Randall Flagg Posted August 28, 2018 Report Share Posted August 28, 2018 They're talking specifically about the stat that measures quality of competition by opposing skaters' relative Corsi or something, I assume? That wouldn't be surprising. i assume they aren't making a claim that staring Casey against every top line versus every 4th line would have minimal difference. Quote Link to comment Share on other sites More sharing options...
TrueBlueGED Posted August 29, 2018 Report Share Posted August 29, 2018 (edited) On 8/28/2018 at 2:54 PM, Randall Flagg said: They're talking specifically about the stat that measures quality of competition by opposing skaters' relative Corsi or something, I assume? That wouldn't be surprising. i assume they aren't making a claim that staring Casey against every top line versus every 4th line would have minimal difference. Mostly yea, but there's more nuance to why. QoC matters while the players are on the ice, but even the most inept coach doesn't have his 4th line against Crosby regularly, so in the aggregate you don't see much effect. It matters, but nowhere near to the degree that conventional wisdom would say. I think a lot of the disconnect is because most would look at it in the micro, shift-to-shift sense, whereas McCurdy et al. are talking about season-long effects. So there's some talking past each other. Hockey is fluid enough that everyone sees time against everyone, so the variance in QoC is pretty tight, and thus it's pretty hard to measure statistical impacts. I think McCurdy overstates things as a whole on this because he doesn't do enough to acknowledge the measurement difficulties of QoC, but I think his point is pretty sound when walked back a notch. What I don't think has been properly assessed are potential cumulative effects of QoC. For example, does Risto look worse against lesser players because he's worn down from the first half of his shift against Crosby? I would think yes, but I don't think this type of question has really been addressed, and doing so with the available data would be super challenging. Edited August 29, 2018 by TrueBlueGED Quote Link to comment Share on other sites More sharing options...
Stoner Posted August 30, 2018 Report Share Posted August 30, 2018 (edited) "Advanced stats" can't be "explained." They are figments of people's imaginations. Like the home team hit count of the stat man on Long Island. I mean, y'all can have your fun, but don't act like the uneducated, unwashed masses of people like me have to "edumucated" on it. Have a "nice" day. Edited August 30, 2018 by PASabreFan nano robot infection Quote Link to comment Share on other sites More sharing options...
LGR4GM Posted August 30, 2018 Author Report Share Posted August 30, 2018 19 hours ago, TrueBlueGED said: Mostly yea, but there's more nuance to why. QoC matters while the players are on the ice, but even the most inept coach doesn't have his 4th line against Crosby regularly, so in the aggregate you don't see much effect. It matters, but nowhere near to the degree that conventional wisdom would say. I think a lot of the disconnect is because most would look at it in the micro, shift-to-shift sense, whereas McCurdy et al. are talking about season-long effects. So there's some talking past each other. Hockey is fluid enough that everyone sees time against everyone, so the variance in QoC is pretty tight, and thus it's pretty hard to measure statistical impacts. I think McCurdy overstates things as a whole on this because he doesn't do enough to acknowledge the measurement difficulties of QoC, but I think his point is pretty sound when walked back a notch. What I don't think has been properly assessed are potential cumulative effects of QoC. For example, does Risto look worse against lesser players because he's worn down from the first half of his shift against Crosby? I would think yes, but I don't think this type of question has really been addressed, and doing so with the available data would be super challenging. We would need a new metric to measure it but I think it does have some impact. The key would be making sure the cummunalitive impact of just playing too much is accounted for. Is Risto bad stats wise because he plays 30minutes a night on a terrible team? Is he bad because he himself isn't good? Does he only face the top 2 lines and after 25-30minutes of that he's tired? Considering forwards play in the low 20's or high teens and Risto is in the high 20's, I wonder how much fatigue could be a factor regardless of opponent. 1 Quote Link to comment Share on other sites More sharing options...
pi2000 Posted August 30, 2018 Report Share Posted August 30, 2018 4 hours ago, LGR4GM said: We would need a new metric to measure it but I think it does have some impact. The key would be making sure the cummunalitive impact of just playing too much is accounted for. Is Risto bad stats wise because he plays 30minutes a night on a terrible team? Is he bad because he himself isn't good? Does he only face the top 2 lines and after 25-30minutes of that he's tired? Considering forwards play in the low 20's or high teens and Risto is in the high 20's, I wonder how much fatigue could be a factor regardless of opponent. He's just not very good defensively, it's that simple. Quote Link to comment Share on other sites More sharing options...
