Advanced stats/ Moneypuck

[Larry Goodenough]

I've really been getting into advanced stats this year and have been fascinated in how they can be used to look at players/teams/management in a whole new light.

The most interesting part is how an evaluation can now be backed up with evidence, supported by scientific method. You have an opinion, the advanced stats geek has proof.

My ongoing question was how much pro teams use advanced stats in their own evaluations. Gillis and Gilman have admited to it somewhat, but I think they don't want to tip their hand too much so as not to give away some trade secrets.

Here's 2 Edmonton newspaper articles on it. Hey, the Oilers have announced they've formed an analytical team to help in decisions. Boston admits to already doing it for a while. That might say enough about advanced stats and their value right there.

http://blogs.edmontonjournal.com/2012/0 ... ck-issues/


http://blogs.edmontonjournal.com/2012/0 ... e-is-here/

[Ktulu]

The big problem with advanced stats in hockey is the human factor in recording the base events. Then there's the stupid names, Corsi? Fenwick? C'mon. Tom Tango didn't call Fielding Independent Pitching "Tango". BIll James didn't name any of his stats after himself.

[Potatoe1]

Fairly obvious the Canucks stay on top of this stuff.

Best example is how AV has changed the way he distributes ice time around face offs.

About 2 years ago behind the net posted an article regarding winning and losing offensive / defensive zone face offs. they went through mounds of data and came out with the theory that every face off win in the offensive zone was the same value as 15 to 20 second of power play time (I have no idea the exact number TBH).

Over the past 2 years AV has migrated from distributing his forwards like most coaches (i.e. line matching) to almost exclusively deploying forwards based on where the face off is.

In other words the Sedins take almost every offensive zone draw, and Manny takes almost every defensive zone draw. AV doesn't even match lines at all anymore, if it's an offensive zone draw and the Sedins are fresh they go out, on the defensive side it's Manny and Lappy, the other 2 lines just fill in the gaps.

This is in stark contrast to not only what basically every other team does but in fact to what we were doing 2 years ago.

[Larry Goodenough]

The big problem with advanced stats in hockey is the human factor in recording the base events. Then there's the stupid names, Corsi? Fenwick? C'mon. Tom Tango didn't call Fielding Independent Pitching "Tango". BIll James didn't name any of his stats after himself.

Well, the stats guys would say the human factor evens itself out the bigger the sample size gets. For every human error one way, another human error will eventually be made the other way.

Human factors are often at play in baseline stats as well.

[Waffle]

Statistics are very interesting, in and of themselves, but also because of the way people use them.

At Hockey Prospectus they use them extensively it seems when their stats support an article’s point, but ignore them when it doesn’t...an article yesterday about the Hodgson trade is a good example of avoiding them (http://www.puckprospectus.com/article.p ... cleid=1265, The Blue Line...Darcy Regier Steals The Show). Posters on these forums have given more reasoned “Corsi event” posts about the trade than Hockey Propectus did.

I was recently amused when someone (I think it was Cam Charron) commented on the Canucks not celebrating goals scored against the Leafs because his take was that the Canucks had come to realize that “a goal is just another Corsi event.” While it is, it also isn’t...Corsi events don’t win games, scoring more goals than the other team does.

I also think that advanced stats can really add to the objectivity of someone’s point of view and really enjoy it when people use them. But there are many knowledgeable hockey fans here whose opinion will never be replaced by stats. I just hope the Board doesn’t develop its own set of advanced stats to determine who they are

[Larry Goodenough]

Fairly obvious the Canucks stay on top of this stuff.

Best example is how AV has changed the way he distributes ice time around face offs.

About 2 years ago behind the net posted an article regarding winning and losing offensive / defensive zone face offs. they went through mounds of data and came out with the theory that every face off win in the offensive zone was the same value as 15 to 20 second of power play time (I have no idea the exact number TBH).

Over the past 2 years AV has migrated from distributing his forwards like most coaches (i.e. line matching) to almost exclusively deploying forwards based on where the face off is.

In other words the Sedins take almost every offensive zone draw, and Manny takes almost every defensive zone draw. AV doesn't even match lines at all anymore, if it's an offensive zone draw and the Sedins are fresh they go out, on the defensive side it's Manny and Lappy, the other 2 lines just fill in the gaps.

