This is the early version of Edmonton’s training camp roster. There are 45 names here and I’ll guess the final total stretches past the 52 names on last fall’s list. The Oilers of this era generally invite players who aren’t really strong contenders for NHL work (Jake Virtanen) along with a plethora of limited if interesting AHL contracts (Dino Kambeitz). I’m always hopeful the club will invite some CHL free agents who could one day play in the league.
The final frontier for the Oilers is to understand winning in the playoffs often means making the simple play, the safe decision, and not forcing offence where there is high risk.https://t.co/l6p9QgV7ND
LD Isaac Menard, Shawinigan Cataractes (QMJHL). Skilled and effective, he’s 19.
LC Justin Cote, Drummondville (QMJHL). Under the radar, but he has skill. July 2004.
AHL CONTRACTS WORTH INVITING
This is always a sticky wicket because not everyone on an AHL contract has a future at the highest levels of hockey’s minor leagues. It gets crowded in a quick hurry. Here are the AHL contracts, and the ones I believe should be invited and are worth tracking this season.
G Tyler Parks, 31. He has a long pro career that includes six ECHL seasons and three in the AHL (but just 27 games). He played in 21 games a year ago (.888 SP) with the Tucson Roadrunners, and will likely tandem with Ryan Fanti in the ECHL (Fort Wayne Komets).
LD Xavier Bernard, 23. He was acquired for winger Graham McPhee last season. Red Line Report had him No. 37 overall in 2018, saying “big, strong, smart and reliable” and all of those things will be in fashion this fall for the organization. He was 1-0-1 in 12 games with the Condors last season, his even-strength goal differential (4-5) didn’t hurt or help the team. He didn’t play much (about 10 minutes a night according to Eric Rodgers’ estimates) and will probably spend most of the season in Fort Wayne (420,000 people in the Metro area).
RD Connor Corcoran, 22. He was drafted by Vegas in 2018’s fifth round, and is a defense-first blue. However, he has seven goals over 25 games in three cups of coffee with Henderson Silver Knights (AHL) so there appears to be some range here. I think this was a solid signing. We’ll see.
LD Jake Johnson, 24. He played for the national championship Quinnipiac side a year ago, his outscoring numbers shine like a diamond. Looked good in 10 games with the Fort Wayne Komets at the end of the year, that probably got him this contract. He is average in size.
LD Alex Peters, 26. To my eye, he was the best of the AHL contracts a season ago. Through the end of February, his on-ice goal share at even strength was 38-23 (62 percent) but he struggled a little (11-15) through the end of the regular season. He’s a lefty, so there’s no room at the inn for NHL contracts, but he’s probably the best actual player among the AHL deals.
RW Ethan De Jong, 24. A two-way winger from North Vancouver, De Jong ended his college career in fine style with a national championship. He has a good shot and some offensive acumen, but I think the big selling point is his responsible play while in pursuit of the puck. His NHLE (22) is good and if he plays a 200-foot game Bakersfield’s coaching staff will play the hell out of him.
RW Dino Kambeitz, 23. I confess to liking this player way more than his talent merits. He plays a determined game, and simple. Kind of like Patrick Russell if you recall him. I don’t think Kambeitz ever plays in the NHL but I’d love to see it. He’s 6.02, 210 and has scored 9-12-21 per 82 AHL games so far in his career.
RW Cameron Wright, 24. He has size (6.01, 200) and coming off a 29-34-63 (in 64 games) ECHL season. I have no idea how he gets into AHL games, with the depth chart so crowded with RW prospects.
SAIL ON DOWN THE LINE
LW Tyler Benson, 25. Sometimes a man needs a second opinion. My Dad always said a change is as good as a rest (Dads say the darndest things) and Benson signing with the VGK affiliate in Henderson is certainly change. He has talent (per 82 AHL games: 16-51-67) but needs more speed and torque on his shot. He is an NHL passer with creativity. He has to find a way to get into NHL games with skill. Patrick Maroon did it, perhaps this young man will, too. I wish him well.
LD Yanni Kaldis, 27. Chaos blue had an offensive impact (8-32-40 per 82 AHL games) but the defensive side of the game slipped in 2022-23 (33-39, 46 percent). Signed with Moscow Dynamo (KHL) and he’ll flourish there based on his skill set.
LD Darien Kielb, 24. He signed with Fort Wayne so is technically in the system and AHL games are possible. I liked him with the Condors last season, a little surprised he didn’t get the AHL portion of the contract. His even-strength goal share (17-12, 59 percent) and size (6.02, 183) make him an attractive Bakersfield option if required.
RW Drake Rymsha, 24. He was chosen in the fifth round of the 2017 draft by LAK, got into one NHL game with them. Played four games with the Condors a year ago but was a point-per-game scorer in the ECHL with Fort Wayne. He has signed with HKM Svolen in Slovakia for the coming season.
Noah Ganske, 24, 6’7 RHD might also be in camp, recent Komets signing out of NCAA Div III, very good player who attended last seasons’ Development Camp. I thought he was the best Dman at that camp.
Great stick, better skater & passer than VD, IMO.
Last edited 1 year ago by €√¥£€^$
Reja
July 11, 2023 8:21 pm
What’s the over-under on Downtown Brown with the Leon push 23>30 Goals.
OriginalPouzar
July 11, 2023 5:08 pm
Don’t get me wrong, I stand my my strong words that Evan Bouchard is an elite offensive d-man, both on the PP and at evens – his production in 2021/22 and over the aggregate of the last two seasons put him there.
At the same time, when looking just at this past season, the most relevant for contract discussions K’Andre Miller had more points than Bouch 5 on 5 (6) – he also let his team in 5 on 5 TOI/G and played almost 2 minutes per game on the PK.
There are some real objective traditional metrics that Holland can point to for Bouch to come in under.
And then Bouch can pop 77 points this coming year!
Always wondered about the luck of schedule difficulty in terms of playing teams when they are hot/cold. Was always too busy to actually execute any analysis on that but with the advent of AI analysis its pretty simple.
The methodology I used is pretty basic, I asked for the winning % of 8 surrounding games (4 before/ 4 after) for each opponent, and then calculate the average winning percentage total for each team’s opponents through the 2022-23 season.
The results were interesting, Toronto had the toughest, Vegas second, Calgary 5th. Edmonton was near the bottom 25th.
Team NumGames
———————————–
Toronto Maple Leafs 33
Vegas Golden Knights 25
Buffalo Sabres 21
Edmonton Oilers 20
Ottawa Senators 20
Washington Capitals 20
Calgary Flames 19
San Jose Sharks 19
Anaheim Ducks 18
Arizona Coyotes 17
Columbus Blue Jackets 17
Detroit Red Wings 17
Dallas Stars 17
Los Angeles Kings 17
Vancouver Canucks 16
Boston Bruins 16
Minnesota Wild 16
St. Louis Blues 16
Pittsburgh Penguins 16
Seattle Kraken 16
Montreal Canadiens 16
Chicago Blackhawks 16
Colorado Avalanche 16
Carolina Hurricanes 15
Nashville Predators 14
Florida Panthers 14
Tampa Bay Lightning 14
Philadelphia Flyers 13
New York Rangers 11
New York Islanders 11
Winnipeg Jets 10
New Jersey Devils 10
This is interesting but any idea how Edmonton was near the bottom in the winning % list but near the top of the 2 or fewer losses list? (TOR, VEG, BUF were at the top of both, and the two lists should be roughly correlated).
Would a 4 or fewer losses list be a helpful sanity check (should correlate even better with winning %)?
