[Article] Scott Walker on Heavy Analytics Approach

[DownUndaCanuck]

I think you clearly need a mix of analytics, good hockey knowledge and at the end of the day - heart and grit.

 

For those using the Arizona comparison, I'm pretty sure Tampa built it's dynasty on analytics. The difference is that Tampa have high end skill, they play with heart and energy and they defend so well as a team.

 

Bruce isn't a particularly analytic-heavy coach it seems so I think bringing some analytics, especially on defence, will help round out his system and hopefully complement him nicely. We have been a poor analytic team for many years, it's worth a shot but I hope they don't get too obsessed with it.

[Grape]

6 hours ago, Elias Pettersson said:

How did teams draft and develop talent in the 80's and 90's without analytics?  Did they actually go to the games and scout them using the eye test?  

 

Would analytics allow you to figure out the character of a hockey player and his desire to win a Stanley Cup at all costs?  

 

Not sure how you can use computers and charts to figure out if someone is going to be a great NHL player.  It's alot easier just to visit and talk to their moms, that will open up alot more information about the player than a WAR chart...

I mean, analytics play a part in why the quality of hockey is much higher than what it was back in the 80's and 90's. Teams didn't suffer from the lack of analytics back then because no one used analytics. That logic is like someone asking: "how did goalies make saves back in the 80's without these new school techniques?" in order to prove that old school techniques are better. Well, they worked back then because new goalie techniques hadn't been discovered.

 

The issues with valuing "the character of a hockey player and his desire to win a Stanley Cup" as a main measure of scouting is that, 1. it is totally subjective and difficult to gauge, and 2. they clearly aren't things that matter as much as product on ice. Analytical measures that drive success are obviously much more present in consistently winning teams than "amount of character per team," however you measure that.

 

I hope you don't actually think that talking to a player's mom is a better way to scout than looking at objective data that drive success (WAR is generally not a stat that works in hockey so I don't think people use it much anyway). I mean, even if you're serious, Analytics also don't prevent things like "talking to the player's parents" as you put it. Discovering who these players are as people still matter of course, and teams still go about doing so - I doubt there is a correlation between "amount of analytics used" versus "amount of parents talked to" in this league.

Edited May 26, 2022 by Grape

[King Heffy]

28 minutes ago, Grape said:

Toronto is such a horrendous example to use if you're against analytics. I think it shows a lack of understanding of the variability and randomness of hockey if you're using them as an example. Toronto just happens to have hilariously come out as disadvantaged from such an aspect of this sport.

 

Also, just because a team isn't built by analytics does not mean that analytical tools will think they are worse. And just because a team is built through analytics does not mean analytics will love them. There are tons of variables in determining the competency of a GM/coach other than: do they use analytics? For example, predictive analytics never saw Arizona as a good team. Predictive analytics did however, see that the cup-winning STL team, despite being dead last in January of that year, was much better than their record indicated.

Toronto's defence is absolutely atrocious and their top forwards refuse to put in an acceptable effort.  They're a great example of what happens when you put together a team without understanding the sport at all.  They got rid of all the competent hockey people and promoted an unqualified moron who has no business working in any hockey related job.  Dubas has done a great job of handicapping the Laffs for years with his contracts to their floaters and promoting a defence-optional culture.

Edited May 26, 2022 by King Heffy

[Grape]

I think the main misunderstanding about analytics is that it is a predictive tool more than anything.

 

People scoff when they're told "yes, you won the game, but analytics say that you were expected to lose," and then deem that analytics are completely worthless since it's more important to win the actual game. Of course, what matters in the moment is getting a W, but what matters for future assessment is deeming what is conducive to sustaining W's, which is why analytics is used.

 

Example: if in game 1, team A beats team B, but the high-danger-scoring-chances were 15-1 for team B, who would you bet wins game 2? Without more information, I think all of us would guess team B.

 

Now, of course, this is such an exaggerated example of one team getting absolutely dominated that the same can be deduced without analytics, and just through the eye-test. But in real-life cases, the eye-tests can fail because of a variety of reasons (emotional bias, misplaced value, memory, and most importantly the impossibility of watching every moment of every game etc...), which opens up the value of analytics - it can help assess the future of a team or player much more efficiently than simple and raw stats such as W's and L's or G's and A's.

 

Of course, as has been said in this thread, analytics should be used as a supporting tool because we haven't reached a point in time where analytics are even close to perfect. But it is something that NEEDS to be used for player and team evaluation in order for modern teams to sustain winning hockey. Teams wouldn't be heavily investing in it if it weren't.

Edited May 26, 2022 by Grape

[Grape]

40 minutes ago, King Heffy said:

Toronto's defence is absolutely atrocious and their top forwards refuse to put in an acceptable effort.  They're a great example of what happens when you put together a team without understanding the sport at all.  They got rid of all the competent hockey people and promoted an unqualified moron who has no business working in any hockey related job.  Dubas has done a great job of handicapping the Laffs for years with his contracts to their floaters and promoting a defence-optional culture.

