Sportsnet: Some interesting points on 31 thoughts somewhat leafs related

[Brobust]

15. After last season, I had a lengthy conversation with Steve Valiquette, the former goalie who works on the Rangers’ broadcasts and is the CEO of Clear Sight Analytics. Like others who try to build strong predictive models, he is constantly searching for the right ingredient or statistic to create that “Eureka!” moment. After Tampa Bay won the 2020 Stanley Cup, he’d zeroed in on something.
“For a few years, I thought the surest predictor of winning was goaltending,” he said Sunday. “Then I was looking at differentials, such as expected goals for and against.”
It wasn’t predictive enough for him — until he zeroed in on high-danger goals against at five-on-five/60 minutes. Goalie errors are not included.
“We leave them out for this exercise,” he added.
Here’s last season’s Final 16 playoff bracket:

As you can see, teams ranked higher in the stat Valiquette pinpointed went 14-1.
“Two years ago, Tampa Bay was 19th; last year [they were] first,” Valiquette said.
(The Lightning slightly ruined this year’s exercise, finishing 18th in 2020–21 as they leaned more on Andrei Vasilevskiy.)

16. Dallas was No. 1 overall in high-danger goals against at five-on-five/60 minutes this season, but didn’t make the playoffs. Guess who was second? Your Stanley Cup finalist Montreal Canadiens.
“I did not put any money on Montreal,” Valiquette laughed. “At face value, I didn’t see it happening, no way.”
We both laughed at the idea he didn’t trust his own model, but the proof is four wins away from a championship. Toronto, by the way, was third, one spot below the Canadiens.

17. The conversation with Valiquette gave me an opportunity to ask him one story I’d heard, but he declined to answer, saying he could not discuss personnel moves because some teams are clients. The story I’d heard is that Minnesota finished first in that metric last season, and targeted Cam Talbot because his specific strengths addressed their specific weaknesses.
I think that’s true, but got no confirmation. I also think he advises teams that finish high in the metric but don’t finish well — Minnesota last year, Dallas and Toronto this year — to address those kinds of specific weaknesses rather than ripping down to the studs and starting over.

Click to read more...

 

[Brobust]

Nothing?

I thought people were very eager to figure out what happened with the leafs this season.

 

[Racer88]

Nothing?

Nothing?

I thought people were very eager to figure out what happened with the leafs this season.

The minute you start talking about expected goals and high danger chances 5vs5 but he left out goalie mistake…..etc etc you lose me and most people.
The game is played on the ice with human beings not on a spread sheet or slide rule. we know what happened to the Leafs. 2 of their highest paid player failed to perform when it mattered yet again so all the regular season stats go out the window when the grind of the playoffs begin

 

[Racer88]

I think I messed up with my previous quote order.
Lol

 

[Brobust]

Hey, you're the guy who got insulted when I said you lost faith in the stars.

I think there's part of rational explanation here in what went wrong. Maybe you'll start believing again.

 

[Racer88]

Hey, you're the guy who got insulted when I said you lost faith in the stars.

Nope you are thinking of someone else, however I have absolutely lost faith in the stars.
That part of the article made absolutely no sense to me.
What I took from it is he kept looking until he found a model that fit his narrative but left things out. Then he uses Tampa as an example for 3 different years 1 year they were at the bottom and lost 1st round then they were much higher and won the cup this year they are way down and probably will win the cup………..looks like his Model is all over the place.
If I am reading it wrong please correct me…….I promise I will not be offended.
Lol

 

[Cap'n Flavour]

The concept of xG seems fine but last time I looked at the three public sites that tracked it, they all had significantly different figures, and presumably one or more of: different data sources, different probabilities for a given shot to go in, and different weighting (often non-existent) to match xG with actual goals.

One thing that particularly irks me is that it's still missing context. For example, a weak, off-balance shot into the pads from close in is usually considered a higher xG event than Ovechkin/Stamkos blasting a one-timer from the top of the circle. I read a hypothesis that Tampa Bay was ranked low in xG this season but consistently outperformed because they were *completing* long lateral passes and firing long-distance, "low percentage" shots... into a half-empty net.

 

[Brobust]

Nope you are thinking of someone else, however I have absolutely lost faith in the stars.
That part of the article made absolutely no sense to me.
What I took from it is he kept looking until he found a model that fit his narrative but left things out. Then he uses Tampa as an example for 3 different years 1 year they were at the bottom and lost 1st round then they were much higher and won the cup this year they are way down and probably will win the cup………..looks like his Model is all over the place.
If I am reading it wrong please correct me…….I promise I will not be offended.
Lol

It's not a model. It's just an interesting observation.

You have to admit that it's interesting that the team with the better net front defence won the series 100% of the time last season.

 

[makbeer]

Jack Campbell had under .600 sv% against high danger chances in the post season against Montreal and price was over .800 for the series. But you can’t bring up bad stats about him because he’s a nice guy and the fans have anointed him as our lord and saviour.

 

[Martin Skoula]

The concept of xG seems fine but last time I looked at the three public sites that tracked it, they all had significantly different figures, and presumably one or more of: different data sources, different probabilities for a given shot to go in, and different weighting (often non-existent) to match xG with actual goals.

One thing that particularly irks me is that it's still missing context. For example, a weak, off-balance shot into the pads from close in is usually considered a higher xG event than Ovechkin/Stamkos blasting a one-timer from the top of the circle. I read a hypothesis that Tampa Bay was ranked low in xG this season but consistently outperformed because they were *completing* long lateral passes and firing long-distance, "low percentage" shots... into a half-empty net.

If your bad shooters are consistently getting opportunities in the slot over thousands of data points over the course of the season, your team is either doing something right with their offensive structure or the other team is just allowing them to shoot from prime real estate for some reason.

Obviously it's not a valuable shooting stat for a single game but it does tell you how effective a team is at getting and defending that real estate over the course of a season. Obviously outliers like Ovi, Laine, and Trotz exist. You cant do anything to stop Ovi from getting shots from the circle, teams have had 15 years to figure out a way to stop it and they haven't without opening up the rest of his team. I don't care if the other team's AHL call up with 0 shot power is the one taking 10 slot shots against my team, he should not have the opportunity in the first place.

 

[Dreakmur]

The concept of xG seems fine but last time I looked at the three public sites that tracked it, they all had significantly different figures, and presumably one or more of: different data sources, different probabilities for a given shot to go in, and different weighting (often non-existent) to match xG with actual goals.

One thing that particularly irks me is that it's still missing context. For example, a weak, off-balance shot into the pads from close in is usually considered a higher xG event than Ovechkin/Stamkos blasting a one-timer from the top of the circle. I read a hypothesis that Tampa Bay was ranked low in xG this season but consistently outperformed because they were *completing* long lateral passes and firing long-distance, "low percentage" shots... into a half-empty net.

Seems pretty obvious to me that not every shot is an equal scoring chance. What kinds of shots score most?

From 2020:
9% deflections
16% in stride
54% off a pass
16% off a rebound
5% broken play

 

[Racer88]

It's not a model. It's just an interesting observation.

You have to admit that it's interesting that the team with the better net front defence won the series 100% of the time last season.

For sure it’s interesting..
I figure in about 3 months I will be over my total disappointment over yet another year of failure…….lol but I will do me best to not put to much faith in how they play during the regular season…….good or bad