All-Purpose Advanced Stats, Player Usage, etc. Thread

[Djp]

Ristolainen is a bust (heave stats related)

They say in one measure Risto is one of the worst Dmen

based on this article:

http://www.habseyesontheprize.com/2...orst-possession-players-in-the-nhl-defencemen

Ehrhoff and Myers are boith in top 25.

My head hurts reading this.....I cant download the spreadsheet now.

 

[tsujimoto74]

Bit overdramatic with the thread title, no? It's nice to see some solid advanced stat work confirming what we already know about how Myers has played this year, though. I'm gonna be doing a whole lot of and if Murray actually trades him.

 

[Mit Yarrum]

Good lord what a terrible title

 

[dotcommunism]

There's absolutely nothing in that link that states that Ristolainen is a bust, so clearly this is how you chose to interpret things. That's utterly ridiculous. He's been a bad possession defenseman to this point in the NHL, but using that to label him as a bust at this point is a tad melodramatic.

He's a 19 year old defenseman, adjusting to a new league and a new country. He also hasn't even spent half the season in the NHL, and most of his sample of NHL games is from before he played a minute in the AHL. This is also a pretty awful possession team, period. The fact that Myers and Ehrhoff are highly ranked there doesn't really reflect on Ristolainen because their circumstances are vastly different

So you're taking data, stripping it of any context, and using it to support a rather wild conclusion. I don't know if you're pushing an agenda here or simply trying to be provocative, but calling Ristolainen a bust based on this is just silly

 

[JLewyB]

Let's not forget that Risto played most of his games under Rolston.

edit: I'm pretty Myers was one of the worst posession D-men in the league until November 15th.

 

[Djp]

So you're taking data, stripping it of any context, and using it to support a rather wild conclusion. I don't know if you're pushing an agenda here or simply trying to be provocative, but calling Ristolainen a bust based on this is just silly

Its called sarcasm.....

Have I ever said Risto is a bust in any other thread.

I think stats are overdone---I am a statistician so I know what I am talking about. Its easy to create measures---but what exactly are you measuring with thgem is up for debate.

 

[Woodhouse]

Its called sarcasm.....

Have I ever said Risto is a bust in any other thread.

I think stats are overdone---I am a statistician so I know what I am talking about. Its easy to create measures---but what exactly are you measuring with thgem is up for debate.

Sarcasm in a new thread title is disruptive though. The topic and your point really should be clear to other users, especially when you joke Ristolainen's a bust because of some warped interpretation of poor possession numbers in a small sample size predominantly under Rolston.

 

[Jim Bob]

http://www.pensionplanpuppets.com/2...m-murray-reads-us-and-offers-some-thoughts-on

Triple P listens to WGR....

 

[Woodhouse]

Unstuck. Please bookmark if needed. Thanks.

 

[struckbyaparkedcar]

NHL scoring leaders in close, 5 on 5 situations.

http://stats.hockeyanalysis.com/rat...d=0&type=individual&sort=ipoints&sortdir=DESC

Considering Thomas Vanek got traded for ****ing nothing the second time around, I'm surprised to see him in the top 15. The on ice goal stats under these limitations are quite interesting as well. Anze Kopitar is un****ingreal.

 

[Paxon]

Bumping because I really wanted to show this cool chart which shows "chemistry" between players. I don't think that's really the accurate way of putting that, more matter-of-factly it demonstrates links in production between players.

http://iguana.cs.toronto.edu/

 

[Jim Bob]

Here is an interesting use of analytics.

 

[Paxon]

NHL website to feature Corsi, Fenwick beginning in February

NHL.com will begin offering Corsi, Fenwick, PDO and other enhanced statistics next month, NHL chief operating officer John Collins confirmed to Yahoo Sports on Wednesday.

The additions are scheduled for late February.

"You’re going to see a big change in the way we present our stats, in terms of the depth and the utility of how to do it," Collins said. "And that’s before the puck tracking (system)."

The new offering will include Corsi, Fenwick, PDO, zone starts, average shot distance, goals and assists per 60 minutes and penalties drawn and taken per period and per game.

It will date back to the 2010-11 season and will be incorporated into the existing player and team statistics pages.

Last summer, multiple NHL teams hired analytics experts, and the league's website changed its terms of service shortly after, prohibiting other sites from re-purposing its own statistical information.

 

[joshjull]

Here is an interesting use of analytics.

If we could only get the knucklehead parents and organizations that resist cross ice to see this.

 

[sjci]

http://www.thehockeynews.com/blog/w...ics-former-nhl-goalie-steve-valiquette-knows/

One night, Steve Valiquette was removing his gear in the dressing room after one of his many games as a backup in the NHL. He remembers the coach of the team approaching one of his teammates and praising the player for having such a great game because he recorded nine shots on goal. Valiquette couldn’t believe what he was hearing, but he knew what he had seen.