Here’s possibly a more informative framing of the results, doing a +/- of total games against hot or cold opponents. Its ranked as most games against hot teams, so Toronto played 17 more games against hot teams than against cold teams. Edmonton were par.
Toronto Maple Leafs +17
Vegas Golden Knights +13
Buffalo Sabres +7
Anaheim Ducks +7
Ottawa Senators +7
Los Angeles Kings +5
Washington Capitals +6
Calgary Flames +6
San Jose Sharks +4
Detroit Red Wings +3
Montreal Canadiens +3
Arizona Coyotes +2
Minnesota Wild +2
Colorado Avalanche +1
Florida Panthers +1
Edmonton Oilers 0
St. Louis Blues 0
Chicago Blackhawks -1
Nashville Predators -1
Vancouver Canucks -3
Dallas Stars -3
Carolina Hurricanes -4
Columbus Blue Jackets -5
Boston Bruins -5
Pittsburgh Penguins -5
Seattle Kraken -5
Winnipeg Jets -7
New Jersey Devils -8
Philadelphia Flyers -8
Tampa Bay Lightning -9
New York Rangers -11
New York Islanders -18
Looking at the top of this chart (the +/- one) and it’s apparent who the hot teams were. All the teams with the most games against hot teams were in either the Atlantic or Pacific divisions. Boston was hot all season and Edmonton went on a massive run in the 2nd half.
Im using the pro version of GPT4 which now has functions (beta). It allows you to import data sets and ask questions. It basically writes python code using pandas, and executes that code on the uploaded dataset.
So you just describe the calculation in text and it pulls the data and does it?
For example: “For each nhl team during the 2022-2023 season, calculate the game win percentage of a team’s opponent for the 4 games preceding and following a given game divided by the opponent’s win percentage for the year. Average this value across all games for a team. Do this for each team and stack rank the result outputting the computed value for each team.”
Pretty much yeah. The prompting can be a skill in itself, but in terms of data science, it’ll do whatever you want to know based on human language questions.
I wouldn’t really call that AI, per say. It’s using AI to understand human language and do some math calculations not computing something using AI.
An example of AI computation would be inputing a bunch of features into an AI model and asking it to do a regression (reason) about which features are more important for answering a question by learning from training data. I suppose you could argue that’s just some math calculations too :-).
Not sure what you’re saying here. It’s an LLM… which is a generative AI.
To expand on how this is working, GPT4 (LLM) reads my request, imports the data, and writes an actual python script that can accomplish the request, then executes the script to calculate the outcomes. So the entirety of the heavy lifting here is 100% generative ai
Yes but it’s like saying I asked chat GPT to compute the sum of the points scored by the top 5 players for each team. AI was used but the calculation itself was rudimentary and not insightful. Any middle school kid could work that out. I mean to apply AI analysis to the hockey stats themselves to get insights that can’t be derived by hand.
Agreed. Also you (Big Mac) said its using AI to understand language, so it sounds like you think its AI, but then you say its not computing something using AI.
Actually understanding human language is quite important for AI I would say. I also think that, by my rudimentary understanding, that cognitive science has been an important feild towards AI progressing, and part of that is understanding how humans think, which is not through computations and reasoning (we are appearently not so good at actual logic) but through conceptual metaphor. But im blathering now.
You’re use of AI has me thinking about my own data analysis. How can i learn more about this? Is there some way it can look at enormous swaths of data that i have stored (in excel) and analyze it for me?
if you want to experiment for free, I’d recommend downloading a free python IDE like PyCharm, and then using free ChatGPT membership, write prompts that ask it to create scripts that analyze your data.
The easiest way is to buy a GPT pro membership (around 25 bucks) and use the new Code Interpreter mode, which allows you to upload your CSVs and it can inspect your data instead of you needing to describe the contents. Then it creates the python scripts and runs them for you, presenting you with the results
Sure AI is involved in understanding your request and your question and analysis was interesting. Here’s an example of what I’m trying to describe. JFresh has a model that computes the relative value of a player. It has it’s own weighting of different attributes. Can you use AI to learn a better model for valuing players? So for your example, instead of it calculating the relative strength of schedule based on your formula, learns it’s own (should be better) formula for computing relative strength and applies it.
Expanding my example a bit. For calculating strength of schedule, your model doesn’t take into account: date, day of the week, start time of the game, back to back, backup goalie, player injuries or slumps, coaching change, travel schedules, all-star break, current standings, weather, elevation, etc. An AI can take all those as inputs and determine a relative importance of each variable (feature) and build it’s own model to calculate strength of schedule. The more interesting inputs (features) and the larger the sample data-set the better the model.
Yes an AI could do all that. You are conflating the idea of more intricate analysis, with AI. A generative AI will do whatever you want. You need to ask it to consider those things, like I mentioned earlier you can get very intricate with the prompts.
You’re more decribing a Narrow AI / specific machine leaning scenario, which would require a massive dataset and iterate every conceivable path to calculate and test probability.
I guess the takeaway is generative AI isn’t sentient. It’s not going to ask its own questions and find the answers itself. It needs prompting. Agents are a layer that are closer to what you’re describing, that can use generative AI to ask itself questions to help generate more multi step solutions.
But that’s still prompting… automated prompting.
Regardless it all goes back to the point, you’re confusing the simplicity of my request as ‘not being ai’
Corcoran, Peters and De Jong are my ones to watch from that list. They could all do well (or well again in Peters case) with the Condors. De Jong in particular interests me as a potential long shot to make the NHL.
Numenius
July 11, 2023 2:37 pm
Interesting argument by Mathew Panchyshyn on Twitter: “Broberg is leading his draft class in ability to play.”
— Mathew Panchyshyn
These are the D drafted in 2019 to play +25 gp in 2022/23…
-Broberg -Byram -Seider -York -Soderstrom -Xhekaj
Conclusion:
Overall: Broberg is leading the draft class in the ability to play. This does consider QoC and QoT. I would be very hesitant to label anything on Broberg other than “Pending” when trying to understand what he is. So far, he’s been nothing but excellent. https://twitter.com/mathewjdp/status/1678508895705214977
—
Coaches choice. I am perplexed on how Oiler coaches use players including pre Woody
Bouchard also has team top stats in a few of them. Yes he makes mistakes. Bro by what that tweet shows is really good at the things that matter most to be a really strong NHL D. The one thing he had issues with was puck retrieval. So pair him with a guy good at it. Both Nurse (historically) and Ek are. I am pretty sure that is also an issue of not enough games, especially consecutive, so he can get used to it at the NHL level
To me players are what they are pre-draft. The players that don’t hit expectations have fatal flaws that were overlooked or missed (can’t skate at an NHL level, no hockey sense, weak puck skills, too small to handle it, are the usuals), assuming no big injury, attitude, or off ice issues
Bouchard is exactly what they drafted. He isn’t Hedman, but he is close to being a very good top 4 D that is a two way player, offensive leaning, as he was in junior. He was the go to all situations D on his junior team
Bro was an elite D for Sweden, captain, their most relied upon D. Able to play men’s league at a young age. He got hurt a few times which is the biggest worry and it set him back, but he’s that player still, he has all the attributes of a strong NHL player, he just needs to play
What perplexes me is that the risk of using the younger players is a hill too far. Because the guys they use instead aren’t as talented, make mistakes too, and really hamper pairs and team play because they have such a limited range. Ceci has repeatedly been shown to only have a stronger trait in being a decent in zone D, and he wasn’t even that last season
The real risk is not being able to make some astute and hard cap decisions and sign these two early and long. It will be a lot more expensive later which will reduce the quality of the team then
To me if Bouch will go 5-6 x 8 you do that all day long. Just don’t give him protections so you can change your mind later. Try it with Bro in a bit. If you can’t replace Foegele with as useful or near a player at 1M and find somehow a decent RD to add depth that is cheaper than Ceci, then you shouldn’t be an NHL GM IMO. These are supposed to be the easiest things to do, find bottom roster players
I’ve played around with a lot of Evolving Wild’s data.
