HFBoards

Go Back   HFBoards > General Hockey Discussion > By The Numbers
Mobile Hockey's Future Become a Sponsor Site Rules Support Forum vBookie Page 2
By The Numbers Hockey Analytics... the Final Frontier. Explore strange new worlds, to seek out new algorithms, to boldly go where no one has gone before.

State of analytics in major junior hockey?

Reply
 
Thread Tools
Old
09-01-2014, 10:47 PM
  #1
sfan
Registered User
 
Join Date: Jun 2013
Location: Ottawa
Country: Canada
Posts: 66
vCash: 500
State of analytics in major junior hockey?

Can anyone here shed insight on the state of analytics adoption in junior hockey? I am of course aware of Dubas' tenure at the Soo and the Ottawa 67s have re-signed young Matt Pfeffer as their advanced stat guy. What other CHL teams are investing? Any sense if the CHL will invest in official data and automated data acquisition anytime soon? Are other international T1 junior leagues taking the lead? Thx

sfan is offline   Reply With Quote
Old
09-03-2014, 07:57 PM
  #2
sfan
Registered User
 
Join Date: Jun 2013
Location: Ottawa
Country: Canada
Posts: 66
vCash: 500
Former Ottawa 67's AGM Misha Donskov has been hired manager of hockey operations, analytics and video for Hockey Canada.
http://www.tsn.ca/canadian_hockey/story/?id=460935

Saskatoon Blades hire Bruce Peter to provide analytics:
http://gdrinnan.blogspot.ca/2014/08/...l-cougars.html

Kelowna Rockets' president & GM “I’m a big believer in that stuff, it’s the way of the future.”
http://www.kelownadailycourier.ca/sp...a4bcf6878.html

Jim Hiller, former WHL coach of the year Tri-City Americans and Chilliwack Bruins, now tagged as the "Analytics Guy" in new role with Detroit Red Wings
the http://
http://www.wingingitinmotown.com/201...r-tony-granato
http://blogs.theprovince.com/2014/07...-coach-search/

Lethbridge Hurricanes coach, Drake Berehowsky “Numbers don’t lie to me”
http://www.theglobeandmail.com/sport...ticle16065376/


Last edited by sfan: 09-03-2014 at 08:29 PM.
sfan is offline   Reply With Quote
Old
09-07-2014, 07:57 AM
  #3
Kevin Forbes
Hockey's Future Staff
 
Join Date: Jul 2002
Location: Nova Scotia
Country: Canada
Posts: 9,184
vCash: 500
Matt Pfeffer handles it for the Ottawa 67s
http://www.thehockeynews.com/blog/author/mattpfeffer/
http://www.progressivehockey.com/

__________________
Twitter / Blog / About.me
Kevin Forbes is online now   Reply With Quote
Old
09-08-2014, 01:45 PM
  #4
sfan
Registered User
 
Join Date: Jun 2013
Location: Ottawa
Country: Canada
Posts: 66
vCash: 500
"Donskov, Hockey Canada committed to analytics"
http://www.sportsnet.ca/hockey/nhl/d...-to-analytics/

sfan is offline   Reply With Quote
Old
09-08-2014, 01:50 PM
  #5
sfan
Registered User
 
Join Date: Jun 2013
Location: Ottawa
Country: Canada
Posts: 66
vCash: 500
Anyone have insight into why the QMJL officially publishes "dangerous shot" and face-off win/loss data while the OHL and WHL do not?

sfan is offline   Reply With Quote
Old
09-08-2014, 07:00 PM
  #6
Kevin Forbes
Hockey's Future Staff
 
Join Date: Jul 2002
Location: Nova Scotia
Country: Canada
Posts: 9,184
vCash: 500
Quote:
Originally Posted by sfan View Post
Anyone have insight into why the QMJL officially publishes "dangerous shot" and face-off win/loss data while the OHL and WHL do not?
Some of it has to do with just long-standing tradition and some of it has to do with the different web properties.

