Advanced Stats

[InjuredChoker]

If the over/under for the Avs next season was set at 99, how many would bet over?

depends on the roster.

 

[Pokecheque]

Oh, and please quit with this "moral victory" garbage. Call the advanced stats crowd smug if you must (certainly some are) but none have ever used that term, nor have they ever implied it.

Fancy stats can't predict games, but they can somewhat accurately predict trends. They were spot-on regarding the Leafs, for instance. Chances are Toronto would've missed the postseason last year as well had it been 82 games.

 

[tigervixxxen]

Hench's point about winning the CF against good teams is valid though. What's the argument against winning CF against possession giants Anaheim and Boston? Fluke? Luck? I think it's an example of one of the holes in the measure.

If people want to call a performance unsustainable that's fine but I want a better argument about why it's happening other than just luck.

I wouldn't have a single problem with a 99 point season next year.

Oh and

Next year, who is more likely to fall down the ladder? Rask or Varly? Given how poor Colorado is with puck possession, it's going to be Varly, which means a bunch of those 1 goal wins will flip the other way, and everyone in hockey is going to say "I told you so".

 

[badtakemachine]

I've got a few problems with the so-called 'advanced stats'. The first is the use of the word 'advanced'. Seriously? What makes them advanced? Try 're-arranged' stats. I see no advanced statistics here, all I see is the use of the regular old statistics compiled in such a way to try to prove something. 'CORSI' is a direct function of shots directed at the net. 'FENWICK' is a function of shots directed at the net and shots blocked, because shot blocking is a 'skill'. Okay, so if I make up the Aerolanche number that doesn't count shots that miss the net because the goalie forcing shots wide due to his positioning is also a skill, will I get my username on NHL.com? Oh wait, that stat already exists, it's called shots on goal. 'QUALCOM' is a direct function of our old friend +/- with no other parameters, which by itself, if you ask any fan of Taylor Hall and Nail Yakupov, supposedly means nothing in the first place. Adding regularly kept statistics together does not make them advanced.


Let's see someone sit there every game and count every second the team is in possession of the puck. At least that would be a 'new' statistic. Sure, the 'CORSI' and 'FENWICK' numbers may have some relation to what these actual numbers would be, but I'm going to go out on a limb and say that goals for and goals against would have a just as good, if not better, relation to the real data for time of possession over a statistically significant sample. Even then, is this the correct way to predict a game? Is the team that wins the time of possession battle more likely to win? Maybe, but not if you have Jacob Markstrom in net and you're shooting on Semyon Varlamov. Garth Snow would be offering Johnny Mitchell $6M per season after leading the league in time of possession. Just because there is a correlation doesn't mean there isn't a better way.


I'm all for trying to find patterns to further understand the game. But to put any more stock in these so-called 'advanced statistics' while disregarding 'regular statistics' (or as I like to call them, the 'actual statistics') and to use the argument "well you just don't understand" is laughable. Hey, if you want to use these numbers for betting on the game, be my guest. Good luck with that. Hockey is an incredibly complex game with a lot of moving parts. It seems that we are nowhere near having a way to account for four officials, ten skaters, two goaltenders, each with their own size, strength, equipment, age, experience, playing with injuries, coaching system, line change frequency, momentum from the last shift, momentum from the last game, and position in the standings, all sliding around on 17,000 square feet of ice. To simply make the assumption that some or all of these simply don't matter, to my satisfaction, has never been proven. Right now, the only stat that I can see that actually predicts the strength of a team is wins and losses. Even that can be sketchy if you consider that San Jose and Chicago are tied, yet San Jose is 49-31 while Chicago is 46-34.

 

[henchman21]

There's literally thousands of games of data that strongly support fenwick/corsi, including the current Avs season. You're selectively choosing 10 games to prove otherwise.

You don't see a slight problem with that?

When saying with 80 Avs games this year, that you take the 30 'worst' games and they win 26... you don't see a problem with that? 30 out of 80 is a pretty good sample size to see how corsi has applied to the Avs.

I think there are 2460 games a year. To get a good sample it would take ~100 samples (off the top of my head, probably a tad high... 93 is likely closer) to have a decent amount of confidence. Really they should be randomly picked, but skewing it to the games where corsi should mean the most (worst corsi) you could pick the bottom 100 games and see where it stands.