The basic problem with all of this data (or Natural Stat trick) is that there are different minutes played by players on different teams.
It doesn’t matter if you’re using expected goals, corsi, GF%, Danger Fenwick, or whatever…
if you compare, for example a guy who’s playing sheltered (7d sheltered) 3rd pairing minutes (Broberg) to a guy who’s playing top pairing minutes on the Red Wings (Seider), your results and conclusions are going to be completely invalid.
This is where PuckIQ is unique and stands alone.
At the end of the day, Broberg is playing 11:22 per night (5v5) on the Oilers and Seider is playing 16:10 against the toughest comp on his team.
The moment you try to tell me your stats tell you Broberg (a 12:36 minute per game defensemen) is better than Seider (23 minutes per game all states), you’ve lost me.
It’s the age-old 3rd pairing d problem we’ve stumbled over for years.
Guys like Matt Benning having rock solid fancy stats leaving posters here scratching their heads trying to figure out why the Oilers coach is too stupid to play them more…
not realizing that the coach is working his ass off to hide them. using on the fly starts (or reduced minutes) to keep them away from good players or playing them against good players when the puck is headed in the direction.
The last point is definitely true. But one issue with this is the coaches are so damned risk averse sometimes that they’re either unwilling to give more mins or tougher matchups to the younger players or won’t live with their mistakes when they do. But they’re willing to live with repeated mistakes from veterans that should know better.
A good coach needs to not just shelter young players but to find them consistent opportunities to elevate their game even when they’re making mistakes. If this was done more early last season for Broberg (when healthy) or two years earlier for Bouchard the team would be in a better position to make the call on going long, bridging or trading the player.
Although it’s go time for the team, they are also good enough to win at a playoff pace early in the season even with a couple of rookies making more mistakes than you’d like. It may save you assets trading for players at the deadline as well. This year I really hope Woody takes this approach with Holloway until the deadline. It may pay off in spades in the playoffs.
Yes, sheltered vs unsheltered is the obvious objection.
Though Mathew does say the data already factors in Quality of Comp and Quality of Team. How reliably it does that is is hard to say.
Someone asks him: “How sheltered is each player respectively? Because something tells me Mo plays harder minutes than Broberg”
Answer: “Of all 6. Mo plays the most against the Elites, and Broberg plays the least. This takes that into account. Mo did not well this year against Elites (like most don’t) and was avg at best against other competition. Broberg was exceptional against competition other than Elites.”
Broberg has seemed unusually sheltered this last season. Seemingly more than he should have been.
If I were to guess, a big reason is to give him extra time to learn to protect himself. His ability to sense danger when he first entered the NHL was poor.
I don’t have twitter at work here and you can no longer access it without logging in. He didn’t say where his data is from, but I am assuming it’s from Evolving Hockey.
JP and I went over a bunch of EH player ratings several months ago now. While they do claim to adjust for QOC and QOT, it’s not much. Their total ratings, way over value 3rd pairing offensive d, for example.
Even with PuckIQ, it’s really hard to compare a 3rd pairing d to a top pairing d.
One issue, Tyler Dellow exposed before he got hired was the 3rd pairing d problem.
Coaches shelter their 3rd pairing d, both using toi and on the fly starts.
Let’s say you have a Matt Benning and you’re a coach. You don’t want him out there against Jordan Kyrou, but you only have so much control. But when your team has control of the puck in their zone and they start heading to the other team’s end, Adam Larsson heads to the bench and you throw Benning over the boards. Now Jordan Kyrou is on the ice against Matt Benning, but the Oilers have Mcdavid on the ice and possession of the puck.
You get enough of these situations when Larsson off and Benning on and it starts to look in the data like Benning is some sort of Elite comp savant, even though it’s only because he’s shifted on the fly against elites when we have the puck, not because he’s actually making a contribution.
I just looked at PuckIQ, and that supports your point.
For example, Broberg comes out as much better than Nurse in DFF against Middle and Gritensity, and yet against Elites far worse.
Similarly compared to Ekholm, Broberg rates about the same DFF against Middle and Gritensity, and far worse against Elites.
Yet he’s surely not near their level in reality, as their superiority against elites proves.
You can’t disregard comparison against elites and only compare results between lower comp and think you have something there. That seems to be what Mathew has done.
With defensemen, the first thing I look at are toi/g and on the fly starts/60.
Both of these stats are determined by the coach. They play the players they think are the best the most and the best players have the lowest OTF starts/60.
That usually indicates that the coach is trying to control who they play against or when (on the fly when the puck is headed in the right direction).
Opposing teams also try to target 3rd pairing d like when Broberg got the phantom holding call against Eichel in game 5.
How players like Broberg show up against elites probably depends more on how successful other coaches are getting the Eichels out against them on the attack vs how good their own coaches are getting them out against the Eichels while on the fly rather than truly indicating their innate ability of how good they actually play vs elites.
In other words, how a sheltered player plays against elites probably isn’t indicative of their actual ability. If they were actually good playing against elites, they wouldn’t be sheltered.
I’d have to look it up (I don’t hve time tonight but remember Oscar Klefbom had a ZS push and a pile of OTF in his first full season. He turned out okay, and Broberg has been a fairly close match so far.
If we’re going by his 2014-15 season, Klefbom actually had the lowest OTF starts/60 on the team. He was also #1 in toi/g. (draft +4).
Edit-> in the 2013-14, the Oilers had 11 d who played over 100 minutes at 5v5. Klefbom was 6th in toi/g and 6th in otf starts/60. That was his draft + 3 season.
He did, but the Oilers were the 28th team out of 30 that year.
If you look at #1 d on bad teams, they’re going to get torched against elites.
I think maybe Hampus Lindholm is a #1 who played on some bad teams holding his own against elite comp for a few years, but he fell back to earth before getting traded to Boston.
Yes but the point is he wasn’t playing those minutes and in those situation because he was ready in draft plus 3 but because of the team being awful and, consequently, its not really fair to use him playing those minutes as a comparison to Broberg’s deployment at the same stage, right?
It seems to me like you are making a strong argument that we can’t really analyse Dmen with any analytics. If their stats are dependent on their TOI and on the fly shifts and if TOI and OTF shifts are based on coaches’ subjective opinion of the player’s ability, then aren’t we just analysing the coaches’ opinion of the players?
I would say that defensemen are more of a challenge to analyse with analytics than forwards.
If you want a decent list of who the best forwards are in the league, this isn’t a terrible starting point (here).
I’m not saying you can’t use analytics, but you have to be aware of potential problems. The 3rd pairing OTF issues was identified by Tyler Dellow, not me.
It’s easier to compare say a Nurse to a Makar since you’re comparing two #1 d who each play on competitive teams.
If you are going to compare two players either forwards or defensemen, you have to look at their stats in the context of what type of minutes they play. Context is always important and QOT weighs more than QOC.
If their stats are dependent on their TOI and on the fly shifts and if TOI and OTF shifts are based on coaches’ subjective opinion of the player’s ability, then aren’t we just analysing the coaches’ opinion of the players?
Coaches do a decent job handing out icetime except when they’re William Karlsson’s coach on the Columbus Blue Jackets.