Years ago (maybe five), I believe the WHL and the OHL had their websites designed and managed by NewSport Media/LeagueStat (I believe they currently run the AHL and the ECHL sites). This also provided the interface for league statisticians to enter information and standardized the information collected between the two leagues.

At the same time, the QMJHL's website was proprietary and they were undergoing an ambitious project to digitize their entire history of results and records into that same structure that ran their main site.

Then the CHL put all three leagues under the same technological umbrella, run by a company called MRX. The launch of this new framework and design was pretty horrible, with the MRX sites not being ready for the beginning of season, especially the earlier start of the QMJHL. They eventually worked the kinks out, however, but it was pretty brutal for the first month or so. Plus, the Q had already put in a huge amount of work into their stats project and were forced to use the MRX system.

So now the Q has their big historical database (which was formerly their game-by-game/season stats) located here: http://www.lhjmq-records.qc.ca/ and use MRX on their main site. They must transfer information to the historical database on a regular basis, as it is pretty up to date as well.

The Q's stats are headed by Denis Demers, who's been doing this for 20 years. You can follow him on Twitter here: https://twitter.com/lhjmq_hdd8

I don't know if that really answers your question, but basically the answer I got when I once asked why the Q seems to have stats figured out a lot more than the other leagues was that Demers and his team don't really have equals in the other leagues.

Kevin Forbes is online now   Reply With Quote
Old
09-08-2014, 07:33 PM
  #7
sfan
Registered User
 
Join Date: Jun 2013
Location: Ottawa
Country: Canada
Posts: 66
vCash: 500
Thanks Kevin, that is interesting. It certainly is easy to appreciate how complex this must have been, both technically and politically.

I am coming at this relatively new to both major junior hockey and to hockey analytics. This past weekend's exhibition game between Gatineau Olympiques and the Ottawa 67s introduced me to the DS and FO stats tracked in the Q.

As we are now seeing in the NHL, it would be great to see MJ hockey expand official league stats to support advanced analytics. As more and more teams adopt analytics (publicly or confidentially) there is both redundant and sub-optimal data being generated. The insights and results are thus correspondingly sub-optimal.

Maybe the Q and Denis Demer are the most able to take the lead on such an initiative. My guess is that, among other things, a growing number of NHL teams and scouts will be very supportive.

Quote:
Originally Posted by Kevin Forbes View Post
Some of it has to do with just long-standing tradition and some of it has to do with the different web properties.

Years ago (maybe five), I believe the WHL and the OHL had their websites designed and managed by NewSport Media/LeagueStat (I believe they currently run the AHL and the ECHL sites). This also provided the interface for league statisticians to enter information and standardized the information collected between the two leagues.

At the same time, the QMJHL's website was proprietary and they were undergoing an ambitious project to digitize their entire history of results and records into that same structure that ran their main site.

Then the CHL put all three leagues under the same technological umbrella, run by a company called MRX. The launch of this new framework and design was pretty horrible, with the MRX sites not being ready for the beginning of season, especially the earlier start of the QMJHL. They eventually worked the kinks out, however, but it was pretty brutal for the first month or so. Plus, the Q had already put in a huge amount of work into their stats project and were forced to use the MRX system.

So now the Q has their big historical database (which was formerly their game-by-game/season stats) located here: http://www.lhjmq-records.qc.ca/ and use MRX on their main site. They must transfer information to the historical database on a regular basis, as it is pretty up to date as well.

The Q's stats are headed by Denis Demers, who's been doing this for 20 years. You can follow him on Twitter here: https://twitter.com/lhjmq_hdd8

I don't know if that really answers your question, but basically the answer I got when I once asked why the Q seems to have stats figured out a lot more than the other leagues was that Demers and his team don't really have equals in the other leagues.

sfan is offline   Reply With Quote
Old
09-08-2014, 10:49 PM
  #8
Kevin Forbes
Hockey's Future Staff
 
Join Date: Jul 2002
Location: Nova Scotia
Country: Canada
Posts: 9,184
vCash: 500
Quote:
Originally Posted by sfan View Post
As we are now seeing in the NHL, it would be great to see MJ hockey expand official league stats to support advanced analytics. As more and more teams adopt analytics (publicly or confidentially) there is both redundant and sub-optimal data being generated. The insights and results are thus correspondingly sub-optimal.
I'm not sure I 100% agree with this.