Bottom 50 corsi games in all the NHL teams went 38-12 against it. It could even out in the next 50, I don't feel like counting more than that though. Random would be more accurate, but that shows that the stat isn't the end all be all. If a team is supposed to lose way more often than not when getting 'beat' that badly in possession, why are they winning so much? There should be a stat that can answer that question and not luck or great goaltending.

 

[Avs_19]

The more confounding team to me when looking at advanced stats is the Rangers. They're good defensively, sport one of the best goalies in the game, have good possession #s, plenty of offensive depth, a mobile defense capable of jumping into the play...and yet they simply can't put the damn puck in the net. They also have, IMO, one of the best coaches in the league behind the bench.

The advanced stats guys say it's bad luck, but when your "bad luck" spans years, I don't think you can just chalk it up to that. It's possible it's a question of talent--the Rangers are good, but no real elite game-breakers IMO--or a question of a lack of muckers who can score the dirty goals.

I'm not surprised. Whenever a team is winning but doing poorly on paper according to the advanced stats, they're just lucky and it'll catch up to them. If the advanced stats say a team should be doing a lot better than they actually are, then it's just "bad luck".

There have been a few games this season where, if you're actually watching the game, it looks like the Avs couldn't do anything and deserved to lose. However, they win the corsi battle in that game and earn a moral victory. I'm not knocking anyone who is into advanced stats, to each their own, but there are plenty of cases where that **** is stupid. There have been a lot of games where the Avs keep everything to the outside and don't give up many scoring chances but that makes them look bad according to the stats because of possession and the number of shots they give up. Then there are games where the Avs play like they did under Sacco but don't get many scoring chances and the stats show they played well, even though they lost and looked poorly doing it.

It's pretty obvious for people watching that Varly has bailed the Avs out a lot this season. They need to improve the defense in order to sustain this success and become a true contender. I don't think anyone is denying that but it's not all just "luck".

 

[tigervixxxen]

When saying with 80 Avs games this year, that you take the 30 'worst' games and they win 26... you don't see a problem with that? 30 out of 80 is a pretty good sample size to see how corsi has applied to the Avs.

I think there are 2460 games a year. To get a good sample it would take ~100 samples (off the top of my head, probably a tad high... 93 is likely closer) to have a decent amount of confidence. Really they should be randomly picked, but skewing it to the games where corsi should mean the most (worst corsi) you could pick the bottom 100 games and see where it stands.

Bottom 50 corsi games in all the NHL teams went 38-12 against it. It could even out in the next 50, I don't feel like counting more than that though. Random would be more accurate, but that shows that the stat isn't the end all be all. If a team is supposed to lose way more often than not when getting 'beat' that badly in possession, why are they winning so much? There should be a stat that can answer that question and not luck or great goaltending.

Excellent point. This problem isn't even contained to the Avs who are having a magical season. It's happening league wide.

And as I've said before the Avs are not trying to play a possession game so obviously their possession numbers are going to be poor. They are getting dinged for failing at something they are not even trying at. Yes I know every team wants the puck and to limit shots but not every team plays the same system. These stats seem skewed to a specific style of play that's currently sccessful. We know Roy does things differently, the hybrid man defense is different. The type of offense he's executing is different. It's at Leary something to consider. We all know they need to get better defensively and to improve possession but I guarantee they will never be a top 10 possession team.

 

[BB1133]

When saying with 80 Avs games this year, that you take the 30 'worst' games and they win 26... you don't see a problem with that? 30 out of 80 is a pretty good sample size to see how corsi has applied to the Avs.

I think there are 2460 games a year. To get a good sample it would take ~100 samples (off the top of my head, probably a tad high... 93 is likely closer) to have a decent amount of confidence. Really they should be randomly picked, but skewing it to the games where corsi should mean the most (worst corsi) you could pick the bottom 100 games and see where it stands.

Bottom 50 corsi games in all the NHL teams went 38-12 against it. It could even out in the next 50, I don't feel like counting more than that though.

I get what you're saying, but I just disagree with your conclusion. The stat guys aren't confused "why" the Avs have done so well this year.