If you don’t like TOI as a metric, you probably shouldn’t use PuckiQ data. 🙂
Klefbom’s offensive zone starts in his year 4+ draft (7.87/60) and Broberg’s in his 4+ draft year (8.14/60) are not dissimilar.
One of the things we should take into account is the lack of a Nurse or an Ekholm on the 2014-15 Oilers. That was death valley and had much to do with Klefbom’s deployment (although I did feel Dallas Eakins was innovative in handling him).
Victoria Oil
July 11, 2023 2:03 pm
NYR sign K’andre Miller for 2 x $3.87. Bouchard has more offensive upside, although they both have the same career high in points (43). Miller is better defensively.
This past season, Miller’s coaches trusted him to play 500+minutes against players classified as elite by Puckiq, 2nd most on the Rangers. Bouchard got about half the minutes (246) against the same category of players. Coaches deployment is one significant indicator and Miller’s minutes against elite have been high for a few seasons.
Its a long summer when your team is actually good, roster looks good. Even longer till they get back to playoffs.
Boil-in-the-Oil
July 11, 2023 1:00 pm
Early 70’s I recall hearing Sailin’ Shoes on CKUA, loved it. Drove straight to Sub Records, it wasn’t in stock so I bought Feats Dont Fail Me Now. Looking back, imo for me, they were easily the best band of the 70’s. Lowell George was truly a great song-writer & musician (from The Mothers of invention to Little Feat) who was lost far too early.
Let the fast feet in fast boots of our Oilers conquer all who challenge us. Dont fail us now!
Chico21
July 11, 2023 12:05 pm
Ah Patrick Russell couldn’t hit water if he fell out of a boat.
Couldn’t hit the broad side of a barn with a handful of rice.
Oh but the determination.
I assume you are joking. But this kind of sentiment demonstrates such a myopic understanding of analytics and the value of analytics.
It’s also funny that folks always try to clown on one specific decision maker within an organization and claim to know they were the specific individual who advocated for a transaction. But go off!
No idea if OP is joking or not but to be fair it is pretty standard fare by a large number of posters on this blog to credit Tulsky with every astute move that franchise makes with just as much or little information as to who is driving the decisions.
Not sure how his comment demonstrates a myopic understanding of analytics though.
With respect, many long-time community members laud Eric Tulsky and the Canes’ organization use of analytics and speak to a massive gap between the Oilers and Canes management and front office abilities largely due to this.
I can’t remember a Cane transaction that wasn’t lauded for its reliance on analytics.
Just yesterday there was a piece at The Athletic that suggested the Oilers didn’t have an analytics team that could provide Jay Woodcroft with the Puck IQ data and Woody couldn’t adjust and it cost the team the series.
The initial comment seems prudent for this community.
As luck would have it…the Athletic has just summarized which teams have been the most successful in the salary cap era when the game changed drastically.
The Hurricanes have the 10th best record in the league.
Maybe their analytics department has determined they need some snot to catch up with the Oilers? I mean, since Woodcroft and Manson have taken over the Oiler, they have a better points percentage than the Canes.
All that time since the game changed with the cap era but yet they are currently not even further ahead than the, from accounts, dark-age Oilers.
Meh – its been 6 years since a the Cup Winner had “a McDavid” (Ovie) – its not needed to win.
The Knights just won by trading for their two most impactful players (Eichel and Stone), neither of which are near generational.
If the Canes wanted, they could have Tulsky and a near generational player – they could likely acquire Auston Matthews but it seems Tulsky doesn’t want that.
I find it very improbably that Woodcroft doesn’t receive analytics and that the Oilers didn’t have the ability to get him something like Puck IQ info in real time and it cost the Oilers the cup.
They are thinking outside the box and incorporating a new stat, finger bites/60.
I imagine they have done the analysis to determine that Brendan, at 800k, may be a good investment assuming he bites off a few fingers every now and then. They must have ballparked the amount of punishment Brendan would receive for reoccuring offenses and probably determined he could get away with biting the fingers off of a few divisional teams players, and maybe even a few Stanley Cup contenders fingers as well.
Holland may want to consider it as well. 800k would be a small price to pay if for example, you knew the player could possibly bite off one of say.. Kopitars fingers, Stones fingers, MacKinnon’s fingers.. wouldn’t you agree?
buck yoakam
July 11, 2023 9:03 am
your daddy says I’m no good
your mama says keep away
I gotta tell you truthful girl, you can never make me stay…
OriginalPouzar
July 11, 2023 8:59 am
Here is hoping that Tyler Benson is able to re-set in a new org.
Kudos to the young man for changing his game to try and make it at the NHL level. He wasn’t going to make it as a “producing winger”, at least on this Oilers team, so he increased his “jam level”. He always had some jam but he started to focus on that area and being “a pest”. It had some value at the NHL level but not enough.
For me, he never re-found his true game at the AHL level last season after re-assignment. He was not close to the dynamic offensive player at that level that we’d seen over the years.
I hope he gets back to being a dish-master and high end offensive producer at the AHL level and earns himself a termination of his AHL deal and another NHL contract.
The NYT is a formidable institution, but I’m not convinced their sports page was their leading component. Giving way to The Athletic is a recognition of comparative advantage. If it gets LT more eyeballs I’m all for it.
As always, a great article to read, however, I don’t understand how a tandem pretty much alternates every several games then decides to stick with one goalie. I got the thinking early in the LA series. But I think the equation should have changed after the Campbell showed he found his footing and Skinner wasn’t setting the world on fire. I think the smart approach would be to put trust in the tandem and not just one goalie. Both seem to be better when given a night off to reset, so the best way to help Skinner would’ve been to give him a night off.
Also, what are your thoughts on ability to implement systems? This team is just a better team in general so of course it has a good win record, but I think the best system they have played was under the early games of Tippett. Everyone in the zone came back deeper and there seemed to be layers to the defence. The defence has previously been too shallow (not to mention the forwards) to play tight zone defence, but is that still the case?
I think if this team could play a style closer to vegas then they torch the league. Is your thought that Woodcroft has evaluated the alternatives and what they play is somehow the best approach? I’m not saying he couldn’t coach it but some coaches don’t, so is that a limitation of Woodcroft and might hold this team back? (Not implying it is, just an open question)
Woody cost us the series with his reluctance to start Campbell. If Maurice stays with Lyon they get beat by the Bruins. Sometimes a Coach needs to make hard decisions
On the flip side, Campbell had lots of minutes in the regular season to establish himself as a viable alternative. And he failed to do so, over and over again.
Maybe it would have gone differently if Bjugstad was not matched against Eichel, or if Broberg had played instead of Vinny.
I agree. Crazy Coach also commented on it. He calls it ‘box plus one’. Make Connor the one, it’s easy to use and would play to Nurse’s strengths, and Ekholm’s. Easier for Bouch
Connor as a threat to get a puck and take off. They would be very hard to beat and the goalies would play far better with consistency and no loose opponents in front of them
The other side though is that the Oilers have been far better under Woodcroft than any other coach in recent memory. The personnel has also been strengthened, and I believe in the past you’ve attributed much or all of the team improvement to McDavid and Draisaitl maturing.
There’s some truth to those things, but it’s not cut and dried that the system they’re using is ‘broken’, or that they would necessarily perform better using the box plus one or whatever else. It’s very likely IMO that that the system in place has contributed to the #2 regular season record and the 3 playoff round wins since Woodcroft and Manson arrived.
And that’s not to say the system shouldn’t be tweaked, or added to, or even scrapped entirely. Just that it has actually served the team pretty well over the last year and a half and we simply don’t know if something different would be better.
