Yes but no.
no offense to the skaters I’m naming, but having Bortuzzo and Walman as a defensive pair is not depth.
blues stacked up front, not so much in the back.
2021 Blues Regular Season multi-purpose thread
- Thread starter Stupendous Yappi
- Start date
- Status
Yes but no.
no offense to the skaters I’m naming, but having Bortuzzo and Walman as a defensive pair is not depth.
blues stacked up front, not so much in the back.
Zezel’s Pretzels
Registered User
- May 25, 2019
- 709
- 1,088
For all the bitching about the defense, the Blues GAA is fourth best in the Western conference. One more goal allowed than Winnipeg, so second in the division.
Our goal differential is tied with Colorado for best in the Central (Colorado is finally heating up, as expected).
It’s not as bad as everyone makes it out to be.
Our goal differential is tied with Colorado for best in the Central (Colorado is finally heating up, as expected).
It’s not as bad as everyone makes it out to be.
Bluesnatic27
Registered User
- Aug 5, 2011
- 4,756
- 3,327
Thank you goaltending, because the defense really isn't helping.
Reality Czech
Registered User
- Apr 17, 2017
- 5,840
- 9,431
Obviously they are much deeper up front but I can't see how anyone would fault Army for building this roster. The defense is anchored by 3 top guys who all have different strengths, and since they each make $6.5 million of course they have to fill in the other spots with lesser guys. Perunovich was waiting in the wings, so any major addition might have blocked his path. There is nothing wrong with Bortuzzo as a 6/7 option.
Where would we be without Krug and Faulk? Top 4 defensemen don't grow on trees, so there is no guarantee better options would have come up. Or it would have at least cost us more assets via trade. Could the D group be better? Sure, but it's far from a "fail"
Thallis
No half measures
There's a 10 goal disparity between actual goals conceded and expected and we're still 10th in the league in goals allowed. We've gotten excellent goaltending this season, not good defense.
Zezel’s Pretzels
Registered User
- May 25, 2019
- 709
- 1,088
Just win, baby.
look, I’m a stats guy as well and I get your point. Maybe they’ve been lucky thus far. But the results have been just fine.
also notable - we’re done with Colorado and Vegas until the last two games of the season. Had our roadie with Carolina. Had an Edmonton game. We will get Tampa Bay out of the way shortly. The schedule should soften up considerably, right as we are getting healthy and have assessed which parts are truly spare (cough, Walman, cough).
Many observers have been clamoring for the Blues to get faster and better skaters. Well they have. The team speed is now a core strength of this team, and it creates matchup problems for opponents. Unlike the Hitchcock days when a less talented team had to grind the boards for possession to create scoring opportunities, this team frequently scores off the rush or by possessions where they carry the puck into the zone.
They’ve transitioned from ‘low event’ hockey to the opposite. I’m still unsure of what this will look like in the postseason. It’s curious, because this is the same core group of players who won a Cup playing a check and grind style. But I will say, the games are engaging. It’s a lot of fun to watch a team who can outskate most opponents, with enough skill in the line-up to do something threatening with the puck while doing so.
They’ve transitioned from ‘low event’ hockey to the opposite. I’m still unsure of what this will look like in the postseason. It’s curious, because this is the same core group of players who won a Cup playing a check and grind style. But I will say, the games are engaging. It’s a lot of fun to watch a team who can outskate most opponents, with enough skill in the line-up to do something threatening with the puck while doing so.
Sorry. I don’t but I to expected goals. Seems very random to me. How do you come up with it?
Brian39
Registered User
- Apr 24, 2014
- 7,580
- 14,247
Expected goals take into account the quality/location of each shot attempt based on the league-wide frequency of shots going in from similar locations and sequences.
Let's say goalie A faces 10 shots. All 10 are dump ins from the red line that happen to go on goal. Opposing goalie B faces 4 shots, all of which are 2 on 0s. Which goalie do you expect to allow more goals? If you are evaluating the team play in front of those two goalies, would you say that the team for goalie B did a better job than the team for goalie A since they are outshooting team B 10-4? Or would you say that the team with four 2-0s is outplaying the team that has just dumped the puck on net 10 times?
That is what expected goals are measuring: the quality/danger of the shot attempts. The actual formulas/calculations used are pretty damn complex and obviously the vast majority of shot attempts in hockey fall somewhere in between "dump on net" and "2 on 0."
Basic models only look at shot location. They don't factor in external factors and simply say "league-wide, a shot from this spot on the ice has X% chance of going in." They apply the percentage to every shot taken in a game to conclude that with an average performance by the shooters and goalie, you would expect Y goals to go in. For example, Team A takes 5 shots from an area with 20% goal rate, 10 shots from an area with a 10% goal rate and 20 shots from an area with a 5% goal rate. You would "expect" 3 goals to go in from those 35 shots, so the expected goals for that team would be 3.0 for that game. The formula would be (5 x .2) + (10 x .1) + (20 x .05) = 3 expected goals. If they had taken an additional shot from the area with a 10% goal rate, then the expected goals would be 3.1 for that game (5 x .2) + (11 x .1) + (20 x .05) = 3.1. Models like this are limited in value, but they paint a decent picture of teams that "force opponents to settle for bad shots from the perimeter."