 

[expatriatedtexan]

This whole thread reminds me somewhat of the psychohistory stuff from the Foundation novels written by Asimov in the 40s and 50s.

But not quite as entertaining.

 

[BB1133]

This whole thread reminds me somewhat of the psychohistory stuff from the Foundation novels written by Asimov in the 40s and 50s.

But not quite as entertaining.

It's nothing compared to what Leafs fans went through the past two seasons. Wait until September/October and everyone predicts doom for the Avs. The debates are just getting started...

 

[Pokecheque]

Excellent point. This problem isn't even contained to the Avs who are having a magical season. It's happening league wide.

And as I've said before the Avs are not trying to play a possession game so obviously their possession numbers are going to be poor. They are getting dinged for failing at something they are not even trying at. Yes I know every team wants the puck and to limit shots but not every team plays the same system. These stats seem skewed to a specific style of play that's currently sccessful. We know Roy does things differently, the hybrid man defense is different. The type of offense he's executing is different. It's at Leary something to consider. We all know they need to get better defensively and to improve possession but I guarantee they will never be a top 10 possession team.

Actually, they ARE trying to be a puck possession team. They're just not able to implement that properly because they have a largely immobile, underskilled blueline corps. If Roy was trying to avoid a puck possession game he wouldn't be letting them skate it through the neutral zone like he does. It'd be more like Sacco's dump-and-chase approach.

 

[tigervixxxen]

Actually, they ARE trying to be a puck possession team. They're just not able to implement that properly because they have a largely immobile, underskilled blueline corps. If Roy was trying to avoid a puck possession game he wouldn't be letting them skate it through the neutral zone like he does. It'd be more like Sacco's dump-and-chase approach.

I see more of a defensive first quick transition offensive team, not a cycle team which Corsi loves. They don't have the size to be a cycle team so they build around the speedy skilled forwards and focus on passing to the slot. It's more quality over quantity with the shots they take, even Babcock said this. Where they need to improve is some of the shots against totals, I think we can all agree there but the shot for totals are not going to dramatically move up is my point.

 

[Pokecheque]

I see more of a defensive first quick transition offensive team, not a cycle team which Corsi loves. They don't have the size to be a cycle team so they build around the speedy skilled forwards and focus on passing to the slot. It's more quality over quantity with the shots they take, even Babcock said this. Where they need to improve is some of the shots against totals, I think we can all agree there but the shot for totals are not going to dramatically move up is my point.

There are a lot of different ways to be a good possession team. In fact, you can actually be a dump-and-chase club or a shot-blocking team and still have good possession numbers.

This is one of the best articles by one of the best fancy stats guys out there on how the Kings, the best possession team in hockey, do what they do.

http://www.sbnation.com/nhl/2014/1/9/5278996/los-angeles-kings-stats-shot-differential-really-good

Analytics in hockey tell us the key is neutral zone play and a strong forecheck. If you stay aggressive in the neutral zone and give them hell in the attacking zone you control the pace of the game. That's why when I see Avs defenders keep backing up and the forwards letting opposing attackers enter the zone with speed I want to tear my hair out. When they stand guys up at the blue line consistently that's when they get better results.

In that St. Louis victory you saw a hint of what this team can do. They went after the Blues relentlessly. At least one of the goals was the result of a neutral zone turnover by the Blues. The Avs actually had a 52% Corsi rating in that game when you factor in scoring effects.

Roy has said repeatedly he wants to keep the gap between forwards and defense to keep from getting too wide, and he wants his team to stay aggressive in the neutral zone. But I imagine his efforts to do that have to be balanced with the fact that he's got guys like Nick Holden who can't play that type of defensive style.

Bring in more speed and skill to that blueline so Hejda and EJ aren't so grossly overworked and I think the possession woes of this club will change in a hurry. That's one thing that annoys me about the advanced stats crowd--some act as if these problems can't be fixed or that they're fundamental. I don't think that's the case.

 

[InjuredChoker]

I see more of a defensive first quick transition offensive team, not a cycle team which Corsi loves. They don't have the size to be a cycle team so they build around the speedy skilled forwards and focus on passing to the slot. It's more quality over quantity with the shots they take, even Babcock said this. Where they need to improve is some of the shots against totals, I think we can all agree there but the shot for totals are not going to dramatically move up is my point.

hawks are second best team and have been one of the best possession teams and they aren't cycle team either. quick transition, good puck movement from puck end (best in NHL imo), relentless puck pursuit.. i think that is the team avs should thrive to be. there are other teams who are top 10 in corsi who aren't heavy, cycling teams just speedy skilled teams.