Noah Ganske, 24, 6’7 RHD might also be in camp, recent Komets signing out of NCAA Div III, very good player who attended last seasons’ Development Camp. I thought he was the best Dman at that camp.
Great stick, better skater & passer than VD, IMO.
What’s the over-under on Downtown Brown with the Leon push 23>30 Goals.
Don’t get me wrong, I stand my my strong words that Evan Bouchard is an elite offensive d-man, both on the PP and at evens – his production in 2021/22 and over the aggregate of the last two seasons put him there.
At the same time, when looking just at this past season, the most relevant for contract discussions K’Andre Miller had more points than Bouch 5 on 5 (6) – he also let his team in 5 on 5 TOI/G and played almost 2 minutes per game on the PK.
There are some real objective traditional metrics that Holland can point to for Bouch to come in under.
And then Bouch can pop 77 points this coming year!
Good grief.
Thought you two weren’t allowed to communicate? Time out?
Always wondered about the luck of schedule difficulty in terms of playing teams when they are hot/cold. Was always too busy to actually execute any analysis on that but with the advent of AI analysis its pretty simple.
The methodology I used is pretty basic, I asked for the winning % of 8 surrounding games (4 before/ 4 after) for each opponent, and then calculate the average winning percentage total for each team’s opponents through the 2022-23 season.
The results were interesting, Toronto had the toughest, Vegas second, Calgary 5th. Edmonton was near the bottom 25th.
results visible here:
2 or fewer losses in their surrounding 8 games
Team NumGames
———————————–
Toronto Maple Leafs 33
Vegas Golden Knights 25
Buffalo Sabres 21
Edmonton Oilers 20
Ottawa Senators 20
Washington Capitals 20
Calgary Flames 19
San Jose Sharks 19
Anaheim Ducks 18
Arizona Coyotes 17
Columbus Blue Jackets 17
Detroit Red Wings 17
Dallas Stars 17
Los Angeles Kings 17
Vancouver Canucks 16
Boston Bruins 16
Minnesota Wild 16
St. Louis Blues 16
Pittsburgh Penguins 16
Seattle Kraken 16
Montreal Canadiens 16
Chicago Blackhawks 16
Colorado Avalanche 16
Carolina Hurricanes 15
Nashville Predators 14
Florida Panthers 14
Tampa Bay Lightning 14
Philadelphia Flyers 13
New York Rangers 11
New York Islanders 11
Winnipeg Jets 10
New Jersey Devils 10
This is interesting but any idea how Edmonton was near the bottom in the winning % list but near the top of the 2 or fewer losses list? (TOR, VEG, BUF were at the top of both, and the two lists should be roughly correlated).
Would a 4 or fewer losses list be a helpful sanity check (should correlate even better with winning %)?
I wondered the same thing, and asked GPT. It’s reason was EDM played 20 games against teams with 6 or more losses in the surrounding 8.
0 New York Islanders 29
1 Tampa Bay Lightning 23
2 Columbus Blue Jackets 22
3 New York Rangers 22
4 Philadelphia Flyers 21
5 Seattle Kraken 21
6 Pittsburgh Penguins 21
7 Boston Bruins 21
8 Edmonton Oilers 20
9 Dallas Stars 20
10 Carolina Hurricanes 19
11 Vancouver Canucks 19
12 New Jersey Devils 18
13 Winnipeg Jets 17
14 Chicago Blackhawks 17
15 St. Louis Blues 16
16 Toronto Maple Leafs 16
17 San Jose Sharks 15
18 Nashville Predators 15
19 Arizona Coyotes 15
20 Colorado Avalanche 15
21 Detroit Red Wings 14
22 Buffalo Sabres 14
23 Washington Capitals 14
24 Minnesota Wild 14
25 Montreal Canadiens 13
26 Calgary Flames 13
27 Ottawa Senators 13
28 Florida Panthers 13
29 Los Angeles Kings 12
30 Vegas Golden Knights 12
31 Anaheim Ducks 11
Here’s possibly a more informative framing of the results, doing a +/- of total games against hot or cold opponents. Its ranked as most games against hot teams, so Toronto played 17 more games against hot teams than against cold teams. Edmonton were par.
Toronto Maple Leafs +17
Vegas Golden Knights +13
Buffalo Sabres +7
Anaheim Ducks +7
Ottawa Senators +7
Los Angeles Kings +5
Washington Capitals +6
Calgary Flames +6
San Jose Sharks +4
Detroit Red Wings +3
Montreal Canadiens +3
Arizona Coyotes +2
Minnesota Wild +2
Colorado Avalanche +1
Florida Panthers +1
Edmonton Oilers 0
St. Louis Blues 0
Chicago Blackhawks -1
Nashville Predators -1
Vancouver Canucks -3
Dallas Stars -3
Carolina Hurricanes -4
Columbus Blue Jackets -5
Boston Bruins -5
Pittsburgh Penguins -5
Seattle Kraken -5
Winnipeg Jets -7
New Jersey Devils -8
Philadelphia Flyers -8
Tampa Bay Lightning -9
New York Rangers -11
New York Islanders -18
Thanks for looking into it.
And yeah that makes more sense. Unusual still in that it seems the Oilers faced more polarized (hot or cold) opponents than most other teams.
Looking at the top of this chart (the +/- one) and it’s apparent who the hot teams were. All the teams with the most games against hot teams were in either the Atlantic or Pacific divisions. Boston was hot all season and Edmonton went on a massive run in the 2nd half.
Danny, which AI are you using?
ChatGPT isn’t up to date and I’ve seen hilarious stuff when trying to use Bard… things like Joe Thornton showing up as a goalie or something.
I’m interested in learning your process here. I’ve been fumbling around trying to use AI for analysis myself, but I haven’t gotten very good results.
Im using the pro version of GPT4 which now has functions (beta). It allows you to import data sets and ask questions. It basically writes python code using pandas, and executes that code on the uploaded dataset.
Thanks!
So you just describe the calculation in text and it pulls the data and does it?
For example: “For each nhl team during the 2022-2023 season, calculate the game win percentage of a team’s opponent for the 4 games preceding and following a given game divided by the opponent’s win percentage for the year. Average this value across all games for a team. Do this for each team and stack rank the result outputting the computed value for each team.”
Pretty much yeah. The prompting can be a skill in itself, but in terms of data science, it’ll do whatever you want to know based on human language questions.
I wouldn’t really call that AI, per say. It’s using AI to understand human language and do some math calculations not computing something using AI.
An example of AI computation would be inputing a bunch of features into an AI model and asking it to do a regression (reason) about which features are more important for answering a question by learning from training data. I suppose you could argue that’s just some math calculations too :-).
Got to say it’s a cool idea to try this!
Not sure what you’re saying here. It’s an LLM… which is a generative AI.
To expand on how this is working, GPT4 (LLM) reads my request, imports the data, and writes an actual python script that can accomplish the request, then executes the script to calculate the outcomes. So the entirety of the heavy lifting here is 100% generative ai
Yes but it’s like saying I asked chat GPT to compute the sum of the points scored by the top 5 players for each team. AI was used but the calculation itself was rudimentary and not insightful. Any middle school kid could work that out. I mean to apply AI analysis to the hockey stats themselves to get insights that can’t be derived by hand.
Asking AI to compute easy calculations doesn’t mean it’s not AI. It’s simply doing what you asked it to do.
Agreed. Also you (Big Mac) said its using AI to understand language, so it sounds like you think its AI, but then you say its not computing something using AI.
