Hockey Analytics and the Edmonton Oilers. Does it even exist?

[Panda Bear]

Just dropping in to say that Corsi and Fenwick shouldn't be considered advanced stats.

 

[Tippetts Mustache]

Good scouting is much more important than good analytics. Analytics can be a good complement to scouting but it should never vice versa, i.e. that analytics takes precedent over scouting / is weighted more heavily. That's how you end up with analytics darlings that are actually not very good hockey players. I think this organization's issue has been the scouting, not that they're behind in analytics.

The Edmonton Oilers employ one person for analytics, Justin Mahe. Mahe deals with customs issues too, so he can't even focus 100% on analytics. Chiarelli turfed basically the whole department and Holland has shown no interest in building it back up.

 

[Forgot About Drai]

Its malpractice the way the Oil have approached analytics.

Whether you are fully in on analytics or not, having 1 (who doesnt even do analytics) in the department is so dumb and backwards I dont know how you can defend it.

Every high level org in pro sports, not just hockey, have integrated advanced analytics into their decision making. Oilers are in the stone age and should be embarrassed.

 

[Little Fury]

Here's an old (2019) list of analytics staff across the league. It's a little out of date because I know Tampa has a full time staff of three.

 

[oobga]

Here's an old (2019) list of analytics staff across teh league.

Justin Mahe?? That sounds like a made up name to me!

 

[Little Fury]

After Tyler Dellow its amazing the Oilers would have anybody doing analytics. He's the exact kind of "expert" that causes the same to be subsequently avoided. This guy was a walking migraine.

Also, Tyler seems most tied to the acquisition of Sekera, a subsequent buyout. The list of hits from this guy is slim to none. But hey, he watched Moneyball.

Dellow had very little input on player acquisition when he was here.

 

[McDNicks17]

Big analytics departments are kind of pointless. It doesn't really matter how much you spend on it, you'll likely never come close to replicating what the big third party companies can do and you also probably won't find anything useful to track that they aren't already tracking.

I'm pretty sure every team in the league uses SportsLogIQ now.

Most analytics jobs are just relaying their data to coaches/GMs. If Tipp wants to know how many cross-ice passes or shots from the slot the team is allowing, he asks Justin or whoever it is. If Holland is looking for a guy that move pucks while under forecheck pressure, he asks Justin.

How effective a team is with analytics doesn't really have anything to do with the size of department they have. It really just comes down to if the decision-makers are looking to use it.

 

[McDNicks17]

According to Pierre McGuire, the Oilers are one of the two only teams in the league that are built through analytics haha.

 

[Behind Enemy Lines]

I'm challenged by the chaos reality of hockey vs. baseball and feel the hobbiest nature of many hockey analysts versus hard core math trained quants often leads to wrong or distorted conclusions of data but there is really no reason all professional sports team, most especially smaller market "Oakland A's" types shouldn't be pursuing the best hockey brains and trained math/quant personnel to pursue competitive advantage. Weeding through an arrogance phase (on both sides) is a vital step.

For curiousity, I took a peek at new kid Vegas and some of their personnel in quickly building up a Cup contender team. McPhee is an established veteran GM but I think his biggest coup was hiring two elite hockey personnel guys I argue were the best in the CHL, Kelly McCrimmon and Kelly Kisio who have unparalleled scouting eyes and brains, imo. Building out a contemporary Hockey Operations team from there has included some really solid and eclectic personnel which I'm including some article links for context:

Misha Donskov: Misha Donskov already vital to expansion Las Vegas
Donskov’s responsibilities will be similar in Vegas. He had been prepping to leave for a Team Canada coaching staff retreat when McPhee called, formally asking him to come aboard, and leapt at the “once-in-a-lifetime-opportunity to start at ground zero.” Since then, he’s designed their scouting software, vetted outside analytics firms, and organized their video platform, logistical matters required before the coaching staff gets hired and the team gets on the ice next year. No wonder McPhee calls his hiring “automatic.”

Vaghan Karpan: Sep 2019: Karpan's first big step came in Brandon

Tom Poraszka: Moneyball Meets Hockey In Key Vegas Hire

EDIT: Link to the incredible Western Canadian hockey talent among the Vegas hockey operations team: Vegas franchise has WHL off-ice flavour . . . WHLers brought golden touch to Golden Knights . . . Next up: Stanley Cup final

A quote by Vaghan Karpan is a pretty good summary of how all business should work but most especially imo, in a closed system like the NHL cartel. This is a top-down ownership belief system which is executed by smart, deeply experienced and versatile hockey executives.

