Advanced Statistics vs Team Standings

[Larry Hoover]

I don't know much about advanced statistics so I'm hoping someone can help me out here and then I can edit it into the first post.

What my question is, is what do the advanced statistics say about the NHL standings? Who is getting lucky, who is getting unlucky, etc...

Using PDO, Fenwick close, what can you tell me about all the teams?

Click to read more...

 

[Unaffiliated]

http://stats.hockeyanalysis.com/teamstats.php?db=201213&sit=5v5&disp=1

^ Good site.

Don't have time to sort through all that for you, but it's easy enough to use.

 

[number72]

Fenwick correlates with 40% of winning (P%). So teams that win tend to have a high fenwick (loosely they tend to outshoot or out chance their competition). Correlation doesn't mean it causes winning but that teams that win tend to outshoot. 60% of the game is either not captures by Fenwick (grit, intangibles, clutchiness etc) or is luck driven (50/50 puck battles etc).

PDO is team SH% + SV%. If there is parity in the league then the PDO should be approx 1000. And so a team that has a PDO well above 1000 is either highly skilled (have elite goalie or elite shooters) OR they are benefiting by luck. The goalie SV% is higher because of bounces, a string of good and above average games etc. If the reason is luck then over the course of the season the teams PDO (shooting % and SV%) should move back towards historic average.

They are guidelines to consider but can and do breakdown. However, they may suggest a team like anaheim is playing above their heads and will eventually come crashing down. But they can't tell you when anaheim will crash - this year, next year etc. But simply, the probabilty is that they will start playing more average hockey

 

[MastuhNinks]

I'm a supporter of advanced statistics but lately it seems a lot of people are misunderstanding how to use them. Just because a team or player (e.x. Anaheim or Kunitz) is most likely going to regress to the norm in shooting percentage, doesn't necessarily mean they're just getting lucky. It means they're playing unsustainably good, obviously puck-luck is a thing but it's extremely closed minded to suggest the only way someone can have a high shooting percentage is by being lucky. Chris Kunitz is not getting lucky, he's just playing by far the best hockey of his career, and that combined with playing with the best player in the World has lead to him having a high shooting percetage. Is it likely to go down? Definitely, but that doesn't automatically mean it's luck-driven. There's a reason Steven Stamkos has had such a high shooting percentage throughout his career, and it isn't because he's lucky.

 

[walle]

Fenwick correlates with 40% of winning (P%). So teams that win tend to have a high fenwick (loosely they tend to outshoot or out chance their competition). Correlation doesn't mean it causes winning but that teams that win tend to outshoot. 60% of the game is either not captures by Fenwick (grit, intangibles, clutchiness etc) or is luck driven (50/50 puck battles etc).

They are guidelines to consider but can and do breakdown. However, they may suggest a team like anaheim is playing above their heads and will eventually come crashing down. But they can't tell you when anaheim will crash - this year, next year etc. But simply, the probabilty is that they will start playing more average hockey

I think Ducks are a good example why fenwick is a good and bad stat depending how you use it. Now they're about 46-47% Fenwick team which is almost/about bottom 5 in the league. So they should crash down you could easily assume. But lets take a closer look inside their team and what fenwick tells. Their top6 players are about 49-52% while bottom 6 are 40-45%. So most importantly their top players are performing and evening things up, and as shutting down their top quality opponents they're also limited to 50-50 range thus bottom 6 "getting beat" is not as bad as they're also getting beat by lower quality opponents who's skill and finish aint gonna as likely result in goals against. My point is that using advanced stats should also be used in comparing roles in the teams and players performing those roles so we can have clear picture whats going on and how. Ducks look worse than they're because they've their important roles are working, which in the end are the ones winning games.

 

[Moskau]

Toronto is abysmal in advanced stats. Had no clue.

 

[Theridion]

If I gamble a dollar on coin flips, and a I at it 3 times in a row, you can't tell me I am destined to lose, in order to return to a statistical average of 50/50.

However, if you tell me I am going to flip the coin 5,000 times, we can agree on a pretty much 50/50 split.

If a team has an amazing record (i.e. undefeated blackhawks) through 20 something games, its childish to say that they are destined to lose or they are destined to go on a losing streak. Of course they are. Everyone is bound to lose. No different than a team going 1-10 eventually breaking out of a slump.

People like to poop on other fan's teams, and its way easy to make yourself sound cool if you bring up fancy stats that other people have to google in order to try to refute your claims.

