested in seeing how GPS biometric tracking data will evolve the analytics conversation. I want to see things like:

-Skating stats: total distance traveled, average speed, stops and starts, heart rate etc.
-Collisions.
-Movement vectors, like whether a player attacks the net in straight lines, or avoids heavy traffic, or whatever.
-Time spent in each zone.

I'm surprised time spent in each zone or even time of possession isn't already a widely avaliable stat like they have in soccer.

That type of biometric data will definitely be valuable for player evaluations or optimization of player ice time.

Unfortunately, what we have, publicly available, are just basic applied stats (that are also missing key variables) that tries to model a VERY complex problem (i.e. how good a defence/offense is).

 

[Dekes For Days]

So please tell me what you said I need response to my +/- statement.

I don't know what you're trying to say here.

And please tell me what all these other “useful” elements are that Barrie brings.

I already gave you one. He's a player that's fairly good at zone exits, which were bottom of the league and heavily exploited previously, so that we could more effectively exit our zone and attack instead of endlessly icing pucks like we used to. We also utilized him at times as an effective shooting threat on the PP to open up space. Even with his weaknesses, he was a significant upgrade on the alternatives for that spot in pretty much every way.

 

[Shanwhatplan]

I don't know what you're trying to say here.

I already gave you one. He's a player that's fairly good at zone exits, which were bottom of the league and heavily exploited previously, so that we could more effectively exit our zone and attack instead of endlessly icing pucks like we used to. We also utilized him at times as an effective shooting threat on the PP to open up space. Even with his weaknesses, he was a significant upgrade on the alternatives for that spot in pretty much every way.

I edited my first question. I said you tell me what your response was in regards to my +/- statement (I did say a + is when your team scores when you’re on the ice, missed the short handed element).

In terms of your undying belief in Barrie, we will see how Dubas feels. My belief is if he resigns him, he’s doing it in hopes of trying to at least break even in that horrible trade. If he doesn’t resign him, at least he’s come to his senses on who Barrie really is as a defenceman, which is a 3rd pairing PP specialist.

 

[GirardSpinorama]

You haven't supported your 80% claim, you haven't explained why that wouldn't be good enough, especially over significant samples that minimize the discrepancies, you haven't addressed any of my responses explaining why it is good enough, and you haven't supported ignoring all data and solely using the eye test being better, so I'm going to continue to utilize extremely valuable information that is available to me to best evaluate my team, so that I can more effectively target where the weakness are.

80% is shown in a link that i've already provided.

The reason why 80% accuracy is not good enough is simple.

For example,
In a 45%/55% split of expected goals game (which is not uncommon, as you can see from the TO/Columbus series, 4 games were like that), the model would have to be very accurate to help you determine the outcome of the game. If its only 80% accurate (over a large sample size), its useless. The margin of error is bigger than the difference between two team.

Remember, this is 80% accuracy over a large sample size (i think the models I've looked at uses shots from many seasons); that means over small sample sizes, the error could be even bigger. For example, it could be 100% accurate for 100 shots and then only 60% accurate predicting the next 100 shots. 60% accurate is only 10% better than guess (its a binary outcome).

Games - Natural Stat Trick

In a bigger context,

The difference between even the TOP scoring team and WORST team isn't even that big. Its not 200%, 100%. So if your model is only 80% accurate (again this is how accurate it is over a large sample), how helpful is it to tell you the difference between even the best and worst team (nevermind the teams in the middle)?

The way I see it; xG is only marginally better at telling you which team played better versus the good old shot counter.