During his scouting career, the game has evolved tremendously to more of a speed and skill-based game. He’s constantly had to adjust the way he looks at the game, something he said is helped by Golden Knights owner Bill Foley, president of hockey operations George McPhee and McCrimmon, who officially became general manager on Sunday.
"You have to be a progressive thinker," Karpan said. "What worked and what worked yesterday isn’t going to work tomorrow. You always have to challenge yourself to be better. That’s one of the unique things about this group of people we have in Vegas, beginning with Mr. Foley and George and Kelly, is that we don’t talk about the good old days. We don’t have a lot of good old days. We’re always looking to get better and incorporating new ideas."
He said Vegas has enthusiastically embraced the metrics movement, which is advanced statistical study of the game and its players. That also suits Karpan’s willingness to evolve.
"We continue to reinvent ourselves," Karpan said. "You have to do that. The world isn’t static, never mind hockey, so if you’re not willing to adapt and adjust, you’re going to get left behind."

Now, this is just a high level scan of one progressive, highly successful new NHL model organization. But it is fair to ask and question if the Oilers is doing enough to build out its hockey operations function with the type of high level talent that Vegas believes as a core principle to fostering a winning organization. It is an immense strategic area not restricted by salary cap and one that can contribute in all facets to enhancing business decisions in key areas including acquisition -draft and development; trades and free agency; game planning and execution; etc etc.

The Oilers have a long way to go on the ice to compete among the NHL elites. I think there's substantive risk off-ice as well as complex, fluid decision making is being enhanced through new approaches like the Vegas model which is build upon elite hockey minds and supplemented by analytical tools and processes as part of its process.

 

[McDNicks17]

I'm challenged by the chaos reality of hockey vs. baseball and feel the hobbiest nature of many hockey analysts versus hard core math trained quants often leads to wrong or distorted conclusions of data but there is really no reason all professional sports team, most especially smaller market "Oakland A's" types shouldn't be pursuing the best hockey brains and trained math/quant personnel to pursue competitive advantage. Weeding through an arrogance phase (on both sides) is a vital step.

The chaos argument definitely works for blogger analytics, but actual NHL analytics are basically there to quantify the chaos.

If Holland asks his analytics guy if Caleb Jones is a good puck-mover, he isn't going to get something like "yeah, he has a 3.6 offensive GWAR" back. It's going to be a report with stuff like, "he attempted 12.4 exits/60, was successful on 74% of them, he attempted 68% of his pairing's exits, he saw a 16% drop in success rate when under pressure, etc.".

The Oilers could definitely use more guys on the analytics side, but I think the worry by some is overblown. Pretty much any layman in the organization can understand this stuff, so only having one "analytics guy" doesn't mean he's the only one working with them.

 

[McTonyBrar]

Analytics isn't why these teams are successful. They're successful because they use analytics as a tool ALONG with watching the players play the sport. How will looking at spreadsheets of numbers tell you how skilled the player is? Was Pouliot as good as his analytics? Fayne?

 

[Bryanbryoil]

An important "analytic" that hasn't been mentioned here. "Would play for Edmonton/60". Unless you are using it specifically for drafted players, the rest is well and good at potentially finding players to target, but if they have NMC/NTC's or are UFA's that doesn't exactly help.

 

[Behind Enemy Lines]

The chaos argument definitely works for blogger analytics, but actual NHL analytics are basically there to quantify the chaos.

If Holland asks his analytics guy if Caleb Jones is a good puck-mover, he isn't going to get something like "yeah, he has a 3.6 offensive GWAR" back. It's going to be a report with stuff like, "he attempted 12.4 exits/60, was successful on 74% of them, he attempted 68% of his pairing's exits, he saw a 16% drop in success rate when under pressure, etc.".

The Oilers could definitely use more guys on the analytics side, but I think the worry by some is overblown. Pretty much any layman in the organization can understand this stuff, so only having one "analytics guy" doesn't mean he's the only one working with them.