I love statistics. And I love sports numbers and theory. You always pop into these presidential streaks where tallest or youngest or whatever politician always wins the race... or those born earlier in the year have a better chance of winning an election.... etc. Maybe there is some truth that causes them to be reliable up to this point.

As far as Fenwick, Corsi, and teams like the Ducks go, I have read way too many people say, "Their fenwick is bad, they are going to lose". Yeah. They are. They are going to not only lose some gains, but probably get a losing streak or two. This year, playoffs, next year...

Only actually playing out a season tells the truth, because it makes the truth. If you win games, then at the end of the season, you were a good team. If you lose games, then you were a bad team.

When was the last time a team had a .600 winning percentage with a bottom 15 powerplay? A bottom 10 powerplay? A bottom 10 penalty kill during the regular season?

Lots of stats out there.

 

[Kershaw]

Last year the Rangers had a mediocre fenwick close last year and the experts on the board were saying they were due for regression. The regressed their ways to the division title and ECF appearance.

This season, the Rangers have one of the best fenwick close ratings and they're a mediocre team.

 

[Daz28]

What is the mean of everything? It's half.

Just for kicks what is a teams "historic average"? It's the middle, that's what it is. iow, the mean. People need to quit trying to prove anything with their silly stats, cuz they can't prove a freakin thing with them. Historically, it all evens out for EVERYONE, Right?

Ex: my grampa's old, will he die? Hold on, I'll check my fenwick corsi. Yup, he'll die. Silliness.

Last year the Rangers had a mediocre fenwick close last year and the experts on the board were saying they were due for regression. The regressed their ways to the division title and ECF appearance.

This season, the Rangers have one of the best fenwick close ratings and they're a mediocre team.

Lesson learned is?

Don't listen to them.

 

[Unaffiliated]

What is the mean of everything? It's half.

Just for kicks what is a teams "historic average"? It's the middle, that's what it is. iow, the mean. People need to quit trying to prove anything with their silly stats, cuz they can't prove a freakin thing with them. Historically, it all evens out for EVERYONE, Right?

The Blackhawks are statistically better than the Panthers this season.

Sidney Crosby is statistically better than Zanon Konopka this season.


Silly stats. They don't mean a thing. All the things above are equal. Right? They're all half?

 

[Bear of Bad News]

Just for kicks what is a teams "historic average"? It's the middle, that's what it is. iow, the mean. People need to quit trying to prove anything with their silly stats, cuz they can't prove a freakin thing with them. Historically, it all evens out for EVERYONE, Right?

The fact that you don't understand the difference between meaningful stats and meaningless stats doesn't make all stats meaningless.

 

[burf]

Last year the Rangers had a mediocre fenwick close last year and the experts on the board were saying they were due for regression. The regressed their ways to the division title and ECF appearance.

This season, the Rangers have one of the best fenwick close ratings and they're a mediocre team.

This anecdote is clearly damning to the proven correlation between fenwick and winning.

Oh wait, maybe it's just random chance, which is accounted for in any statistical regression model anyway.

 

[MasterofGrond]

Last year the Rangers had a mediocre fenwick close last year and the experts on the board were saying they were due for regression. The regressed their ways to the division title and ECF appearance.

This season, the Rangers have one of the best fenwick close ratings and they're a mediocre team.

Their possession numbers improved significantly towards the end of the season, for the record.

 

[Kershaw]

Their possession numbers improved significantly towards the end of the season, for the record.

And they played worse down the stretch to end the year relative to the time where people were saying they aren't going to make the playoffs at the rate they were playing because of their fenwick numbers.

The eye test and the stats should both be taken in with context. In the simple measure of the eye test, this years Rangers team is vastly inferior to last year's team, despite their high ranking in fenclose.

 

[GuineaPig]

The eye test and the stats should both be taken in with context. In the simple measure of the eye test, this years Rangers team is vastly inferior to last year's team, despite their high ranking in fenclose.

The problem is that humans freakin' suck at the "eye-test." Not only are we inherently biased, but we're extraordinarily responsive to suggestion, have selective memories, and are extremely outcome-dependent.

I'd argue that the principal differences between this year's Rangers and last year's is that Lundqvist has regressed to his mean and the Rangers aren't getting the bounces at the offensive end.

 

[Daz28]

The Blackhawks are statistically better than the Panthers this season.

Sidney Crosby is statistically better than Zanon Konopka this season.