Sub-optimal data is just a question of collection methods and the resources used to put it all together. Keep in mind that teams might have access to a deeper level of information than is publicly available, if only because they have their own statisticians travelling with the team and the ability to review raw video. That doesn't mean they're right or using the resources properly, just more to say that at the NHL level, teams aren't likely to just be scraping the play-by-play feed off NHL.com, they're deriving their own analytics from their own observations and the information they have on hand. Things like Corsi and Fenwick are still just predictors for puck possession, which will be thrown out the window whenever spatio temporal data comes into play. Teams might rate things like scoring chances or shot quality differently based on internal criteria, etc.

But a big part of this whole conversation is competitive advantage.
Sure, it's great for fans to better understand the game (and maybe do their own research, etc) if advanced stats are readily available in a public forum, but there's no added benefit for the teams that are working deeply in these areas.

If there are teams assigning funds and effort to work in the advanced stats area, be it analysts, statisticians, technology resources, etc, (we certainly know there are at the NHL level and we can assume there are at least some at the junior level), they clearly would want to stay ahead of that curve or, at the very least, make their opponents work just as hard to gain the same insights.

I think that's same reason why teams are shutting down websites after they hire the guys behind them. Why make it easy for your opponents to even take a half-step into this arena, regardless of whether you're a step ahead of that or five steps ahead.

It's hard to say whether someone like M. Demers has the staff (or the desire) to get into this space too much. Dangerous Shots is an attempt at shot quality/scoring chances. I suppose the next step might be something with QoC/QoT or even PDO.

Kevin Forbes is online now   Reply With Quote
Old
09-09-2014, 09:43 AM
  #9
sfan
Registered User
 
Join Date: Jun 2013
Location: Ottawa
Country: Canada
Posts: 66
vCash: 500
Thanks for your thoughtful reply. I agree with much of what you say. While analytics seems to be going mainstream in the NHL, is no obvious evidence that it has penetrated more than early adopters in MJ hockey so far. I agree there certainly is a current opportunity for MJ teams that invest in this area to gain some competitive advantage, at least for now. I also agree that publicly available advanced stats plays a role in deepening fan understanding of the game, building interest, and even possibly growing the base.

I am an unequivocal supporter of hockey analytics, but I am not convinced that it will always be a sustainable competitive advantage for any given team. Once mainstream, and I see this as inevitable, it simply becomes the new normal, table-stakes to running all but the most negligent hockey ops.

Like every other aspect of drafting, team formation, player development, and game strategy, analytics has an important role to play but ultimately, judgment drives decision making. Whether it’s the decisions of GMs, coaches, or most crucially, the conscious and instinctive decisions players make on the ice. This is where the real magic happens that makes the game electric.

If we can agree that analytics is inevitable in pro & MJ hockey, that data collection is resource intensive and error prone, and recognize that not all teams have comparable budgets, might it not be better for the business of MJ hockey, at the league level, to play a role in leveling the playing ice by taking on advanced stat collection?

MJ hockey has important differences in its business model compared to the NHL and this can play a role in the business case for analytics both at the team and league level. MJ teams seem to increasingly need to compete for top CDN draftees, and top US talent, that have choices about where, and under what terms, they are prepared to play. Yet the productive life of a top roster player is at most four years, often much less. However a strong track record in producing future NHL players can be a sustainable way of deferentially attracting both talent and market dollars. Finally, all other things constant, teams & leagues that lead in the adoption of analytics, it would seem, will be able to produce a better product for fans and sponsors and will be in a better position to see their players differentially scouted and confidently ranked by the NHL.

Quote:
Originally Posted by Kevin Forbes View Post
I'm not sure I 100% agree with this.