“I was on the bench that night and I said to myself, ‘Are you kidding me? He just played catch with goalie from 75 feet,’ ” Valiquette said. “ ‘He should be reprimanded, not applauded for that game.’ ”

And so began Valiquette’s quest to debunk the theory that all shots are created equal. They aren’t, and Valiquette is frustrated that people look blindly at things like shots on goal and save percentage and take them at face value. What he wants to see is goaltenders rated on the basis of how many difficult shots they face.

 

[Jim Bob]

A goalie coach I know has three rules for goaltending:

1 - Set your feet
2 - Get square
3 - Challenge when you can

Shots that come so quickly that you can't do 1 and/or 2 are much tougher to stop.

Also, traffic in front ups the level of difficulty.

 

[Jim Bob]

If we could only get the knucklehead parents and organizations that resist cross ice to see this.

I really think one thing that will help is when we get to the stage where no U8/Mite kids have older siblings that played full ice at that age.

One of the most maddening parents I know on this topic were in that group with an older kid that played full ice and had a younger kid that fought the cross ice idea because that's not what their older sibling did.

The same thing goes for moving checking to Bantams.

 

[mikemcburn]

Bumping because I really wanted to show this cool chart which shows "chemistry" between players. I don't think that's really the accurate way of putting that, more matter-of-factly it demonstrates links in production between players.

http://iguana.cs.toronto.edu/

As I mentioned in another thread when posting this link, I think you're right that it does not really show "chemistry".

But if you combine it with the context of who generally plays with who, it can maybe show a lack of chemistry?

In the other thread I flagged Moulson's thick connections to Ennis& Girgensens, Girgensens/everyone & Hodgson/Stewart as examples of this. Looking at Girgensens and knowing that he's played with just about everyone in every situation, it's interesting to see that his "connection lines" are just about everywhere. That suggests to me that he finds a way with whoever.

Comparably, given how much Hodgson & Stewart have been stuck together and yet the tenuous line between them, I think it shows the wholesale lack of chemistry between these guys. The fact that Stewart's lines spray out to other guys shows he's been more effective (offensively anyway, lol) when being given more offensive opportunities. Gee, go figure.

For Moulson as the other example, sure he's got thick lines with Ennis & Girgensens, but is that really "chemistry" or simply demonstrating that he generally plays with those guys?

Really I think this just shows, with context, where there might be 1) a lack of chemistry (ie: Stewart/Hodgson), 2) guys who can produce with anyone in any situation (ie: Girgensens) and *possibly* 3) guys will be more effective with anyone if plugged into prime offensive opportunities (ie: take away just Stewart's PP time, and he'll have 3 more points (goals) than Hodgson, who doesn't see an ounce of PP time anymore).

Context is everything. But it's a really cool chart!

Speaking of Stewart again though... they gotta be pumping a trade. The guy is our arguably our worst defensive player and certainly the give-away king of the team, but this is never mentioned... Poor Ennis and Girgensens who have to deal with him now, lol.

 

[Jim Bob]

http://www.tsn.ca/why-analysis-of-goalies-is-so-difficult-1.348286

The Takeaway

Goaltenders are voodoo. Statistical analysis has proved only mildly helpful as it pertains to identifying goaltending talent and forecasting future performance. Teams also struggle mightily with goaltender evaluation – from the time of the draft, to the time in which they decide to fork over the first (or second) lucrative contract. Down the road, technological improvements may pay significant dividends here, as metrics will be crafted to better identify goaltending talent, and separate skill from the noise.

http://ingoalmag.com/analysis/voodoo-madness-mysticism-excuses-for-goaltending-ignorance/

What we need is clear. Anyone who follows goaltending knows that what the puck and goalie do before the shot determines how difficult the save will be. Unfortunately, the projects dedicated to tracking shot quality are proprietary, meaning they are, ironically, creating even more mystery instead of providing clarity. Ryan Stimson’s fascinating, publicly-available passing project will be tracking many goaltender-relevant variables this year, but any lone endeavor will necessarily account for only part of what goes into shot quality.

A general lack of goaltending knowledge means that many with the intelligence, creativity, and technical savvy to move the analysis of the position forward feel unqualified. Those without a basic understanding of goaltending will have a hard time forming and testing theory, and may well fail to see whatever distinctions their data is trying to show them.

There is no easy solution to the current problem advanced statistical analysis has with goaltenders. A first step is to recognize that the mythology of the incomprehensible goaltender is just that: pure myth. The position is neither madness, nor mysticism, nor magic. The tacit permission these myths have given hockey professionals to remain ignorant about goaltending is hereby rescinded. If your job is to understand, analyze, and explain the game, it’s no longer acceptable to know nothing about one of its most important aspects.

