This day on November 20
None

Happy Birthday To: d_ottosen, lambofgad, Ofer, JayStyles, oilpride5

F.A.Q. Terms of Use F.A.Q. F.A.Q.
Members Members   Search Search     Register Register   Login Login   Home Home
 Speculation » Advanced MetricsPages (10): [ «  <  1  2  3  4  5  6  7  8  9  10]
Switch to flat viewSwitch to tree viewCreate a new topicSubmit Reply
 Re: Advanced Metrics [message #698836 is a reply to message #698831 ]
Mon, 11 September 2017 09:13 Go to previous messageGo to next message
Kr55  is currently offline Kr55
Messages: 11351
Registered: May 2002
Location: Edmonton

6 Cups

ziltoid wrote on Mon, 11 September 2017 08:11

Adam wrote on Sun, 10 September 2017 23:56

Wanted to weigh in on the below items, but thought I'd clutter the media thread if I didn't move this:

Kr55 wrote on Sun, 10 September 2017 22:50

GabbyDugan wrote on Sun, 10 September 2017 19:58

I commit the cardinal sin of trusting my own eyes, and I view advanced statistics as very interesting and probably useful in many ways, but I'm still not convinced advanced stats are predictive of everything all the time...

Chia's "black box" analytics would be fascinating to know more about, especially as they pertain to players like Russell (was it Matt Henderson at Oilers Nation that thought Russell was the worst thing to happen to the Oilers ever?) ...to my eye, Russell is okay for the next year or two, but long than that? I think another Benning or two will fall into the Oilers lap....the hockey gods don't despise the Oilers any more

I gave up my subscription to the Journal a few months ago.... Paula Simons and David Staples are about the only veteran writers they have left, and I'm not much of a fan of either one....



I definitely agree hockey analytics are not predictive of everything. I do still find it interesting though for looking at the past. And it's nice to be able to confirm guys you know are awesome (McDavid!) are in a few dozen different ways :)

I'd actually say analytics are predictive of very little, especially for that middle group of players that don't either suck, or are clearly elite. The black box stuff made by armies of guys watching videos are probably a lot closer though than the data the average joes can get a hold of.


I think what's in the black box has its roots in a lot of what we see as advanced stats. There has been a big evolution here over the last few years, but it's still in it's infancy, and it's a struggle to come up with metrics in hockey, where everyone is always in motion, as compared to baseball - where most of the game is static at any given point in time and it's more one-or-the-other outcomes.

In hockey, even faceoffs are challenging to evaluate cleanly - if one centre draws the puck back, but the opposing winger reads the play and jumps in and reaches the puck first, who wins? What about a situation where the center just wants to advance the puck a zone, possibly to achieve a change, or clear the defensive zone - if he pushes it forward and achieves his objective, but the other team gets possession of the puck, who won the draw? Who gets credit for a scrambled draw where the wingers or defencemen ultimately determine possession?

Possession stats and shooting metrics are even more challenging. Not all shots are created equal, and yet any of them can go in the net. What's the value of a blocked shot? What about a missed shot?

I personally think there's something to Valiquette's system looking at puck movement before a shot. I remember once being out at the rink with a friend who's a goalie. There wasn't anyone else around, so I was just taking shots. I had a bunch of pucks in the slot about 10-15 feet out from the goal. The first time around, I just teed them up and shot them directly from where they sat. He stopped almost every one. The next round, I simply did a quick pull of the puck from far out, to closer to my feet and then immediately snapped them. The effect was huge and his save percentage was much worse. Now, I'm not an elite shooter, and he's not an elite goalie (and I was probably closer to the former than he was to the latter) but I still think it's logically true that changing the angle the shot is coming from quickly before releasing is going to make a big difference. If you think about the best goal-scorers - Ovechkin, Stamkos, Hull before them - they score a large percentage of their goals off one-timers. It's harder to track a quick moving puck laterally that then immediately is going the other direction.

The challenge of course is that with a development age in analytics, there will be good analysis and bad analysis. I do think that Russell does some of what the Oilers value well - he penalty kills, blocks shots and dumps the puck out of the zone proficiently. I think there's a high value on puck possession though, and to use a football analogy, I think Russell (and several others of his ilk) basically hedge towards punting on the first down. There's no magic ranking formula that conclusively shows who is best, so if the Oilers' preferred supplier ranks some of Russell's attributes high, and doesn't spend a lot of time on completed passes, or advancing zone-to-zone with control, or surrendering the blueline to attackers...well, then he's going to rank higher with them than he does with others (including me).