Sub-optimal data is just a question of collection methods and the resources used to put it all together. Keep in mind that teams might have access to a deeper level of information than is publicly available, if only because they have their own statisticians travelling with the team and the ability to review raw video. That doesn't mean they're right or using the resources properly, just more to say that at the NHL level, teams aren't likely to just be scraping the play-by-play feed off NHL.com, they're deriving their own analytics from their own observations and the information they have on hand. Things like Corsi and Fenwick are still just predictors for puck possession, which will be thrown out the window whenever spatio temporal data comes into play. Teams might rate things like scoring chances or shot quality differently based on internal criteria, etc.

But a big part of this whole conversation is competitive advantage.
Sure, it's great for fans to better understand the game (and maybe do their own research, etc) if advanced stats are readily available in a public forum, but there's no added benefit for the teams that are working deeply in these areas.

If there are teams assigning funds and effort to work in the advanced stats area, be it analysts, statisticians, technology resources, etc, (we certainly know there are at the NHL level and we can assume there are at least some at the junior level), they clearly would want to stay ahead of that curve or, at the very least, make their opponents work just as hard to gain the same insights.

I think that's same reason why teams are shutting down websites after they hire the guys behind them. Why make it easy for your opponents to even take a half-step into this arena, regardless of whether you're a step ahead of that or five steps ahead.

It's hard to say whether someone like M. Demers has the staff (or the desire) to get into this space too much. Dangerous Shots is an attempt at shot quality/scoring chances. I suppose the next step might be something with QoC/QoT or even PDO.


Last edited by sfan: 09-09-2014 at 09:55 AM.
sfan is offline   Reply With Quote
Old
09-09-2014, 10:30 AM
  #10
Kevin Forbes
Hockey's Future Staff
 
Join Date: Jul 2002
Location: Nova Scotia
Country: Canada
Posts: 9,184
vCash: 500
Great points!

Some quick things to consider as well:

Junior hockey has limited player movement, compared to pro hockey. There are only limited windows during the year when most players can be traded.

In addition, because of age restrictions, a junior hockey career is limited to a maximum of 4 or 5 years.

This period of a player's career is arguably one of the most crucial development phases of their entire hockey experience, from both physical and skill development.

In addition to winning games, organizations are tasked with developing players, as players who are drafted into the NHL turn into development fees back to the junior teams.

So when we're talking about analytics in junior hockey, I'm wondering if there might not be just as strong of a need for the organization to have ways to monitor (and ultimately improve) a player development plan as there would be for coaching staff to have access to information about in-game information.

If a team can improve development efforts, they might be able to create more NHL draft picks (and win more hockey games), thus increasing revenue (and more importantly exposure), which helps with recruiting.

Recruiting is tempered somewhat with the draft, but that's always a bit of a sideshow when it comes to junior hockey with players trying to pick where they want to go, etc. But then you think of Import players, etc.

There's definitely incentive on the team side to find ways to consistently improve the way they assess, develop and produce hockey players.

Kevin Forbes is online now   Reply With Quote
Old
09-09-2014, 11:51 AM
  #11
Doctor No
Retired
 
Doctor No's Avatar
 
Join Date: Sep 2005
Posts: 23,446
vCash: 500
It's fantastic that the Q has really taken ownership of their league history - their historical archives are tremendous.

And the support staff are very responsive. I'm not even sure why they have support staff, but I made some corrections to their records (most notably, the first few posts of this thread: http://hfboards.hockeysfuture.com/sh....php?t=1465761) and they seemed genuinely interested and invested.

Doctor No is offline   Reply With Quote
Old
09-09-2014, 12:08 PM
  #12
Canadiens1958
Registered User
 
Canadiens1958's Avatar
 
Join Date: Nov 2007
Posts: 11,056
vCash: 500
Further

Quote:
Originally Posted by sfan View Post
Anyone have insight into why the QMJL officially publishes "dangerous shot" and face-off win/loss data while the OHL and WHL do not?
QMJHL arose out of the merger of the MMJHL and the QJHL in 1969. NHL head offices were in Montréal at that time so the Q simply carriesd on the tradition that the old leagues had, especially the MMJHL which was more or less a Canadiens developmental league.