 

[InjuredChoker]

Bring in more speed and skill to that blueline so Hejda and EJ aren't so grossly overworked and I think the possession woes of this club will change in a hurry. That's one thing that annoys me about the advanced stats crowd--some act as if these problems can't be fixed or that they're fundamental. I don't think that's the case.

i agree. stat crowd predicted that jackets miss the playoffs this season as they were all bobrovsky last season. and early on it looked like that way and they were meh possession team. then they started to turn it around since last december or january.

or vancouver before their cup run.

also that article is one of the best examples on how to use stats.. look them to see what's wrong.. and then fix it. in this case, look what kings do in neutral zone so well and learn to play against it. like hawks have seemingly done.

 

[PepsiCenterMagic]

http://hockeyanalysis.com/2014/03/12/lets-talk-corsi/

Great article explaining the limitations and use of Corsi. It stresses the need to somehow incorporate "shot quality" in analytics, something that Colorado is very good at doing. In their relatively low chances, I believe they have created great chances.

It also shows evidence that shooting percentage is not subject to regress if above average, dispelling the notion that PDO is automatically going to lower right away.

 

[InjuredChoker]

http://hockeyanalysis.com/2014/03/12/lets-talk-corsi/

Great article explaining the limitations and use of Corsi. It stresses the need to somehow incorporate "shot quality" in analytics, something that Colorado is very good at doing. In their relatively low chances, I believe they have created great chances.

It also shows evidence that shooting percentage is not subject to regress if above average, dispelling the notion that PDO is automatically going to lower right away.

sh% does regress if it's a lot above average but it will likely stay above average.

 

[expatriatedtexan]

From grantland.com Your Guide to the Stanley Cup Playoffs bit on the AVs/Wild matchup...

Line no. 2: Professor Fancystats says …
It’s hard to overstate how much the analytics crowd hates the Avalanche, a terrible possession team that’s spent the whole year riding unsustainable percentages and would have been eaten alive by the Blackhawks if the Blues hadn’t collapsed.*​

The * is a footnote: The Avalanche are well aware of all this, by the way. They don’t particularly care.

This kinda cracked me up.

 

[InjuredChoker]

DGB is awesome.

 

[tigervixxxen]

I didn't really want to bump this thread but I didn't really want to make a new one. Wish we had a statistic thread or something.

Anyway, found this link interesting. It shows where all the shots came from this season. Interesting to compare it to other teams or last season.

http://legion.stat.cmu.edu:3838/plotsberry-teams/

 

[BB1133]

Interesting article about how the Hawks (and Blues) are utilizing the "fancy" stats for their evaluations.

http://www.suntimes.com/sports/hock...-secret-formula-for-success.html#.U2PC_FfRYX1

Here was the mention of the Avs...

To the advanced stats crowd, the Colorado Avalanche are the exception that proves the rule. All season long, they defied the math by having an abnormally high shooting percentage and an abnormally high save percentage. The stat known as PDO adds these two numbers together, and the average is about 100. So if a team has a very high or very low PDO, it should eventually regress to the mean of 100 over the course of the season. Sure enough, the first-place Avalanche (with a 102.2 PDO in the regular season that offset a subpar Corsi percentage) lost in the first round to the fourth-place Minnesota Wild after getting outshot by an average of 33-25 each game.

Like a gambler on a hot streak at the blackjack table, the Avalanche defied the math for months. But in the long run, the math almost always wins.

 

[Hennessy]

People saying that a game 7 loss vindicates their stat predictions are jumping the gun. 82 games and a division title is luck, but a 7 game series, well, that's the pudding hiding the proof right there.

I do not expect the Avs to repeat as division champs next season. That said, neither do I believe they just got lucky all year long and are doomed for misery next year. Maybe they'll keep being lucky. Maybe the team adds some essential pieces. Maybe the guys mature and are that much better for it.

If the stats say you suck while winning the toughest division in the league, the stats need some fine-tuning.