TrueBlueGED Posted August 30, 2018 Report Share Posted August 30, 2018 5 hours ago, PASabreFan said: "Advanced stats" can't be "explained." They are figments of people's imaginations. Like the home team hit count of the stat man on Long Island. I mean, y'all can have your fun, but don't act like the uneducated, unwashed masses of people like me have to "edumucated" on it. Have a "nice" day. Nice to see you engaging in your usual effective manner. Quote Link to comment Share on other sites More sharing options...
LGR4GM Posted October 29, 2018 Author Report Share Posted October 29, 2018 Does anyone know if this would be a good book on the subject? "Hockey Analytics: A Game-Changing Perspective" - Shea and Baker Quote Link to comment Share on other sites More sharing options...
IKnowPhysics Posted October 29, 2018 Report Share Posted October 29, 2018 I haven't read that one yet. The authors wrote books on basketball analytics, this is their first hockey book. Someone did a terribly written review of that book, more from the perspective of whether analytics has any value or not: https://www.wingingitinmotown.com/2018/4/24/17272436/book-review-hockey-analytics-a-game-changing-perspective I've read Rob Vollman's stuff, and I like it: http://www.hockeyabstract.com/fan-s-guide-to-hockey-analytics Quote Link to comment Share on other sites More sharing options...
pi2000 Posted October 29, 2018 Report Share Posted October 29, 2018 (edited) If Risto maintains his current pace... he'll finish the season a -60. I ran some TRpm's. BUF has 21 goals for at even strength or worse (17 5v5, 3 5v6, 1 6v5), and 25 goals against at even strength or better (20 5v5, 1 5v4, 1 3v3, 2 6v5, 1 6v4). That's a team +/- of -4. Player TRpm's (actual vs expected +/- based on minutes played). Top 3: Skinner +11 Pommer +6 Beaulieu +3 Bottom3: Risto -7 Sheary -5 Okposo -5 Reinhart -4 Mitts -4 Dahlin +1 Eichel 0 Sure 11 games is a relatively small sample size, but there's already a big gap between the have's and have not's. Keep in mind, a TRpm of 0 on a good team means you're an equal contributor (when compared to teammates) to the team's success. A TRpm of 0 on a bad team, means you're part of the problem. To have a negative TRpm approaching double digits at this point of the season is alarming. And if it wasn't obvious, without Skinner, this team probably close to where it was last season. Edited October 30, 2018 by pi2000 Quote Link to comment Share on other sites More sharing options...
pi2000 Posted November 1, 2018 Report Share Posted November 1, 2018 Interesting article for those with The Athletic subscriptions... https://theathletic.com/626729/2018/11/01/qa-steve-valiquettes-radical-approach-to-evaluating-shots-is-catching-on-in-the-nhl/ ...love the idea of sanitizing the shots data. That's been my biggest gripe with Corsi, all shots are not equal. Quote Link to comment Share on other sites More sharing options...
sabills Posted November 1, 2018 Report Share Posted November 1, 2018 33 minutes ago, pi2000 said: Interesting article for those with The Athletic subscriptions... https://theathletic.com/626729/2018/11/01/qa-steve-valiquettes-radical-approach-to-evaluating-shots-is-catching-on-in-the-nhl/ ...love the idea of sanitizing the shots data. That's been my biggest gripe with Corsi, all shots are not equal. Thats the good stuff. Love seeing that guys whole process of how it came together, super interesting. Quote Link to comment Share on other sites More sharing options...