I think the biggest challenge as fans is that NHL teams have been quick to snap up anyone doing any kind of groundbreaking work. There's been a thirst for knowledge on evaluation metrics, and teams have bought up a lot of good analysts. That has left fans with the early day metrics, without access to a lot of the actual advanced work. I expect that some teams are using quite sophisticated metrics now, while others languish in the past...


A lot of this comes down to what a person believes constitutes good performance. That is, what metrics accurately reflect "offense" and "defense" etc. To this end, there are statistical tools we can use to assign these metrics to abstract things like "offense" and "defense", as well as determine how big a role each metric plays in determining "offense" and "defense" etc. The question is whether these tools are being used, or if people are still relying on traditional knowledge, so-to-speak.



IMO the ultimate analytics would come from a combination of collecting metrics by using traditional knowledge. Not traditional knowledge guys like Matheson and Spector have, but with experienced minds that know what good and bad hockey plays look like. If you could develop a set of rules to follow to assign credit for good/bad plays and watch every game, every play and assign good/bad points for each player on every play and who was on the ice for each, you probably have the ultimate data set to derive the worth of each player (in the past tense still of course).

That still isn't enough to be certain of how that player would perform if you acquired him, dropped him in a new coaching system with new teammates that you have to hope he can mesh with, but it's probably as close to knowing as you can get. Bowman with Chicago has said they do something like this. They have their army of guys watching video, and they try to bring in guys with specific skill sets that fill a need in Quenneville's system. I doubt Chicago's analytics guys watch every single game and play of course. Probably have some high level stats to try to identify some targets, then they really dig in on the potential guys they could go after.


Hockey is a heck of a game to try to analyse. As much as I enjoy the analytics stuff, there is certainly lots that you can't figure out with past data. Is there any stat you can find that will tell you if you keep taking runs at Eberle in the playoffs that he's gonna bail out and give you the puck for free in his own zone? Not really, but players figure this out over time. Is there anything that will tell you if you keep tripping and hacking McDavid in a 7 game series you can keep him under ppg? The physicality factor in hockey is huge and almost impossible to track in this game where any player can be any place on the ice at any given time and physical contact is allowed, and even illegal contact can be completely ignored by refs.

[Updated on: Mon, 11 September 2017 09:25]


"The Edmonton Oilers are not where they should be right now and that is unacceptable. We need to get better immediately. That starts today"
-Kevin Lowe, April 2013


"Next year (15/16) I would forecast as another developmental year"
- #2, April 2015

Send a private message to this user  

 Re: Advanced Metrics [message #698837 is a reply to message #698836 ]
Mon, 11 September 2017 09:41 Go to previous messageGo to next message
RDOilerfan  is currently offline RDOilerfan
Messages: 2169
Registered: January 2016

2 Cups

While there is definitely a place for advanced stats and I imagine that as time goes on and there is more historical data to look at and how the stats are calculated will be refined, the stats will get better. However, I do think hockey is a game where there are some things you just can't measure with a stat that impact the game and how players play. Players, past and present talk about intangibles, they talk about how intimidation and toughness still play a role in the game, are important and help your team.
Every stat guy will say that statistically it is a myth. But if the players say it and believe it, unless they are all lying all the time, it's worth something.



Send a private message to this user  

 Re: Advanced Metrics [message #699490 is a reply to message #571967 ]
Tue, 03 October 2017 22:36 Go to previous message
Goose  is currently offline Goose
Messages: 1617
Registered: October 2006
Location: Vancouver

1 Cup

For anyone interested in anything beyond goals and assists, Corsica.hockey is back.

The Nation Network has actually partnered with Emmanuel Perry (@MannyElk) to provide a resource for all of the Nation writers and the public to access. Kudos to the Nation for funding what I think is a great resource and making it available to everyone (and of course to Manny for doing all the work).

There are a bunch of new features, charts and live game tools that will give you even more things to play around with.




Send a private message to this user  

Pages (10): [ «  <  1  2  3  4  5  6  7  8  9  10]  
Previous Topic:Who is the best player in the NHL?
Next Topic:How many points does Drai need to earn 8.5M?
Oilers NHL Minors Speculation For Sale 


Copyright © OilFans.com 1996-2017.
All content is property of OilFans.com and cannot be used without expressed, written consent from this site.
Questions, comments and suggestions can be directed to oilfans@OilFans.com
Privacy Statement


Hosted by LogicalHosting.ca