Analytics doesn't ignore defense whatsover. In fact, defense is paramount to solid analytics for most models, especially in a sport where perfect defense beats perfect offense.

 

Blaming their lack of defense on analytics is a pretty good indicator that you're just looking for reasons to attribute blame towards analytics. Dubas may not be a good GM, but what does any of what you said have to do with analytics?

 

I can tell you definitively that analytics doesn't support optional defense or cap mismanagement lol

Edited May 26, 2022 by Grape

[DSVII]

I feel this discussion is just clickbait by the media. Walker may be biased against using it but he's an assistant coach at the end of the day. If the GM and head coach use the tools given to them. No biggie. Analytics are important and they are just one component of a successful team. You need all sorts of people in the room to succeed.

 

 

[DSVII]

6 hours ago, King Heffy said:

Same with Toronto.

The Tavares signing didn't strike me as an analytics based decision. 

 

And it's more with Dubas being a rookie and letting his RFAs like Nylander take him to the cleaners on setting the tone for their extensions.

[Canucklehead73]

If analytics were everything they would not scout players in person to watch actual games played.... Always need the eye test to confirm the data, especially when you take into account that it is a team sport with 5 other players on the ice affecting your "data" both positively and negatively.

[Alflives]

2 hours ago, DSVII said:

The Tavares signing didn't strike me as an analytics based decision. 

 

And it's more with Dubas being a rookie and letting his RFAs like Nylander take him to the cleaners on setting the tone for their extensions.

Dumbass also traded away Kadri (a great player, who contributes more to winning than Tavares) in return for a soft Dman and Alex "the useless" Kerfoot.  Dumbass, and his analytics, got taken to the woodshed on that trade.  

Edited May 26, 2022 by Alflives

[Down by the River]

36 minutes ago, Canucklehead73 said:

If analytics were everything they would not scout players in person to watch actual games played.... Always need the eye test to confirm the data, especially when you take into account that it is a team sport with 5 other players on the ice affecting your "data" both positively and negatively.

Who said they were?

[Down by the River]

10 hours ago, Elias Pettersson said:

Scott Walker was a head coach in junior hockey for 5 years.  He was also the player development consultant for the Canucks for 2 years, working directly with our prospects.  He was also the assistant coach of the Canadian Olympic team in South Korea in 2018.

 

You don't acquire those positions by being dumb and not knowing how to evaluate players.  Scott Walker is old school.  So are alot of other coaches in the NHL, namely Bruce Boudreau.  Boudreau was the one who hired Walker in the first place, so obviously he thinks highly of him.

 

I'm not saying that analytics doesn't have a place in hockey or in sports, but alot of the so called analytic computer nerds who know nothing about hockey think analytics is the only thing that matters, which of course is not true.

Give me an example of someone who feels that way.

 

It seems like you and others just don't like analytics and want to misrepresent what it is so that you can call people nerds and call others old school to make yourself sound a certain way. 

 

What you don't seem to understand, or choose to ignore, is that Scott Walker being a good coach without using analytics is not necessarily a good thing. Because maybe if he kept his same coaching style but also considered analytics he could eke out a little bit more success from his players that might be the difference in winning one more game per year. These are the details that matter. Nobody is saying you can't be a hard-nosed, motivating coach AND also consider analytics (except for you and some other people in this thread). 

[Maddogy]

My suspicion is that whenever the coaching staff is complaining about analytics, they are saying that the data they have been fed were useless. The data looks neat on the screen but there was nothing helpful to improve the PK, solve the struggling PP or help players to play to their potential. I don't think any coach is going to complain about the analytics guys if all of the data contributed to winning. 

 

To put it differently, with analytics, why did the Panthers score only 3 goals in 4 games? With analytics why did one of the best defensive team as well as goalie in the regular season $&!# their pants against the Oilers?

 

In fact, my speculation is that the reason Travis Green changed his system and confused the heck out of Canucks players is because the analytics guys told him how bad the Canucks were. 

Edited May 26, 2022 by Maddogy

[awalk]

15 hours ago, Bissurnette said:

Article on CanucksArmy which was surprisingly objective and well-written. Scott Walker doesn't believe in the pure analytics approach. If we're looking for difference of opinions in Hockey Ops, losing Walker behind the bench is definitely going to hurt us. Hockey is a fast sport, the data metrics will never be perfect. You need a pitbull like Walker motivating the players when things aren't going well. I'm very disappointed that he's not coming back.

 

https://canucksarmy.com/2022/05/25/scott-walkers-interview-canucks-analytics/

 

 

EDIT: Mods please move if this thread is not in the right place. Thanks.