Randall Flagg Posted November 1, 2018 Report Share Posted November 1, 2018 50 minutes ago, pi2000 said: Interesting article for those with The Athletic subscriptions... https://theathletic.com/626729/2018/11/01/qa-steve-valiquettes-radical-approach-to-evaluating-shots-is-catching-on-in-the-nhl/ ...love the idea of sanitizing the shots data. That's been my biggest gripe with Corsi, all shots are not equal. Which is why it's been known for years that expected goals, which are a first attempt at weighting shots based on quality, are the best offensive predictive stats out there Catch: didn't read the article Quote Link to comment Share on other sites More sharing options...
pi2000 Posted November 2, 2018 Report Share Posted November 2, 2018 9 hours ago, Randall Flagg said: Which is why it's been known for years that expected goals, which are a first attempt at weighting shots based on quality, are the best offensive predictive stats out there Catch: didn't read the article Not sure how expected goals are calculated, but this guy's shot evaluation is pretty unique and interesting.... and he has his own unique algorithm for calculating expected GF. Quote Link to comment Share on other sites More sharing options...
That Aud Smell Posted September 23, 2019 Report Share Posted September 23, 2019 The chart itself says "don't pay attention to these results." So, with a big dose of FWIW, the Sabres have been verging between bad and fun during the preseason. Quote Link to comment Share on other sites More sharing options...
Randall Flagg Posted September 23, 2019 Report Share Posted September 23, 2019 Lightning bottom 3 finish incoming And yeah you can throw these results away because their sample sizes are so small that one or two games could flip it completely, and none of these teams have used lineups that they'll ever see again in the regular season. Also, NST, the site that the data gets scraped from, has been rife with errors this preseason - for example, in a Ducks game they had Getzlaf as playing the first six minutes of the 3rd period with no breaks. I believe they also had Eichel with no offensive zone starts this preseason. This is the first time the site has included preseason stats so it's really buggy Quote Link to comment Share on other sites More sharing options...
LGR4GM Posted September 23, 2019 Author Report Share Posted September 23, 2019 That's what happens when you have bad goaltending. That's why the Sabres are down there. Quote Link to comment Share on other sites More sharing options...
Randall Flagg Posted September 23, 2019 Report Share Posted September 23, 2019 17 minutes ago, LGR4GM said: That's what happens when you have bad goaltending. That's why the Sabres are down there. Nope, expected goals against doesn't track goals against, it only tracks shots against and their locations and types 1 1 Quote Link to comment Share on other sites More sharing options...
LGR4GM Posted September 23, 2019 Author Report Share Posted September 23, 2019 21 minutes ago, Randall Flagg said: Nope, expected goals against doesn't track goals against, it only tracks shots against and their locations and types interesting. Quote Link to comment Share on other sites More sharing options...
dudacek Posted September 23, 2019 Report Share Posted September 23, 2019 I think it’s basically saying the Sabres are the worst in the league at preventing chances. But they have been so good at creating them that it almost evens out. The second sentence is all you need to know about the validity of the data. Quote Link to comment Share on other sites More sharing options...
Curt Posted September 23, 2019 Report Share Posted September 23, 2019 It says that the Sabres’ offense has been average and defense has been worst in the league, but it’s only for the 4 preseason games, some of which had more AHL players than NHL players, so the chart is meaningless. It honestly makes more sense if you just read the numbers and ignore the “good, bad, fun, dull” part. Quote Link to comment Share on other sites More sharing options...
That Aud Smell Posted September 23, 2019 Report Share Posted September 23, 2019 28 minutes ago, Curt said: It says that the Sabres’ offense has been average and defense has been worst in the league, but it’s only for the 4 preseason games, some of which had more AHL players than NHL players, so the chart is meaningless. It honestly makes more sense if you just read the numbers and ignore the “good, bad, fun, dull” part. The chart says so itself. Mostly just a funny thought exercise. It's funny - my experience is precisely the opposite. 1 Quote Link to comment Share on other sites More sharing options...
That Aud Smell Posted October 10, 2019 Report Share Posted October 10, 2019 Very interesting. Quote Link to comment Share on other sites More sharing options...
triumph_communes Posted October 10, 2019 Report Share Posted October 10, 2019 What’s it going to cost to query that dataset? Quote Link to comment Share on other sites More sharing options...
rakish Posted October 10, 2019 Report Share Posted October 10, 2019 1 hour ago, triumph_communes said: What’s it going to cost to query that dataset? Do you think the interaction is going to be sql? Quote Link to comment Share on other sites More sharing options...
triumph_communes Posted October 11, 2019 Report Share Posted October 11, 2019 1 hour ago, rakish said: Do you think the interaction is going to be sql? I don’t know. I’m just curious what access will be. Simple stats was public but this is different Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.