Pretty much every model used today is substantially more complex than that. Modern goalies are good enough to stop the vast majority of unscreened shots when they are set and modern hockey offense is based around generating rebounds, forcing lateral movement, deflections, etc. A cross crease pass to the back door has a much higher chance of going in than a shot from the same location where the puck carrier walks the puck out from the corner and shoots on a goalie who is set on his post. To account for this, modern expected goal models use dozens of additional data points about what happened leading up to the shot attempt. In addition to weighting shots by location, they assign value to things that consistently increase the chance of shots going in. Shots taken directly after a zone entry are weighted as higher value because that corresponds to odd man rushes and breakaways. Rebound shots are weighted as higher value because they go in at a higher rate than non-rebound attempts. Shots immediately following a pass that crosses the "royal road" (essentially meaning a pass that forces a goalie to move side to side) are weighted higher. One-timers, deflections, screens, etc are considered.
The model is simply a way to use all of these data points to assign a numerical value to support what we already know from the eye test: That 2 on 0 has a much higher chance of going in than the dump in from the red line. The unscreened shot from the point probably isn't going in, but it has a much better chance of going in if it is deflected. Stuffing a puck into a goalie's pad when he's hugging the post probably isn't going in, but tapping a back door pass on net from the same spot probably is.
Expected goals are far from random. They define low, medium, and high danger chances based on tens of thousands of data sets that suggest how often pucks go in the net following a sequence of events. They are far from perfect and in small sample sizes are prone to being "off" because boiling down the cumulative actions of 10 players constantly moving within a 6,000 square foot offensive zone to 20 data points is imperfect. However, over large sample sizes, they are generally very good at determining the quality of chances teams are getting and giving up because they are based purely on averages. Teams that are constantly allowing a greater number of odd-man rushes, failing to break up back door passes, failing to clear rebounds, etc are going to have higher expected goals against numbers than teams who do all those things well, clear traffic in front of their goalie, and let teams settle for a bunch of low-danger shots from the point. As a goalie, I'll take 45 unscreened shots from the perimeter all day every day. I don't want to see 23 shots where 2 of them are breakaways, 6 of them are cross crease plays to the backdoor and I'm dealing with traffic all night.
This is the best intro-level explanation
And this is an insanely dense deep dive that genuinely requires a background in statistical analysis to fully understand.
If you are saying that you don't buy into expected goals as a concept, what you are essentially saying is that you believe every shot attempt in hockey is created equal. These values/inputs aren't just randomly assigned. The people who created these models look at an input and say "looking at hundreds of thousands of shot attempts across every NHL game for years, input X increased the chances of shot Y scoring by Z%. Given that, our model should value input X accordingly." There is plenty of room for debate about how well the expected goals models measure the quality of scoring chances, but there really just isn't a debate that they are at least generally an effective way to gauge whether teams are defending well or getting good goaltending. If your team is allowing an above average number of shots from locations/situations that consistently beat goalies at a higher rate than other shot types, then it is a pretty good indication that you are doing something wrong defensively. And if that isn't leading to a higher rate of actual goals scored against, then it is a good indication that your goalie is the reason why.
Last edited:
joe galiba
Registered User
- Apr 16, 2020
- 2,186
- 2,453
Somewhat good week for ex-Blues and I’d add on, Doug Armstrong. Doug isn’t going to be able to do what he does if he just fleeces everybody. Enjoyable for me that his trades aren’t screwing his gm bros.
The Good:
Tage Thompson expected goals has flipped positive and 10 of his 13 points are 5v5. His 8 goals are on a slightly high shooting % but not crazy.
Zach Sanford collects a hatty but more importantly - his ice time is growing from earlier games where he was hardly being used.
The Bad:
Dominick Bokk isn’t doing so hot to start the year in the AHL. GMDA may have dodged one there, but is also cool to see him come to NA.
Pietrangelos expected goals have flipped negative - rare for his career. Shrug.
The Ugly:
Poor Sammy Blais. From rangers fans I’ve chatted with they were all loving him. A bummer.
The Good:
Tage Thompson expected goals has flipped positive and 10 of his 13 points are 5v5. His 8 goals are on a slightly high shooting % but not crazy.
Zach Sanford collects a hatty but more importantly - his ice time is growing from earlier games where he was hardly being used.
The Bad:
Dominick Bokk isn’t doing so hot to start the year in the AHL. GMDA may have dodged one there, but is also cool to see him come to NA.