Yes, I agree. Regarding the Vegas hockey ops model, I highlighted among Donskov's responsibilities it includes "vetting outside analytics firms" which caught my eye that big data is obviously being crunched by experts and marketed as a service to NHL teams. @Fourier mentioned HockeyTech in Kitchener Waterloo which sounds like an interesting high growth big data firm that's actually located on the U. Waterloo campus and also trial and testing hardware applications like MLB Trackman. Talk of wearable technology and in puck data collection seems future bets as well. This sophistication both in talent (real quants vs. first wave hobbyists) and tools to analyze big data is fascinating where it can go. Early HockeyTech article here: Hockey analytics firm coming to Waterloo

Regarding the Vegas Hockey Ops model, I was just curious to skim how they are structured, the people and experience, and to some degree the belief system defining what has been an out of the box success. What I see (superficially) are deeply experienced hockey people with keen intelligence, growth mindset to pursue innovation and unafraid of change not limited by status quo thinking in a pretty traditional, low risk industry. This belief begins at the top with a bit of a maverick owner - it is a core cultural belief that reverberates throughout the organization. But of course I'll also 'out' my personal bias for championing the Western Canadian roots of many of them (didn't mention Murray Craven too). Another powerful tool in the toolkit.

I'm curious about the cultural change aspect and where teams are in the spectrum of this significant disruption to how things were always done and the embrace of new tools and approaches to guide decision making.

 

[Paralyzer008]

Analytics isn't why these teams are successful. They're successful because they use analytics as a tool ALONG with watching the players play the sport. How will looking at spreadsheets of numbers tell you how skilled the player is? Was Pouliot as good as his analytics? Fayne?

Pouliot was a fine player here, injuries stopped him from back-to-back 20 goal years and even when he couldn't produce anymore, he was a valuable PKer.

I have seen you post the same reply four times and all you keep saying is Mark Fayne as your example.

Many analytic writers at the time outside of Edmonton understood what the Oil were thinking, but also pointed out that Andy Greene was propping up his numbers. Andy Greene, of course, is still playing in the NHL today.

You're never going to bat 100%, but to disregard analytics because of Mark Fayne is odd. Bit of a cherry pick.

You could go the other direction and say Florida is an example of why analytics are the only thing you could use because Carter Verhaeghe profiled strong in his analytics in TBL and look at him now.

I totally agree that you should be looking at the analytics, then watching the player to see how he reaches those stats. Is he a driver? Is he complimentary? Is someone else the reason why he looks good? Is someone holding him back? Analytics can provide a context for what we may not notice on the ice or see on the ice and can't explain. I don't think anyone's eyes can understand the entire game and every player at once. A good organization understands context and has a team to help line everything up.

The Oilers have one guy doing "a report" - it's hard for me to believe in their process right now.

 

[Behind Enemy Lines]

These are the 'types' of sophisticated hires I anticipate are coming. Seattle is interesting to watch as they chart their course:



NHL Seattle Hires Award Winning Quantitative Analyst

NHL Seattle analytics specialist eager to help build expansion team

Gary Roberts sports performance and others: Seattle Kraken Announce Seven New Hockey Operations Hires

EDIT: This will be a direct competitor to Edmonton as is Vegas. Where and do the Oilers match up to this organizational mindset and talent? (Rhetorical question ... kinda)

 

[McDNicks17]

Yes, I agree. Regarding the Vegas hockey ops model, I highlighted among Donskov's responsibilities it includes "vetting outside analytics firms" which caught my eye that big data is obviously being crunched by experts and marketed as a service to NHL teams. @Fourier mentioned HockeyTech in Kitchener Waterloo which sounds like an interesting high growth big data firm that's actually located on the U. Waterloo campus and also trial and testing hardware applications like MLB Trackman. Talk of wearable technology and in puck data collection seems future bets as well. This sophistication both in talent (real quants vs. first wave hobbyists) and tools to analyze big data is fascinating where it can go. Early HockeyTech article here: Hockey analytics firm coming to Waterloo

Regarding the Vegas Hockey Ops model, I was just curious to skim how they are structured, the people and experience, and to some degree the belief system defining what has been an out of the box success. What I see (superficially) are deeply experienced hockey people with keen intelligence, growth mindset to pursue innovation and unafraid of change not limited by status quo thinking in a pretty traditional, low risk industry. This belief begins at the top with a bit of a maverick owner - it is a core cultural belief that reverberates throughout the organization. But of course I'll also 'out' my personal bias for championing the Western Canadian roots of many of them (didn't mention Murray Craven too). Another powerful tool in the toolkit.