 

[PepsiCenterMagic]

Interesting article about how the Hawks (and Blues) are utilizing the "fancy" stats for their evaluations.

http://www.suntimes.com/sports/hock...-secret-formula-for-success.html#.U2PC_FfRYX1

Here was the mention of the Avs...

I'll never completely agree with analysis like this.

PDO could mean MANY THINGS. Common sense tells me a high S% and SV% could come from skilled players. The combination of both could come from CONFIDENCE as well as luck. And the assertion that every team is the same is noted when comments like "everyone will regress/approach 1000 at some point", and is wrongly used as an absolute, when it shouldn't logically be used.

As for the comment, "the math almost always wins", I have a couple problems with that.

1. It's not the math that is right, the math was created to quantify certain theories about success deriving from teams who have the puck more. DERP. Don't pretend, math appliers, that you figured out hockey.

2. Though it is true that the more you have the puck, the better you are probably going to be, there still can be some anomalies, even though Colorado wasn't that TO AN EXTENT, but they were in a certain SAMPLE SIZE. Sample size is extremely important when analyzing data.

3. I don't get the fascination with the search for the "all-knowing" stat. As far as I know, hockey is fun to watch because YOU DON'T KNOW WHO WILL WIN. Once we get to the point where the "math" can tell us who is going to win before the game happens, then what will be the fun of watching hockey anymore?

 

[Foppa2118]

All these stats and math that prove why the Avs lost are completely pointless. People pick and choose what they want to use them for. They try to use them in the regular season for the Avs, and they don't work. Then you try to use them in the playoffs and all of a sudden there's the proof.

You don't lose in the playoffs because of a stat, you lose because you made mistakes, and also didn't react the right way in "key" moments. There's no stat for those things, and many of them are mental, not what you're doing on the ice that leads to it.

The mistakes being the defense failing to get the puck out, and losing all the battles down low because the Wild studied them during a long series and knew their weaknesses. The key moments being their special teams failing, and being too passive after scoring goals, and letting the Wild respond with goals of their own rather than having the killer instinct to keep pounding on them.

There's a multitude of things that lead to a win or a loss in hockey, and trying to boil that down to a stat is ridiculous. They're fun to look at for some people, but they have very little importance IMO unless you're using them with the sample of a whole career.

What most of us said last summer would be the Avs problem in the playoffs, was the Avs problem in the playoffs. The defense. Injuries didn't help, and Varly wasn't able to cover for the mistakes in every single game.

You don't need inaccurate advanced stats to pick and choose from, when all you have to do is look at the team, and how it's playing to see it's weaknesses.

 

[BB1133]

I'll never completely agree with analysis like this.

PDO could mean MANY THINGS. Common sense tells me a high S% and SV% could come from skilled players. The combination of both could come from CONFIDENCE as well as luck. And the assertion that every team is the same is noted when comments like "everyone will regress/approach 1000 at some point", and is wrongly used as an absolute, when it shouldn't logically be used.

As for the comment, "the math almost always wins", I have a couple problems with that.

1. It's not the math that is right, the math was created to quantify certain theories about success deriving from teams who have the puck more. DERP. Don't pretend, math appliers, that you figured out hockey.

2. Though it is true that the more you have the puck, the better you are probably going to be, there still can be some anomalies, even though Colorado wasn't that TO AN EXTENT, but they were in a certain SAMPLE SIZE. Sample size is extremely important when analyzing data.

3. I don't get the fascination with the search for the "all-knowing" stat. As far as I know, hockey is fun to watch because YOU DON'T KNOW WHO WILL WIN. Once we get to the point where the "math" can tell us who is going to win before the game happens, then what will be the fun of watching hockey anymore?

I agree that the Wild series isn't a good example of how the Avs "prove the rule". It is too small of a sample size, but when you're going up against better puck possession teams night in and night out, the chances of winning aren't great. If they didn't lose to the Wild, they would've lost to the Hawks.

True exceptions to the rule? Probably Boston since they have high puck possession, quality depth at forward, and the best goalie in the league. There is no smoke and mirrors with their numbers.

The Ducks are the only other team left standing with an inflated PDO due to a high team shooting %. Will that continue against the Kings and Hawks all the way to the cup? Highly unlikely.