This is in stark contrast to not only what basically every other team does but in fact to what we were doing 2 years ago.

Yes, AV has the Sedins taking 80% of their total draws in the offensive zone and Malhotra has been taking 88% of his draws in the d-zone. These are the most severe deployments possibly in the history of hockey. However, this changed in the last 2 games since Pahlsson showed up. It's putting your best offensive and defensive players in the best positions to succeed. Ron Wilson's highest Ozone starts went to Lupul and Kessel - but they weren't even 55%. One guy is in 1st place again, the other is gone. Enuff said.

Where I find advanced stats most interesting is the puck possession stats of Corsi (differential between total shots directed at the opponents net vs your net) and Fenwick (total shots minus blocked shots). The stats guys claim the clear empirical evidence shows Corsi/Fenwick are the stats that most directly correlate with winning games over a large sample. The league leaders in Corsi every season since these stats started being counted have all made it to the cup finals. Vancouver led the league last year. Not so much this year.

Simply put, if you have 55% of the Corsi events, you have possession of the puck for 55% of the time. 55% of the time you have a chance to score and only 45% of the time you can be scored on.

Raymond is everyone's favourite whipping boy, but AV keeps sending him out there on the 2nd line. Why does he not see what the average fan sees? He must be an idiot or plays favourites. But how does he have his team 1st overall again if he's an idiot?

Oddly, Raymond had the best Corsi rating on the team last year and again has strong Corsi stats this year. So, when Raymond is on the ice, he may be be falling down alot and losing the puck in the offensive zone, but his corsi numbers show he's also got the puck alot and his defence is suppressing shots against his goaltender. Simply put, his team has a greater chance to score than to be scored against when he's on the ice to the highest degree on the team. A team that is filled with some pretty good players. I assume that's what value he brings and that keeps him in the top 6 or 9.

[Larry Goodenough]

Statistics are very interesting, in and of themselves, but also because of the way people use them.

At Hockey Prospectus they use them extensively it seems when their stats support an article’s point, but ignore them when it doesn’t...an article yesterday about the Hodgson trade is a good example of avoiding them (http://www.puckprospectus.com/article.p ... cleid=1265, The Blue Line...Darcy Regier Steals The Show). Posters on these forums have given more reasoned “Corsi event” posts about the trade than Hockey Propectus did.

I was recently amused when someone (I think it was Cam Charron) commented on the Canucks not celebrating goals scored against the Leafs because his take was that the Canucks had come to realize that “a goal is just another Corsi event.” While it is, it also isn’t...Corsi events don’t win games, scoring more goals than the other team does.

I also think that advanced stats can really add to the objectivity of someone’s point of view and really enjoy it when people use them. But there are many knowledgeable hockey fans here whose opinion will never be replaced by stats. I just hope the Board doesn’t develop its own set of advanced stats to determine who they are

Good points. As with all stats, you can somewhat twist them to fit your agenda.

I think Charron was commenting more about his obsession with Corsi than anything else.

[Waffle]

I am not an expert on the advanced stats, but thought it might be interesting to compare Mason Raymond and Chris HIggins. I picked this data off Behind the Net for 2011-2012 so far
http://www.behindthenet.ca/

Relative Corsi for 5 vs 5..........Goals for On/60..........Goals Against On/60
Daniel Sedin..........23.3............................3.26........................1.83
Henrik Sedin..........20.6............................3.38........................1.91
Alex Burrows..........16.6............................3.46........................1.91
David Booth............15.3............................2.42.......................2.53
Ryan Kesler.............14.4............................2.96.......................1.63
Mason Raymond.........4.5.............................2.82.......................2.00
Chris Higgins............2.0.............................3.13.......................2.18
All other forwards have negative relative Corsi

Mason Raymond has 58.3% of his starts in the offensive zone, but finishes in the offensive zone 48.9% of his shifts. Chris Higgins has 46.7% offensive zone starts, but finishes in the offensive zone 50.1%. Chris Higgins plays against somewhat tougher competition. There does not appear to be much difference in Corsi numbers overall related to the teammates they have played with.