John Horman was the initial mover in the stats area MMJHL/QMJHL. QMJHL trophy named after him:

Bio:
http://www.rds.ca/1.261865

Trophy:
http://fr.wikipedia.org/wiki/Troph%C3%A9e_John_Horman


Last edited by Canadiens1958: 09-09-2014 at 02:30 PM. Reason: added data and link
Canadiens1958 is offline   Reply With Quote
Old
09-10-2014, 08:46 AM
  #13
sfan
Registered User
 
Join Date: Jun 2013
Location: Ottawa
Country: Canada
Posts: 66
vCash: 500
Thanks Kevin, I agree with you on all these points below.

Do you feel that advanced data collection should remain a private team choice & cost? Or is there value in the league doing this officially and publicly?

Any thoughts on when automated, real-time player & puck data acquisition may be implemented?

Quote:
Originally Posted by Kevin Forbes View Post
Great points!

Some quick things to consider as well:

Junior hockey has limited player movement, compared to pro hockey. There are only limited windows during the year when most players can be traded.

In addition, because of age restrictions, a junior hockey career is limited to a maximum of 4 or 5 years.

This period of a player's career is arguably one of the most crucial development phases of their entire hockey experience, from both physical and skill development.

In addition to winning games, organizations are tasked with developing players, as players who are drafted into the NHL turn into development fees back to the junior teams.

So when we're talking about analytics in junior hockey, I'm wondering if there might not be just as strong of a need for the organization to have ways to monitor (and ultimately improve) a player development plan as there would be for coaching staff to have access to information about in-game information.

If a team can improve development efforts, they might be able to create more NHL draft picks (and win more hockey games), thus increasing revenue (and more importantly exposure), which helps with recruiting.

Recruiting is tempered somewhat with the draft, but that's always a bit of a sideshow when it comes to junior hockey with players trying to pick where they want to go, etc. But then you think of Import players, etc.

There's definitely incentive on the team side to find ways to consistently improve the way they assess, develop and produce hockey players.

sfan is offline   Reply With Quote
Old
09-10-2014, 08:56 AM
  #14
Kevin Forbes
Hockey's Future Staff
 
Join Date: Jul 2002
Location: Nova Scotia
Country: Canada
Posts: 9,184
vCash: 500
Quote:
Originally Posted by sfan View Post
Thanks Kevin, I agree with you on all these points below.

Do you feel that advanced data collection should remain a private team choice & cost? Or is there value in the league doing this officially and publicly?

Any thoughts on when automated, real-time player & puck data acquisition may be implemented?
There is value in the league doing it officially and publicly, I just think it is rare that you would see a league take on the expense (manpower, technology, etc) to do something of that nature. That leads into M. Demers and his work with the QMJHL, where we see a league that has taken on similar efforts in the past. If a league was going to somehow take a stab at advanced stats as a league initiative, I think the QMJHL could be the one to do it.

There's obvious benefits to individual teams doing it privately and not sharing information with their competitors.

Real-time player/puck data is being bandied about quite a bit (Mirtle had an article this week mentioning a possible wearable device in NHL sweaters), but the SportVu system as it stands now would need some modifications to work for hockey due to things like more frequent shift-changes, more players involved in the game and a smaller puck moving at faster speeds than a basketball.
We still don't have a reliable way to tell if a puck is in the net and yet we want to track its exact position 10 times a second or whatever the refresh rate is?

Kevin Forbes is online now   Reply With Quote
Old
09-10-2014, 09:16 AM
  #15
sfan
Registered User
 
Join Date: Jun 2013
Location: Ottawa
Country: Canada
Posts: 66
vCash: 500
Crowdsourcing advanced data

Until automated real-time puck and player data is a reality, there is possibly a "third way" alternative to either leaving advanced data capture to entirely individual teams (including private analysts) or instituting official, centralized league collection of advanced data.