Until the comfortable old false narratives are dead, nothing will change.

An interesting point-counterpoint on the attempt to predict goaltender performance using analytics.

 

[Zip15]

View attachment 85719

This diagram shows our 5-on-5 shot rates for and against vs league average. A number <1 means fewer shots are generated from these three respective areas--essentially, those are "outside," "medium distance," and "home plate (high danger)"--while a number >1 means more shots than league average are being generated from an area.

As you can see in hour shots for graph on the left, we're doing a pretty good job generating chances in the most dangerous area, while we take fewer shots from medium distance and just below league average from outside. Meanwhile, on the right, that graph shows we're doing an excellent job preventing shots in that home plate area and an above-average (good) job in the medium and long distances, as well.

But, we're a -10 at 5-on-5, and you'll see why here...

View attachment 85721

These are the shooting percentages relative to league average. Once again, a number <1 means the shooting percentage is less than league average, while a number >1 means the shooting percentage is higher than league average.

Let's look first at Buffalo's shooting percentage relative to league average, which is illustrated by the left graph. Despite taking an above-average amount of shots in that "home plate" area, our relative shooting% (compared to the league) is abysmally low--.375 is way, way below the league average represented by 1.00. In addition, our shooting percentage from medium distance is well below league average, as well, while our long distance shooting% is below average, though not by nearly as much as the other two areas. We can reasonably conclude that we should start scoring more at ES as our shooting percentage from the high-danger area starts moving towards league average. For a frame of reference, in 2014-15, our relative shooting% number was 1.12, meaning we shot above league average from the home plate area. Accordingly, I expect that .375 number to significantly improve over time, and the goals should get more frequent provided we continue generating high-danger shots at 5-on-5.

Next, just the opposite is happening with our 5-on-5 shooting percentage against (i.e., our save percentage). Though we do a good job of suppressing shots in that high-danger area, our goaltending has sucked on these shots--I mean it has really sucked on those shots--as demonstrated by our 1.84 number. In addition, our goaltending has been very poor on medium distance shots, as well, as that 1.5 number shows (i.e., giving up a shooting% well above league average from that distance). Finally, our goaltenders do a slightly above-average job at saving long-distance shots. In sum, if our goaltenders can get closer to league average (or at least not be one of the worst in the league) at saving home plate and medium distance shots, that would improve our ES GA number.

These graphs make me optimistic. We're getting to the high-danger areas more than the league average, while doing a good job of suppressing those dangerous shots. Further, those numbers should correct a bit over time (more GF) while our SV% should also get a bit closer to league average (fewer GA).

TL;DR - At 5-on-5, we're creating high-danger scoring chances and suppressing high-danger chances against. However, we are doing a poor job of finishing those high-danger chances, and our goaltenders are doing a poor job of keeping the puck out of the net, relative to league average, when we do surrender shots from close and medium distances.

 

[jc17]

TL;DR - At 5-on-5, we're creating high-danger scoring chances and suppressing high-danger chances against. However, we are doing a poor job of finishing those high-danger chances, and our goaltenders are doing a poor job of keeping the puck out of the net, relative to league average, when we do surrender shots from close and medium distances.

Thanks, this is exactly why the people who were wondering about another tank this year are crazy. Our team is playing well, just unlucky so far.

 

[Woodhouse]

 

[B U F F A L O]

We're going to be the anti-leafs from a couple seasons ago when they made the run to the playoffs despite terrible numbers, all thanks to our PK and goaltending. A statistical outlier where all advanced stat guys shake their heads at our great numbers and terrible results.

 

[FearTheBeard]

We're going to be the anti-leafs from a couple seasons ago when they made the run to the playoffs despite terrible numbers, all thanks to our PK and goaltending. A statistical outlier where all advanced stat guys shake their heads at our great numbers and terrible results.

On the plus side we know where we need to make offseason improvements and we can add a stud in the draft to go with our already impressive young core. Assuming we go all year like this that is...

 

[Zip15]

We're going to be the anti-leafs from a couple seasons ago when they made the run to the playoffs despite terrible numbers, all thanks to our PK and goaltending. A statistical outlier where all advanced stat guys shake their heads at our great numbers and terrible results.

If you listen to Pens fans--and you really shouldn't, because they are angry, angry people when it comes to Bylsma--the wins won't be coming "because Bylsma." You know, because those Bylsma teams in Pittsburgh had real problems winning games.

According to War on Ice, here are our CF% ranks by score situation:

Close: 50% (14th)
Score-Adjusted: 51.2% (11th)
All: 53.3% (5th)

That last number is due in no small part to how bad we killed the Pens while down two goals for 17 1/2 minutes in the 3rd period, which is not considered a "close" situation. Still, we're way ahead of where I thought we'd be, and this is without Bogosian all season, without Kane for the last two games, and without Ennis last night.