For every 60 minutes that Mason Raymonds plays 5 on 5 versus every 60 minutes that he doesn't play, the Canucks are better off by a net 4.5 shots attempted than their opponent, and Chris Higgins only 2. Despite this, Vancouver actually scores more goals when Chris Higgins is on the ice, though they also give up slightly more (but remember Chris Higgins has been playing against tougher competition overall).

Mason Raymond is an advanced stat anomaly. When he is on the ice there are twice as many positive Corsi events compared to Chris Higgins, but fewer of them result in goals.

My take on this is pretty much what people on these forums have been saying. Vancouver would probably do better with Chris Higgins on the second line if it is expected to produce goals.

[Hockey Widow]

Yes, Higgins should be on the second line. But what would that do to our third line having Raymond on it. Would it still be a strong enough checking line? I think it would be fine is my answer so I am all for the change.

I don't want to see Kassian moved off the 4th line at this point in time. I think it is good for his game and good for the team to have him play there for as long as possible, this season, if the 4th line keeps playing the way it did against St. Louis.

[Larry Goodenough]

I am not an expert on the advanced stats, but thought it might be interesting to compare Mason Raymond and Chris HIggins. I picked this data off Behind the Net for 2011-2012 so far
http://www.behindthenet.ca/

Relative Corsi for 5 vs 5..........Goals for On/60..........Goals Against On/60
Daniel Sedin..........23.3............................3.26........................1.83
Henrik Sedin..........20.6............................3.38........................1.91
Alex Burrows..........16.6............................3.46........................1.91
David Booth............15.3............................2.42.......................2.53
Ryan Kesler.............14.4............................2.96.......................1.63
Mason Raymond.........4.5.............................2.82.......................2.00
Chris Higgins............2.0.............................3.13.......................2.18
All other forwards have negative relative Corsi

Mason Raymond has 58.3% of his starts in the offensive zone, but finishes in the offensive zone 48.9% of his shifts. Chris Higgins has 46.7% offensive zone starts, but finishes in the offensive zone 50.1%. Chris Higgins plays against somewhat tougher competition. There does not appear to be much difference in Corsi numbers overall related to the teammates they have played with.

For every 60 minutes that Mason Raymonds plays 5 on 5 versus every 60 minutes that he doesn't play, the Canucks are better off by a net 4.5 shots attempted than their opponent, and Chris Higgins only 2. Despite this, Vancouver actually scores more goals when Chris Higgins is on the ice, though they also give up slightly more (but remember Chris Higgins has been playing against tougher competition overall).

Mason Raymond is an advanced stat anomaly. When he is on the ice there are twice as many positive Corsi events compared to Chris Higgins, but fewer of them result in goals.

My take on this is pretty much what people on these forums have been saying. Vancouver would probably do better with Chris Higgins on the second line if it is expected to produce goals.

Nice work. I've been thinking Higgins should be bumped up to the 2nd line as well. Good to see the stats back that up.

I hope he's being held back just so he can recoup some energy after all the infection stuff. Then onto the 2nd come April.

[ESQ]

The big problem with advanced stats in hockey is the human factor in recording the base events.

I don't know much about advanced stats, but I'm curious what you mean by this?

The human factor in things like hits and faceoffs is huge, and I don't think it will "even out" in the averages. For example, Minnesota's hit stats will be completely skewed because their stats keepers at XCel Energy Center love recording stats. Similarly, the NYI's faceoff stats are flat-out wrong. I remember the game there last year where Manny wasn't credited with a single faceoff win, maybe because they had an agenda to get Zenon Kenopka the #1 faceoff guy in the League? When you play half your games in a lopsided building, it won't even out.

But I don't know if I consider those to be advanced stats. I haven't fully wrapped my head around Corsi, but its based on ice time and goals/shots? Seems to me those have a minimal human element.

I think the Canucks have their own system, and you hear it in AV's post-games when he talks about "Grade A scoring chances". I think their own stats keepers use their own criteria to grade scoring chances for and against as their own version of Corsi.

[Arachnid]

The big problem with advanced stats in hockey is the human factor in recording the base events.

I don't know much about advanced stats, but I'm curious what you mean by this?