This third way is a cooperative crowd-sourcing. This already kind-of informally/partially exists for the NHL, where all games are televised and a richer official stat feed can be scraped. A more compete crowd-sourcing for MJ hockey would have a single data repository of record and tools & process for helping productivity ensuring quality. The heavy lifting could be done by both motivated volunteers as well as paid staff at teams and the league.

The objective is to improve the variety and quality of data available and to reduce the redundant and total costs related to acquiring data. The underlying premise is that the end goal is not private data, rather it is to maximize the effort and resources on what can be learned from the data and applying these insights into improved improved player development league-wide and thus a better product at every game.

sfan is offline   Reply With Quote
Old
09-10-2014, 09:23 AM
  #16
sfan
Registered User
 
Join Date: Jun 2013
Location: Ottawa
Country: Canada
Posts: 66
vCash: 500
It would be very interesting to know M. Demers' thoughts on the matter or how best to explore it with him.

Quote:
Originally Posted by Kevin Forbes View Post
... That leads into M. Demers and his work with the QMJHL, where we see a league that has taken on similar efforts in the past. If a league was going to somehow take a stab at advanced stats as a league initiative, I think the QMJHL could be the one to do it...

sfan is offline   Reply With Quote
Old
09-24-2014, 06:45 AM
  #17
sfan
Registered User
 
Join Date: Jun 2013
Location: Ottawa
Country: Canada
Posts: 66
vCash: 500
Pointstreak is rolling out an advanced video analytics platform:
http://performance.pointstreak.com/n...ory.html?id=86

sfan is offline   Reply With Quote
Old
09-24-2014, 08:18 AM
  #18
Kevin Forbes
Hockey's Future Staff
 
Join Date: Jul 2002
Location: Nova Scotia
Country: Canada
Posts: 9,184
vCash: 500
Quote:
Originally Posted by sfan View Post
Pointstreak is rolling out an advanced video analytics platform:
http://performance.pointstreak.com/n...ory.html?id=86
Sounds very similar to the PUCKS system (http://www.tvti.net/products.html) which is used by almost half the NHL.

...

Also the Saint John Sea Dogs have announced that they are putting together an analytics program: http://www.saintjohnseadogs.com/arti...lytics-program

Based primarily on Paul Boutilier's Defense Development Program and supported by analytics work from Steven Cocker.

Kevin Forbes is online now   Reply With Quote
Old
09-24-2014, 08:39 AM
  #19
sfan
Registered User
 
Join Date: Jun 2013
Location: Ottawa
Country: Canada
Posts: 66
vCash: 500
Yes, and here's a 2012 video of how the Calgary Flames are using PUCKS for scouting junior draft candidates: http://video.flames.nhl.com/videocen...sole?id=181189

Good to see Saint John announce their analytics program! Maybe one more reason the Q is well suited to take the lead on league compiled advanced stats.

Quote:
Originally Posted by Kevin Forbes View Post
Sounds very similar to the PUCKS system (http://www.tvti.net/products.html) which is used by almost half the NHL.
...
Also the Saint John Sea Dogs have announced that they are putting together an analytics program: http://www.saintjohnseadogs.com/arti...lytics-program

Based primarily on Paul Boutilier's Defense Development Program and supported by analytics work from Steven Cocker.

sfan is offline   Reply With Quote
Old
09-26-2014, 06:49 AM
  #20
Mathletic
Registered User
 
Mathletic's Avatar
 
Join Date: Feb 2002
Location: St-Augustin, Québec
Country: Canada
Posts: 11,864
vCash: 500
Analytics in college hockey

http://www.collegehockeynews.com/new...lytics_era.php

Mathletic is offline   Reply With Quote
Reply

Forum Jump


Bookmarks

Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off



All times are GMT -5. The time now is 08:48 AM.

monitoring_string = "e4251c93e2ba248d29da988d93bf5144"
Contact Us - HFBoards - Archive - Privacy Statement - Terms of Use - Advertise - Top - AdChoices

vBulletin Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
HFBoards.com is a property of CraveOnline Media, LLC, an Evolve Media, LLC company. ©2014 All Rights Reserved.