Actually understanding human language is quite important for AI I would say. I also think that, by my rudimentary understanding, that cognitive science has been an important feild towards AI progressing, and part of that is understanding how humans think, which is not through computations and reasoning (we are appearently not so good at actual logic) but through conceptual metaphor. But im blathering now.
You’re use of AI has me thinking about my own data analysis. How can i learn more about this? Is there some way it can look at enormous swaths of data that i have stored (in excel) and analyze it for me?
if you want to experiment for free, I’d recommend downloading a free python IDE like PyCharm, and then using free ChatGPT membership, write prompts that ask it to create scripts that analyze your data.
The easiest way is to buy a GPT pro membership (around 25 bucks) and use the new Code Interpreter mode, which allows you to upload your CSVs and it can inspect your data instead of you needing to describe the contents. Then it creates the python scripts and runs them for you, presenting you with the results
Thx
Sure AI is involved in understanding your request and your question and analysis was interesting. Here’s an example of what I’m trying to describe. JFresh has a model that computes the relative value of a player. It has it’s own weighting of different attributes. Can you use AI to learn a better model for valuing players? So for your example, instead of it calculating the relative strength of schedule based on your formula, learns it’s own (should be better) formula for computing relative strength and applies it.
I think that’s machine learning, not AI.
I guess machine learning is a form of AI, but AI need not involve ‘learning’.
This is not at all my field though.
AI is a popular layman term for machine learning.
That may be, but I’m pretty sure they’re not synonymous (for non-laymen).
I suppose that’s a question for chatGPT!
Expanding my example a bit. For calculating strength of schedule, your model doesn’t take into account: date, day of the week, start time of the game, back to back, backup goalie, player injuries or slumps, coaching change, travel schedules, all-star break, current standings, weather, elevation, etc. An AI can take all those as inputs and determine a relative importance of each variable (feature) and build it’s own model to calculate strength of schedule. The more interesting inputs (features) and the larger the sample data-set the better the model.
Yes an AI could do all that. You are conflating the idea of more intricate analysis, with AI. A generative AI will do whatever you want. You need to ask it to consider those things, like I mentioned earlier you can get very intricate with the prompts.
You’re more decribing a Narrow AI / specific machine leaning scenario, which would require a massive dataset and iterate every conceivable path to calculate and test probability.
I guess the takeaway is generative AI isn’t sentient. It’s not going to ask its own questions and find the answers itself. It needs prompting. Agents are a layer that are closer to what you’re describing, that can use generative AI to ask itself questions to help generate more multi step solutions.
But that’s still prompting… automated prompting.
Regardless it all goes back to the point, you’re confusing the simplicity of my request as ‘not being ai’
I do what I can.
Lol. I was confused by this, but now I realize that Al and Ai are indecipherable when using a capitalized I 😂
Hahaha! I see what you did there.
Corcoran, Peters and De Jong are my ones to watch from that list. They could all do well (or well again in Peters case) with the Condors. De Jong in particular interests me as a potential long shot to make the NHL.
Interesting argument by Mathew Panchyshyn on Twitter: “Broberg is leading his draft class in ability to play.”
—
Mathew Panchyshyn
These are the D drafted in 2019 to play +25 gp in 2022/23…
-Broberg -Byram -Seider -York -Soderstrom -Xhekaj
I will go over the EV off/def, overall off/def, xGAR, and GAR. These already factors QoC/QoT. 0 is NHL level.
https://twitter.com/mathewjdp/status/1678506856287748096
Conclusion:
Overall: Broberg is leading the draft class in the ability to play. This does consider QoC and QoT. I would be very hesitant to label anything on Broberg other than “Pending” when trying to understand what he is. So far, he’s been nothing but excellent.
https://twitter.com/mathewjdp/status/1678508895705214977
—
Highly sheltered minutes, but encouraging nonetheless. Thanks for that!
Coaches choice. I am perplexed on how Oiler coaches use players including pre Woody
Bouchard also has team top stats in a few of them. Yes he makes mistakes. Bro by what that tweet shows is really good at the things that matter most to be a really strong NHL D. The one thing he had issues with was puck retrieval. So pair him with a guy good at it. Both Nurse (historically) and Ek are. I am pretty sure that is also an issue of not enough games, especially consecutive, so he can get used to it at the NHL level
To me players are what they are pre-draft. The players that don’t hit expectations have fatal flaws that were overlooked or missed (can’t skate at an NHL level, no hockey sense, weak puck skills, too small to handle it, are the usuals), assuming no big injury, attitude, or off ice issues
Bouchard is exactly what they drafted. He isn’t Hedman, but he is close to being a very good top 4 D that is a two way player, offensive leaning, as he was in junior. He was the go to all situations D on his junior team
Bro was an elite D for Sweden, captain, their most relied upon D. Able to play men’s league at a young age. He got hurt a few times which is the biggest worry and it set him back, but he’s that player still, he has all the attributes of a strong NHL player, he just needs to play
What perplexes me is that the risk of using the younger players is a hill too far. Because the guys they use instead aren’t as talented, make mistakes too, and really hamper pairs and team play because they have such a limited range. Ceci has repeatedly been shown to only have a stronger trait in being a decent in zone D, and he wasn’t even that last season
The real risk is not being able to make some astute and hard cap decisions and sign these two early and long. It will be a lot more expensive later which will reduce the quality of the team then
To me if Bouch will go 5-6 x 8 you do that all day long. Just don’t give him protections so you can change your mind later. Try it with Bro in a bit. If you can’t replace Foegele with as useful or near a player at 1M and find somehow a decent RD to add depth that is cheaper than Ceci, then you shouldn’t be an NHL GM IMO. These are supposed to be the easiest things to do, find bottom roster players
“There are three kinds of lies: lies, damned lies, and statistics.”
I’ve played around with a lot of Evolving Wild’s data.
The basic problem with all of this data (or Natural Stat trick) is that there are different minutes played by players on different teams.
It doesn’t matter if you’re using expected goals, corsi, GF%, Danger Fenwick, or whatever…
if you compare, for example a guy who’s playing sheltered (7d sheltered) 3rd pairing minutes (Broberg) to a guy who’s playing top pairing minutes on the Red Wings (Seider), your results and conclusions are going to be completely invalid.
This is where PuckIQ is unique and stands alone.
At the end of the day, Broberg is playing 11:22 per night (5v5) on the Oilers and Seider is playing 16:10 against the toughest comp on his team.
The moment you try to tell me your stats tell you Broberg (a 12:36 minute per game defensemen) is better than Seider (23 minutes per game all states), you’ve lost me.
It’s the age-old 3rd pairing d problem we’ve stumbled over for years.
Guys like Matt Benning having rock solid fancy stats leaving posters here scratching their heads trying to figure out why the Oilers coach is too stupid to play them more…
not realizing that the coach is working his ass off to hide them. using on the fly starts (or reduced minutes) to keep them away from good players or playing them against good players when the puck is headed in the direction.
The last point is definitely true. But one issue with this is the coaches are so damned risk averse sometimes that they’re either unwilling to give more mins or tougher matchups to the younger players or won’t live with their mistakes when they do. But they’re willing to live with repeated mistakes from veterans that should know better.
A good coach needs to not just shelter young players but to find them consistent opportunities to elevate their game even when they’re making mistakes. If this was done more early last season for Broberg (when healthy) or two years earlier for Bouchard the team would be in a better position to make the call on going long, bridging or trading the player.
Although it’s go time for the team, they are also good enough to win at a playoff pace early in the season even with a couple of rookies making more mistakes than you’d like. It may save you assets trading for players at the deadline as well. This year I really hope Woody takes this approach with Holloway until the deadline. It may pay off in spades in the playoffs.