The human factor in things like hits and faceoffs is huge, and I don't think it will "even out" in the averages. For example, Minnesota's hit stats will be completely skewed because their stats keepers at XCel Energy Center love recording stats. Similarly, the NYI's faceoff stats are flat-out wrong. I remember the game there last year where Manny wasn't credited with a single faceoff win, maybe because they had an agenda to get Zenon Kenopka the #1 faceoff guy in the League? When you play half your games in a lopsided building, it won't even out.

But I don't know if I consider those to be advanced stats. I haven't fully wrapped my head around Corsi, but its based on ice time and goals/shots? Seems to me those have a minimal human element.

I think the Canucks have their own system, and you hear it in AV's post-games when he talks about "Grade A scoring chances". I think their own stats keepers use their own criteria to grade scoring chances for and against as their own version of Corsi.

I think the Human Factor is the women. Wives & girlfriends have a big impact on the players.
I volunteer to test this theory! A simple sampling of them will determine if they are a positive or negative influence.
Something I like to call PuckControl

[Ktulu]

The big problem with advanced stats in hockey is the human factor in recording the base events.

I don't know much about advanced stats, but I'm curious what you mean by this?

The human factor in things like hits and faceoffs is huge, and I don't think it will "even out" in the averages. For example, Minnesota's hit stats will be completely skewed because their stats keepers at XCel Energy Center love recording stats. Similarly, the NYI's faceoff stats are flat-out wrong. I remember the game there last year where Manny wasn't credited with a single faceoff win, maybe because they had an agenda to get Zenon Kenopka the #1 faceoff guy in the League? When you play half your games in a lopsided building, it won't even out.

But I don't know if I consider those to be advanced stats. I haven't fully wrapped my head around Corsi, but its based on ice time and goals/shots? Seems to me those have a minimal human element.

I think the Canucks have their own system, and you hear it in AV's post-games when he talks about "Grade A scoring chances". I think their own stats keepers use their own criteria to grade scoring chances for and against as their own version of Corsi.

Ya that's pretty much what I meant. A lot of the base events used to compile advanced stats are based on subjective human factors. Even basic stuff like shots on goal, we've all seen games where the shots simply don't make sense. While faceoffs due to tend to even out with large sample sizes, there's a lot of subjectivity there. How often do you see someone "lose" the faceoff but the team actually gains possession? Then there's just straight up bias. I remember when Van went into Washington last year and Manny had something like a 35% faceoff success rate and Steckel dominated. You look at Manny's game log and in the surrounding dozen games he wasn't under 50% once... something fishy there.

[mathonwy]

Cool beans guys. Great analysis!

This thread really shows the quality of the posters on this board.

Here's my non-statistical analysis. What Chris Higgins brings to the 2nd line is puck retrieval pure and simple. Chris has the strength and tenacity to be able to go into the corners, retrieve the puck and pass it back to the point or to re-cycle.

Raymond has the tenacity but not the strength so he goes into the corner and starts puck handling. Inevitably, someone says, fuck you little man and knocks him on his ass. I do give Raymond credit though as his game this year is a lot more dynamic than last year. He's creating more chances and he's using more than just his speed.

It's a bit of a catch 22. The stats show that when Raymond is holding onto the puck, we get scored on less. But because he's holding on to the puck more, our 2nd line is less effective.

So we send Raymond down to the 3rd line and bring the Amex line back together. Our 2nd line becomes more effective but our checking line suffers. Raymond just doesn't have the strength to hold up forwards that are stronger than him nor can he win the grinding battles along the board.

Well.... fuck.

I think I just talked myself into leaving Raymond on the 2nd line and coaching him to stop puck handling so much.

[Larry Goodenough]

Some more blogs related to Burke/Chiarelli and their comments at the advanced stats symposium last week. Here's the interesting quote...

But when you hear that Chiarelli has internalized some data that he hasn’t released to the public, and Burke goes off pretty convincingly against the statistics, (calling much of it “horseshit” and falsely states that “nobody has ever won a championship with Moneyball”) and looking at the position of those two teams in the standings, it’s apparent that there’s at least some value to this.

http://blogs.thescore.com/nhl/2012/03/0 ... -peverley/

http://blogs.thescore.com/nhl/2012/03/0 ... part-deux/