I'm curious about the cultural change aspect and where teams are in the spectrum of this significant disruption to how things were always done and the embrace of new tools and approaches to guide decision making.

Yup. Third party firms do the heavy lifting now. SportsLogIQ is the big one. They do some really cool stuff.

And I 100% agree with the second part. You need guys like that at the top.

 

[MessierII]

Pouliot was a fine player here, injuries stopped him from back-to-back 20 goal years and even when he couldn't produce anymore, he was a valuable PKer.

I have seen you post the same reply four times and all you keep saying is Mark Fayne as your example.

Many analytic writers at the time outside of Edmonton understood what the Oil were thinking, but also pointed out that Andy Greene was propping up his numbers. Andy Greene, of course, is still playing in the NHL today.

You're never going to bat 100%, but to disregard analytics because of Mark Fayne is odd. Bit of a cherry pick.

You could go the other direction and say Florida is an example of why analytics are the only thing you could use because Carter Verhaeghe profiled strong in his analytics in TBL and look at him now.

I totally agree that you should be looking at the analytics, then watching the player to see how he reaches those stats. Is he a driver? Is he complimentary? Is someone else the reason why he looks good? Is someone holding him back? Analytics can provide a context for what we may not notice on the ice or see on the ice and can't explain. I don't think anyone's eyes can understand the entire game and every player at once. A good organization understands context and has a team to help line everything up.

The Oilers have one guy doing "a report" - it's hard for me to believe in their process right now.

Pouliot had a couple of good years then fell off a cliff. Surprised the “predictive value” of analytics didn’t pick that up.

 

[Oilhawks]

Further details on Oilers and analytics from Holland care of Daniel Nugent-Bowman of The Athletic

On how the Oilers use analytics to make decisions: “We don’t make any decisions without them. Justin Mahe and his brother Shaun Mahe do analytics for us. From a coaching perspective, there’s an analytical report that gets emailed to Dave Tippett in advance of every game that we play. There’s a postgame analytic report from a coaching perspective.

“Last year, (director of amateur scouting) Tyler Wright and (chief amateur scout) Bob Green spent a lot of time with Justin Mahe doing analytics on the draft, leading into the draft. (Holland said that’ll happen before the 2021 draft, too.)

“We’re going to our pro meetings. I’ve asked Justin Mahe to have an analytical report on the players that we’re talking about from an unrestricted (free agent) perspective.

“It’s a tool. It’s part of the process. It’s not the chief decision-maker by any stretch of the imagination. Our sport, it’s not a stationary sport. It’s a fluid sport where there’s a lot of people moving in a whole bunch of different directions. There’s a whole bunch of different ways to analyze.

“In talking with Dave Tippett, he’s got his own things that he analyzes — who’s on the ice for chances for, chances against. He’s been doing that for 20 years as a coach.”

“Is analytics a part of our decision-making process? Yes. In every decision that we make from an amateur standpoint, from a trade standpoint, from a pro-evaluation UFA standpoint, from a coaching standpoint — there’s an analytical dive, and there’s an analytical report. That is a piece of the puzzle.”

His overarching view on analytics: “All you’ve got to do is look at the goal differential at five-on-five for the real legit teams. That’s the analytics that I look at. Everybody in this league looks at analytics. I looked at analytics 25 years ago when I became a general manager. … It was rudimentary analytics, but I’ve always run a team on analytics. It’s more involved analytics now. But it’s a part of the process.”

On whether he’d want to grow the analytics team in Edmonton: “You can hire 15 people, but at the end of the day, you’re going to present an analytical report. Players have good years and bad years. It’s a piece of the decision-making process whether you’ve got 12 people hired in your analytics department or you’ve got two people hired.”

 

[Perfect_Drug]

If they have the budget to develop software, I'd take it a different direction.

I'd create VR/AR app of thousands of in-game scenarios to get them to fix reaction time, and get the players better at in-game decision making.

 

[The Safe Play]

"All you’ve got to do is look at the goal differential at five-on-five for the real legit teams. That’s the analytics that I look at."

Well he did a shitty job at putting together the bottom six then.

 

[Oilhawks]

"All you’ve got to do is look at the goal differential at five-on-five for the real legit teams. That’s the analytics that I look at."