Yes, sheltered vs unsheltered is the obvious objection.
Though Mathew does say the data already factors in Quality of Comp and Quality of Team. How reliably it does that is is hard to say.
Someone asks him: “How sheltered is each player respectively? Because something tells me Mo plays harder minutes than Broberg”
Answer: “Of all 6. Mo plays the most against the Elites, and Broberg plays the least. This takes that into account. Mo did not well this year against Elites (like most don’t) and was avg at best against other competition. Broberg was exceptional against competition other than Elites.”
https://twitter.com/mathewjdp/status/1678513768786198528
Broberg has seemed unusually sheltered this last season. Seemingly more than he should have been.
If I were to guess, a big reason is to give him extra time to learn to protect himself. His ability to sense danger when he first entered the NHL was poor.
I don’t have twitter at work here and you can no longer access it without logging in. He didn’t say where his data is from, but I am assuming it’s from Evolving Hockey.
JP and I went over a bunch of EH player ratings several months ago now. While they do claim to adjust for QOC and QOT, it’s not much. Their total ratings, way over value 3rd pairing offensive d, for example.
Even with PuckIQ, it’s really hard to compare a 3rd pairing d to a top pairing d.
One issue, Tyler Dellow exposed before he got hired was the 3rd pairing d problem.
Coaches shelter their 3rd pairing d, both using toi and on the fly starts.
Let’s say you have a Matt Benning and you’re a coach. You don’t want him out there against Jordan Kyrou, but you only have so much control. But when your team has control of the puck in their zone and they start heading to the other team’s end, Adam Larsson heads to the bench and you throw Benning over the boards. Now Jordan Kyrou is on the ice against Matt Benning, but the Oilers have Mcdavid on the ice and possession of the puck.
You get enough of these situations when Larsson off and Benning on and it starts to look in the data like Benning is some sort of Elite comp savant, even though it’s only because he’s shifted on the fly against elites when we have the puck, not because he’s actually making a contribution.
Make sense?
Thanks, Ryan. Yeah, that makes sense.
I just looked at PuckIQ, and that supports your point.
For example, Broberg comes out as much better than Nurse in DFF against Middle and Gritensity, and yet against Elites far worse.
Similarly compared to Ekholm, Broberg rates about the same DFF against Middle and Gritensity, and far worse against Elites.
Yet he’s surely not near their level in reality, as their superiority against elites proves.
You can’t disregard comparison against elites and only compare results between lower comp and think you have something there. That seems to be what Mathew has done.
With defensemen, the first thing I look at are toi/g and on the fly starts/60.
Both of these stats are determined by the coach. They play the players they think are the best the most and the best players have the lowest OTF starts/60.
Broberg has the highest OTF starts/60 on the Oilers (excluding Nemo).
That usually indicates that the coach is trying to control who they play against or when (on the fly when the puck is headed in the right direction).
Opposing teams also try to target 3rd pairing d like when Broberg got the phantom holding call against Eichel in game 5.
How players like Broberg show up against elites probably depends more on how successful other coaches are getting the Eichels out against them on the attack vs how good their own coaches are getting them out against the Eichels while on the fly rather than truly indicating their innate ability of how good they actually play vs elites.
In other words, how a sheltered player plays against elites probably isn’t indicative of their actual ability. If they were actually good playing against elites, they wouldn’t be sheltered.
I’d have to look it up (I don’t hve time tonight but remember Oscar Klefbom had a ZS push and a pile of OTF in his first full season. He turned out okay, and Broberg has been a fairly close match so far.
If we’re going by his 2014-15 season, Klefbom actually had the lowest OTF starts/60 on the team. He was also #1 in toi/g. (draft +4).
Edit-> in the 2013-14, the Oilers had 11 d who played over 100 minutes at 5v5. Klefbom was 6th in toi/g and 6th in otf starts/60. That was his draft + 3 season.
Last year was Broberg’s draft + 4 season.
and Klefbom got mascaraed.
He did, but the Oilers were the 28th team out of 30 that year.
If you look at #1 d on bad teams, they’re going to get torched against elites.
I think maybe Hampus Lindholm is a #1 who played on some bad teams holding his own against elite comp for a few years, but he fell back to earth before getting traded to Boston.
Yes but the point is he wasn’t playing those minutes and in those situation because he was ready in draft plus 3 but because of the team being awful and, consequently, its not really fair to use him playing those minutes as a comparison to Broberg’s deployment at the same stage, right?
It seems to me like you are making a strong argument that we can’t really analyse Dmen with any analytics. If their stats are dependent on their TOI and on the fly shifts and if TOI and OTF shifts are based on coaches’ subjective opinion of the player’s ability, then aren’t we just analysing the coaches’ opinion of the players?
I would say that defensemen are more of a challenge to analyse with analytics than forwards.
If you want a decent list of who the best forwards are in the league, this isn’t a terrible starting point (here).
I’m not saying you can’t use analytics, but you have to be aware of potential problems. The 3rd pairing OTF issues was identified by Tyler Dellow, not me.
It’s easier to compare say a Nurse to a Makar since you’re comparing two #1 d who each play on competitive teams.
If you are going to compare two players either forwards or defensemen, you have to look at their stats in the context of what type of minutes they play. Context is always important and QOT weighs more than QOC.
Coaches do a decent job handing out icetime except when they’re William Karlsson’s coach on the Columbus Blue Jackets.
If you don’t like TOI as a metric, you probably shouldn’t use PuckiQ data. 🙂
TOI > 75% percentile is one of their criteria.
I don’t think that should impact a hockey discussion.
Klefbom’s offensive zone starts in his year 4+ draft (7.87/60) and Broberg’s in his 4+ draft year (8.14/60) are not dissimilar.
One of the things we should take into account is the lack of a Nurse or an Ekholm on the 2014-15 Oilers. That was death valley and had much to do with Klefbom’s deployment (although I did feel Dallas Eakins was innovative in handling him).
NYR sign K’andre Miller for 2 x $3.87. Bouchard has more offensive upside, although they both have the same career high in points (43). Miller is better defensively.
By what metrics is Miller better defensively?
https://www.naturalstattrick.com/playercompare.php?fromseason=20222023&thruseason=20222023&stype=2&sit=5v5&score=all&stdoi=oi&rate=y&p1=8480817&p2=8480803&loc=B&gpfilt=none&fd=&td=&tgp=410&lines=single
A crude exercise, but Bouchard looks better by almost every metric.
This past season, Miller’s coaches trusted him to play 500+minutes against players classified as elite by Puckiq, 2nd most on the Rangers. Bouchard got about half the minutes (246) against the same category of players. Coaches deployment is one significant indicator and Miller’s minutes against elite have been high for a few seasons.
About 3:30 more total TOI per game too.
18 comments
Its a long summer when your team is actually good, roster looks good. Even longer till they get back to playoffs.
Early 70’s I recall hearing Sailin’ Shoes on CKUA, loved it. Drove straight to Sub Records, it wasn’t in stock so I bought Feats Dont Fail Me Now. Looking back, imo for me, they were easily the best band of the 70’s. Lowell George was truly a great song-writer & musician (from The Mothers of invention to Little Feat) who was lost far too early.
Let the fast feet in fast boots of our Oilers conquer all who challenge us. Dont fail us now!
Ah Patrick Russell couldn’t hit water if he fell out of a boat.
Couldn’t hit the broad side of a barn with a handful of rice.
Oh but the determination.