Pietrangelos expected goals have flipped negative - rare for his career. Shrug.
The Ugly:
Poor Sammy Blais. From rangers fans I’ve chatted with they were all loving him. A bummer.
Schenn back in today? I don’t think he went on LTIR did he, Although hasn’t he been out of the lineup long enough to qualify? Is there some Cap rules reason it’s advantageous not to use LTIR in this situation?
542365
2018-19 Cup Champs!
- Mar 22, 2012
- 22,540
- 8,993
He is back today. JR said he wanted to play Monday, but Chief was wary of it.
Brian39
Registered User
- Apr 24, 2014
- 7,580
- 14,247
He wouldn't be eligible to play today if he had gone on LTIR. You have to miss 10 games AND 24 days when you are put on LTIR. Today is day 20 and game 10 since he last played, so he wouldn't have hit either benchmark. If we had put him on LTIR, he wouldn't have been able to play until the Tampa game on 11/30.
Brian39
Registered User
- Apr 24, 2014
- 7,580
- 14,247
Most do not as a player's ability is largely baked into the inputs (area where the shot comes from, puck movement prior to the shot, taking the shot off the rush, etc). Think about the highlght reel goals scored by the league's best players. They are breakaways, odd man rushes, pretty pass plays, etc. Barring a handful of incredible shooters, the great players score because they generate quality chances and not because they convert poor chances at a higher rate than others. As a team, you should give up a low percentage shot from a star instead of a high danger chance from someone else every time. Draisaitl is the league's best shooter over the last several years. Look where his shots and goals come from:
Shots:
Goals:
He shoots a ton from basically everywhere in the zone, but only has consistent scoring success from the high danger area. Preventing him from taking a low danger shot at the expense of allowing a lesser player to get a high danger one is not good defense.
Ovechkin's heat map widens out to cover literally the entire zone:
Shots:
A good defense lets him take non-one-timers from the perimeter all day because look at how useless those perimeter shots are at actually going in. Ovi wants to blast away from above the faceoff circle? Great. Just make sure your goalie sees it and it isn't a one-timer.
Goals:
Anyone who has watched Ovi knows that the large majority of the red bunches at the top of the circles comes from him blasting home one-timers (which would be labeled a high danger chance already).
It doesn't matter how good of a shooter you are. A shot from distance that doesn't have any of the characteristics of a quality chance has little chance of beating an NHL goalie. These guys score so damn much because they create tons of chances that the models expect to go in.
Here are the top 10 players in individual expected goals since the start of the 2019/20 season: Matthews, B Tkachuk, Draisaitl, Aho, Ovechkin, Connor, McDavid, Tavares, Zibanejad, and M Tkachuk. That group makes up 7 of the top 10 goal scoring leaders in that stretch. The other 3 are Pastrnak (16th in expected goals), Marchand (30th in expected goals), and Debrincat (19th in expected goals). Again, none of these models are perfect, but they do a pretty damn good job. You have guys like the Tkachuks who have inflated expected metrics because they jam a lot of low-percentage rebounds into the pads of goalies, but for the most part the models are good.
Last edited:
execwrite1
Registered User
- Mar 30, 2018
- 1,523
- 1,472
Per Lou
Ivan Barbashev-Ryan O'Reilly-David Perron
Brandon Saad-Brayden Schenn-Jordan Kyrou
Pavel Buchnevich-Robert Thomas-Vladimir Tarasenko
Klim Kostin-Tyler Bozak-Oskar Sundqvist
Marco Scandella-Colton Parayko
Torey Krug-Justin Faulk
Niko Mikkola-Scott Perunovich
Ville Husso will start in goal;
Beauterham
Registered User
- Aug 19, 2018
- 1,718
- 1,557
I really like those forward lines. If you look at the first 3 lines, I haven't got a clue which line should be viewed as our 1st, 2nd or 3rd line. Should be a nightmare for an opponent to match their lines against us.
stl76
No. 5 in your programs, No. 1 in your hearts
- Jul 2, 2015
- 9,461
- 9,042
Ovechkin's heat map widens out to cover literally the entire zone:
Shots:
Literally lol'd.
BadgersandBlues
Registered User
- Jun 6, 2011
- 1,893
- 1,420
stl76
No. 5 in your programs, No. 1 in your hearts
- Jul 2, 2015
- 9,461
- 9,042
I just wasn't expecting it, caught me off guard
shpongle falls
Ass Möde
Tough stretch coming up for the Blues, we get the Blue Jackets then the Lightning twice then Panthers twice.
Chicago and Detroit was tough stretch way we are playing. If we don’t pick it up we gonna have worse time with sunshine state than anyone since al gore.
Important stretch here. If the Blues only win one or two of those 5 games, there will be changes.
ChicagoBlues
Terraformers
- Oct 24, 2006
- 15,756
- 6,617
- Status