Silly stats. They don't mean a thing. All the things above are equal. Right? They're all half?

So you're saying if ya have good players, your stats are good? I'm saying there's no substitute for knowing your stuff, and watching games. Moneyball all day if ya want, but that's for betting, not for fact.

btw, points aren't really a stat(hawks better than panthers). Points are the whole object of the game, and basically all that's relevant, except to maybe GM's(or bored people).

I'll oversimplify my originally oversimplified point. If the NHL exist for the next 200 years, what's the odds each year my team wins the cup. It's 1 in 30. PIT was good, then they were bad, then they were good again, then later on, they'll be bad again. I couldn't imagine doing this with the Canadiens "historically". Call Don Stark for that one.

The fact that you don't understand the difference between meaningful stats and meaningless stats doesn't make all stats meaningless.

Ya know which stat isn't meaningless, points(which, as I said isn't really a stat at all). As I also said, they can be a tool for GM's and arbitrators, or gamblers, but just slingin em around assuming their correlation is within some degree of error to be causation is meaningless to me. My current favorite is 'chance of making the playoffs', right down to a tenth of a percent. Awesome.

(I'm not saying you personally sling them like that either, but that's what most people use them for).


Unrelated, but along the same lines is the "eye test" that Kershaw mentions. If you haven't seen something with your own eyes, you ought not claim it, and if you did see it happen, then you don't need any stats.

Ya know, I apologize, because I forgot all about fantasy hockey. If people are using advanced stats for that, then my apologies. I just didn't think of it, because I don't play anything fantasy.

 

[BoHorvatFan]

The problem is that humans freakin' suck at the "eye-test." Not only are we inherently biased, but we're extraordinarily responsive to suggestion, have selective memories, and are extremely outcome-dependent.

I'd argue that the principal differences between this year's Rangers and last year's is that Lundqvist has regressed to his mean and the Rangers aren't getting the bounces at the offensive end.

And I'd argue that they traded away depth for a first line star, lost their toughness edge/identity, and probably have injuries affecting them... do those show up in advanced stats? no they don't because as far as you're concerned it's just a goalie regressing and a bit of luck that have impacted the team. That's a ridiculous way of looking at hockey IMO. There are many non number variables that come into play over the course of a hockey season that won't even show up in excel.

 

[JoelWarlord]

And I'd argue that they traded away depth for a first line star, lost their toughness edge/identity, and probably have injuries affecting them... do those show up in advanced stats? no they don't because as far as you're concerned it's just a goalie regressing and a bit of luck that have impacted the team. That's a ridiculous way of looking at hockey IMO. There are many non number variables that come into play over the course of a hockey season that won't even show up in excel.

So how much toughness are they missing? How much toughness would be enough? Do they need Prust back? Do they need a goon? Do they need 5 more fights or 10 more fights? This is the problem with these kinds of narratives, it's impossible to quantify and you can make the claim all day long and there's no way to prove or disprove it.

The depth part is definitely understandable, but it's not as if Nash has been a scrub for them this year. They got the best player in the deal and have a very talented group of forwards. The Rangers are shooting about 2% less than the league average, does this pass the sniff test? Does a forward core consisting of Richards, Gaborik, Nash, etc etc really shoot 2% off the mark?

 

[garret9]

Oh dear, some comments...

FenwickClose and PDO give a greater understanding on the standings because they do help you see the team performance against others and where the wins are coming from.

They do help indicate if a team's points is beyond what their performance should indicate, since more often than not if the whole thing was repeated multiple times with all other variables being equal they would be unlikely to have that.

We can then guess that they (team or individuals) are likely to regress to their mean because they have been helped by "luck".

Some people have issues with this because of two things: luck and regression.

The idea of luck bothers people as it seems to take responsibility away from the players work and effort, making it either controlled by fates or making it completely random. In actuality it describes the natural variance in life. A premium sniper can take the exact same shot, in the exact same location and situation, against the same goalie, multiple times and get different results. Maybe 9 times out of 10 that would go in, but in that one time the goalie gets it you could say it is lucky.

The problem with regression is sometimes people think that we're indicating that due to their "good luck" they are in run for some "bad luck". Merely what we mean is that teams that have low Fenwick tend to lose, more often than not. So far the team has won, but the probability of being like that is unlikely. It's not demanding that the world will be straightened out, just merely working with weighted probabilities.