His Goal that was called back was robbery.
What analytical information do we think the much-lauded Eric Tulsky provided to advise for the signing of Brendan Lemieux?
I assume you are joking. But this kind of sentiment demonstrates such a myopic understanding of analytics and the value of analytics.
It’s also funny that folks always try to clown on one specific decision maker within an organization and claim to know they were the specific individual who advocated for a transaction. But go off!
No idea if OP is joking or not but to be fair it is pretty standard fare by a large number of posters on this blog to credit Tulsky with every astute move that franchise makes with just as much or little information as to who is driving the decisions.
Not sure how his comment demonstrates a myopic understanding of analytics though.
Tulsky runs the Carolina Pro Scouting and Analytics departments so would have been intimately involved in signing Lemieux.
https://theathletic.com/3171453/2022/03/08/how-do-the-hurricanes-blend-analytics-and-scouting-what-was-chicagos-gm-search-like-ask-eric-tulsky/?source=user_shared_article
This decision might be based to some degree on Keegan Kolesar destroying the opposition goaltender on the Knights march to the cup.
I have no doubt they have their reasons & that Tulsky was involved in the decision.
With respect, many long-time community members laud Eric Tulsky and the Canes’ organization use of analytics and speak to a massive gap between the Oilers and Canes management and front office abilities largely due to this.
I can’t remember a Cane transaction that wasn’t lauded for its reliance on analytics.
Just yesterday there was a piece at The Athletic that suggested the Oilers didn’t have an analytics team that could provide Jay Woodcroft with the Puck IQ data and Woody couldn’t adjust and it cost the team the series.
The initial comment seems prudent for this community.
As luck would have it…the Athletic has just summarized which teams have been the most successful in the salary cap era when the game changed drastically.
The Hurricanes have the 10th best record in the league.
https://theathletic.com/4593586/2023/06/20/nhl-franchise-rankings/
Maybe they needed some more snot on the team.
Maybe their analytics department has determined they need some snot to catch up with the Oilers? I mean, since Woodcroft and Manson have taken over the Oiler, they have a better points percentage than the Canes.
All that time since the game changed with the cap era but yet they are currently not even further ahead than the, from accounts, dark-age Oilers.
Carolina’s problem is they don’t have a McDavid or a Draisaitl. Edmonton’s problem is they don’t have a Tulsky.
One problem is impossible to overcome unless lady luck blesses you. The other is a guy you hire with cash money.
Oilers apparently don’t have the cash money.
Losing Svechnikov to injury last season was a massive blow going into the playoffs.
Interesting that the bookies have them as second favourites to win the cup with him back next season.
And you have to think at some point they will move some of that bounty on D for additional scoring,
Meh – its been 6 years since a the Cup Winner had “a McDavid” (Ovie) – its not needed to win.
The Knights just won by trading for their two most impactful players (Eichel and Stone), neither of which are near generational.
If the Canes wanted, they could have Tulsky and a near generational player – they could likely acquire Auston Matthews but it seems Tulsky doesn’t want that.
I find it very improbably that Woodcroft doesn’t receive analytics and that the Oilers didn’t have the ability to get him something like Puck IQ info in real time and it cost the Oilers the cup.
You’ve completely lost me on this, OP. I’ll move along.
No, it was a joke about Tre in TOR taking over for the Computer Whiz Dubas. I figured you’d get it, maybe my delivery was off.
Tomorrow’s another day.
They are thinking outside the box and incorporating a new stat, finger bites/60.
I imagine they have done the analysis to determine that Brendan, at 800k, may be a good investment assuming he bites off a few fingers every now and then. They must have ballparked the amount of punishment Brendan would receive for reoccuring offenses and probably determined he could get away with biting the fingers off of a few divisional teams players, and maybe even a few Stanley Cup contenders fingers as well.
Holland may want to consider it as well. 800k would be a small price to pay if for example, you knew the player could possibly bite off one of say.. Kopitars fingers, Stones fingers, MacKinnon’s fingers.. wouldn’t you agree?
your daddy says I’m no good
your mama says keep away
I gotta tell you truthful girl, you can never make me stay…
Here is hoping that Tyler Benson is able to re-set in a new org.
Kudos to the young man for changing his game to try and make it at the NHL level. He wasn’t going to make it as a “producing winger”, at least on this Oilers team, so he increased his “jam level”. He always had some jam but he started to focus on that area and being “a pest”. It had some value at the NHL level but not enough.
For me, he never re-found his true game at the AHL level last season after re-assignment. He was not close to the dynamic offensive player at that level that we’d seen over the years.
I hope he gets back to being a dish-master and high end offensive producer at the AHL level and earns himself a termination of his AHL deal and another NHL contract.
Agreed.
I always wonder what could have been if he was healthier during his prime development years.
Best of luck, Tyler.
crazy to hear the news regarding New York times sports dissolving and them using the Athletic.
Things are really changing in that space.
The NYT is a formidable institution, but I’m not convinced their sports page was their leading component. Giving way to The Athletic is a recognition of comparative advantage. If it gets LT more eyeballs I’m all for it.
Jay Woodcroft should coach the Oilers, next year and beyond.
https://theathletic.com/4677219/2023/07/10/oilers-coach-jay-woodcroft/
As always, a great article to read, however, I don’t understand how a tandem pretty much alternates every several games then decides to stick with one goalie. I got the thinking early in the LA series. But I think the equation should have changed after the Campbell showed he found his footing and Skinner wasn’t setting the world on fire. I think the smart approach would be to put trust in the tandem and not just one goalie. Both seem to be better when given a night off to reset, so the best way to help Skinner would’ve been to give him a night off.
Also, what are your thoughts on ability to implement systems? This team is just a better team in general so of course it has a good win record, but I think the best system they have played was under the early games of Tippett. Everyone in the zone came back deeper and there seemed to be layers to the defence. The defence has previously been too shallow (not to mention the forwards) to play tight zone defence, but is that still the case?
I think if this team could play a style closer to vegas then they torch the league. Is your thought that Woodcroft has evaluated the alternatives and what they play is somehow the best approach? I’m not saying he couldn’t coach it but some coaches don’t, so is that a limitation of Woodcroft and might hold this team back? (Not implying it is, just an open question)
Woody cost us the series with his reluctance to start Campbell. If Maurice stays with Lyon they get beat by the Bruins. Sometimes a Coach needs to make hard decisions
On the flip side, Campbell had lots of minutes in the regular season to establish himself as a viable alternative. And he failed to do so, over and over again.
Maybe it would have gone differently if Bjugstad was not matched against Eichel, or if Broberg had played instead of Vinny.
I agree. Crazy Coach also commented on it. He calls it ‘box plus one’. Make Connor the one, it’s easy to use and would play to Nurse’s strengths, and Ekholm’s. Easier for Bouch
Connor as a threat to get a puck and take off. They would be very hard to beat and the goalies would play far better with consistency and no loose opponents in front of them
The other side though is that the Oilers have been far better under Woodcroft than any other coach in recent memory. The personnel has also been strengthened, and I believe in the past you’ve attributed much or all of the team improvement to McDavid and Draisaitl maturing.
There’s some truth to those things, but it’s not cut and dried that the system they’re using is ‘broken’, or that they would necessarily perform better using the box plus one or whatever else. It’s very likely IMO that that the system in place has contributed to the #2 regular season record and the 3 playoff round wins since Woodcroft and Manson arrived.
And that’s not to say the system shouldn’t be tweaked, or added to, or even scrapped entirely. Just that it has actually served the team pretty well over the last year and a half and we simply don’t know if something different would be better.