 

 "I mean, show me an analytical team that’s won the Stanley Cup"

 

 

 

I like Scott, I have a picture of me as a very small kid with him at one of the first Canucks for Kids events. But like... what?

TBay has this nerd helping them win cups, seems to be helping ^ 

 

Like with most things, it's all about a balanced approach.

[Zhukini]

13 hours ago, IRR said:

This comment makes zero sense!! I don’t know why that would even remotely apply to him when he has a constant & consistent winning record. It’s not like his teams do really well for a short period, then s*** the bed! 

In the regular season then crap the bed in the playoffs? That's what he does. 

 

And it checks out, sure you need some extra motivation to get your A game against Buffalo in January, if you're a motivator and that's your calling card, you're screwed come playoff time. 

[TheRussianRocket1994]

25 minutes ago, Zhukini said:

In the regular season then crap the bed in the playoffs? That's what he does. 

 

And it checks out, sure you need some extra motivation to get your A game against Buffalo in January, if you're a motivator and that's your calling card, you're screwed come playoff time. 

Lol! As if that’s all he is as a coach. Come on!! 
 

I guess the “Rah Rah” wasn’t turned up enough in the playoffs and players tuned it out, yet come the next reg season it’s turned up again and the team is doing well again. It doesn’t work that way…if it eventually wears thin with players, that’s the end. There is way more to winning and losing in the playoffs and only so many teams / player succeed. There are many great coaches / players who may not have success in the playoffs, but that doesn’t mean they were never capable of it…there’s a million factors! 

Edited May 26, 2022 by IRR

[Canucklehead73]

1 hour ago, Down by the River said:

Who said they were?

Uhhh just sharing my opinions on analytics, there was a conversation going on about the pros and cons of relying too much on it, was not calling anyone out so not sure what you mean...

[MikeyD]

I find it funny how everybody has been pointing out Tampa Bay as a heavy analytics team yet I've never heard them preach analytics in any interviews before. I've heard plenty about how they're going out and obtaining good traditional hockey guys to fill the roles they need but I've never once heard about how they're acquiring players based on metrics. 

 

Every single team in the league has an analytics department. There's obviously gonna be a position there. But to claim T-Bay is an analytics based team is something I find very odd when they specifically target certain players in the league that don't run good numbers analytics-wise. Their success is likely contributed away from a lot of metrics that are used to look at successful teams. They're a risky and extremely creative hockey team, which is going to skew a lot of analytics metrics. 

 

Guys like JT Miller are like this. On paper heading into this season, none of the metrics pointed him out to be a guy who could post nearly 100 points, but he passed the eye test as to how he could do this and that's because he plays a risky and creative offensive game. But playing that game will lower things like your passing %, your offensive zone transition %'s, all of the entry %'s and whatnot. Most of the Tampa Bay offense is constructed in this manner. All of the elite offensive teams have the one same trait and that's flair. Teams that play basic simple hockey just don't succeed in the NHL consistently. A big part of the game is out-thinking your opponent and Tampa is elite at this. 

 

Can anybody even link me to a post-game or some type of interview of Tampa Bay attributing their success to analytics? 

[Ghostsof1915]

It's a tool like anything else.

Used properly, as long as it helps you. I'm fine with it.

I do understand the concept that it's not the end all/be all answer.

[Gawdzukes]

Scott Walker is an absolute stud. I'll miss him and the character he brings. It's a little over the top but I love his take. Analytics dweebs are quite dumb and so are the inferences they try to make more often than not. Any person who uses the eye test also most likely uses an eye test combined with memory to get to the same results, except with the eye test you have context. For instance I have a pretty good idea Horvat is better at finishing chances in front of the net than Myers and I don't need some silly mathematical breakdown to tell me who I'm putting in front of the net on the PP because I've already run the numbers in my head and gathered that from watching him play. Then there is this stuff like what they call high danger chances which in reality is just shots close to the net. They have a real bad habit of labelling analytic measurements incorrectly with different labels then what they're actually measuring in a vain attempt to sound smart.

 

Don't get me wrong analytics are useful but you need a good understanding of hockey to use them as a tool to narrow down finer points that one may be missing (over time) or just not notice in an eye test. Trying to claim one guy is a better forechecker because he he got 60% retrieval one game (3/5) and another guy got 20% (1/5) is silly without any context. To just think you can look at a spreadsheet and know the game without putting in any work is just wrong.

 

 

[brownky]

17 hours ago, King Heffy said:

As someone who personally agrees with Walker on this, I'm not pleased to seeing this at all.  We've seen what happened in Toronto and Arizona when the qualified hockey people were driven out by the analytics dweebs.

Your CF% on that comment shows you have negative influence in the defensive zone.

 

You should be traded for a bag of pucks and the corpse of Michael Grabner because he was fast and shoots lots at the goalie.

 

(I too loathe analytics. They're just stats - pick what you want to tell the story you want to tell...)