Well he did a shitty job at putting together the bottom six then.

To be fair, players that produce at 5v5 usually cost more than league minimum. Let’s see how it goes with some cap that can be spent on actual 3rd liners like a 3C to anchor the line now that there is some available cap

 

[Behind Enemy Lines]

From Where We Were To Where We Are:

BENCHMARK: Sloan Conference 2012 with peak geniuses (Chiarelli post Cup; Burke Maple Laffs; Milbury Milburing, academic Michael Schuckers ): MIT Sloan Sports Analytics Conference Speaker | Michael Schuckers

I thought it interesting to look at one strategic area we know is vital to a hockey team's success and factors into million dollar decisions projecting teenagers into NHL hockey players. The NHL draft is somewhat the R&D function within other industries. Do this vital function better than your competitors nets a significant competitive advantage. Over time before the NHL became a global talent market, we saw teams exploit undeveloped geographic markets to great success: Oilers in Finland: Nordique in Czechoslovakia: Detroit in Sweden and Russia. Having internal data scientists has to have the potential to support this hockey operations function ... doesn't it?

NHL DRAFT DEVELOPING WORK - SAMPLE:

• NHL Draft Probability Tool (2015): NHL Draft Probability Tool

• Drunks, Lampposts and Hockey Analytics (2015): https://www.statsportsconsulting.com/wp-content/uploads/SCHUCKERS_OTTANALYTICS.pdf

• Draft by Numbers (incl. prediction of 2015 NHL Draft top picks): Michael Draft By Numbers Michael - ppt download

• Text Mining of Scouting Reports as a Novel Data Source for Improving NHL Draft Analytics (2017): https://www.statsportsconsulting.co...Reports_for_Improving_NHL_Draft_Analytics.pdf

(Note: weirdly and sloppily this academic paper cites a Vancouver Canuck blog that simulated an intern picking the Canuck draft versus the team's Scouting Director Ron Delorme. The bad methodology issues of which were thoroughly debunked here: Churko Vision: Sham Sharron is a Sham ).

• * What Does It Mean To Draft Perfectly? An Evaluation Of Draft Strategy In The National Hockey League (2017): What Does It Mean To Draft Perfectly? An Evaluation Of Draft Strategy In The National Hockey League
• NHL Draft Analytics: A Study of NHL Central Scouting (2018): NHL Draft Analytics: A Study of NHL Central Scouting - Neutral Zone

*Namita Nandakumar is now employed by the start up Seattle Kraken team (per my prior post above).

The NHL is a $3 billion+ industry and growing. The fee to join this closed cartel is currently $650 million and annual business operation per member is over one hundred million dollars with on-ice talent the big expense at $81 million top-end. Doesn't it make sense to integrate common business practices and functions like quants and data science to supplement the human expertise within the management function? I'm intrigued to watch Seattle as a start up that is integrating all of these components into their culture. The potential to nurture this professional expertise into competitive advantages in many core hockey operations areas seems significant.

 

[McDNicks17]

To be fair, players that produce at 5v5 usually cost more than league minimum. Let’s see how it goes with some cap that can be spent on actual 3rd liners like a 3C to anchor the line now that there is some available cap

Production isn't really the issue. Getting murdered defensively is where most of the bottom six come up short in goal differential.

I'd say it is a pretty funny comment from Holland considering who he's brought in or kept around. Both Turris and Sheahan haven't had a positive goal differential season since 2015. Ennis has had one since 2012. Khaira, Archibald and Granlund have never had one in their careers. Athanasiou had been getting dumpstered since his sophomore season. The list goes on and on. It's so consistent that it seems like Holland is looking at 5v5 goal differential to find players with a history of being terrible at it haha.

 

[Forgot About Drai]

Analytics isn't why these teams are successful. They're successful because they use analytics as a tool ALONG with watching the players play the sport. How will looking at spreadsheets of numbers tell you how skilled the player is? Was Pouliot as good as his analytics? Fayne?

Yes... exactly..?

Thats like the whole point man. The argument is either a) Oilers dont use this tool as much as other teams or b) arent good at using this tool.

 

[Little Fury]

Production isn't really the issue. Getting murdered defensively is where most of the bottom six come up short in goal differential.

It's both. Only one Oilers bottom sixer cracked double digits in 5v5 points this season (Khaira with 11).