Good quote from NHL numbers:

Outcomes in professional sports are weighted probabilities, not destinies, so it's entirely possible for the better team to lose on any given night or even over a brief sample of games, like a best of seven series, for no other reason beyond variance. There are also other influences beyond the control of the players, coaches and GM's of course: the officiating, injuries to key players, etc. Sports are interesting not only because of the action, competition and violence, but because they are a boiling cauldron of uncertainty. Sometimes the underdog wins. And sometimes it's not because of any particular failing of the favorite.

Also, remember that not all individuals are created equal. Stamkos will always have a better career SH% and OnIceSH% then Chris Thorburn as he has the talent to create the higher percentage plays. A team with more PP TOI will have a higher team SH%. A team with a stronger goalie will have a better team SV%.

But, when a team who has been mostly the same over the last few seasons has been the same (in makeup, SF, SA, Fenwick, etc) except for their PDO... don't expect your Thorburn to be turning into a Crosby.
(I used Thorburn because he's on the team I follow so not to hurt other people's feelings)

 

[TheDevilMadeMe]

Fenwick correlates with 40% of winning (P%). So teams that win tend to have a high fenwick (loosely they tend to outshoot or out chance their competition). Correlation doesn't mean it causes winning but that teams that win tend to outshoot. 60% of the game is either not captures by Fenwick (grit, intangibles, clutchiness etc) or is luck driven (50/50 puck battles etc).

PDO is team SH% + SV%. If there is parity in the league then the PDO should be approx 1000. And so a team that has a PDO well above 1000 is either highly skilled (have elite goalie or elite shooters) OR they are benefiting by luck. The goalie SV% is higher because of bounces, a string of good and above average games etc. If the reason is luck then over the course of the season the teams PDO (shooting % and SV%) should move back towards historic average.

They are guidelines to consider but can and do breakdown. However, they may suggest a team like anaheim is playing above their heads and will eventually come crashing down. But they can't tell you when anaheim will crash - this year, next year etc. But simply, the probabilty is that they will start playing more average hockey

This is the best assessment I've seen of the currently used advanced hockey stats. Thank you.

So often I see everything not captured by the formula du jour thrown out fans of these formulas as "just luck" and I don't think that attitude helps anything.

 

[BoHorvatFan]

So how much toughness are they missing? How much toughness would be enough? Do they need Prust back? Do they need a goon? Do they need 5 more fights or 10 more fights? This is the problem with these kinds of narratives, it's impossible to quantify and you can make the claim all day long and there's no way to prove or disprove it.

The depth part is definitely understandable, but it's not as if Nash has been a scrub for them this year. They got the best player in the deal and have a very talented group of forwards. The Rangers are shooting about 2% less than the league average, does this pass the sniff test? Does a forward core consisting of Richards, Gaborik, Nash, etc etc really shoot 2% off the mark?

A big part of their problem is having so much money tied up in 3 players and terrible scoring depth.

Toughness can't be measured by a number, but it did make up a big part of their identity last year and anyone that watched them knew it.

Hockey is a game based on thousands of split second decisions it's not a simplistic repetitive numbers game like baseball that is easy to break down, yet still not perfectly.

 

[BoHorvatFan]

Thousands of split second decisions that occur within various different systems used by teams and many other variables.

Saying things like ''the team that gets more shots is the better thing'' wow... really.... you need a bunch of stats to tell you that? obviously the team that controls the puck, spends more time in the offensive zone, and gets more shots is usually the better team. But there are ways to beat a better team that have to do with systems and positioning and decision making that won't show up on the score sheet.

 

[garret9]

Just because there are thousands of variables and split second decisions which act almost in randomness doesn't mean it can't be statistically modelled. Especially when there are particular end goals in mind that are defined to very particular acts.
In sciences like Ecology and Evolutionary Medicine, this is done all the time.
You aren't adding up all the micro-events together in order to determine a macro-model, but instead are starting with the macro-goals/endresults and working backwards.

 

[HanSolo]

According to advanced stats, Anaheim should be looking at a lottery pick in a real strong draft class.

What up Alex Barkov?

 

[Bear of Bad News]

According to advanced stats, Anaheim should be looking at a lottery pick in a real strong draft class.

Please elaborate (and be specific). Hockey Prospectus had the Ducks as on the bubble at the start of the year, and obviously didn't see Fasth coming to help tandem with Hiller.

Who (or which advanced stats) are suggesting that the Ducks would be a solid lottery team?