Confirmed with Link: Sens hire Pierre McGuire

[FunkySeeFunkyDoo]

Leafs have drafted quite a few Russian and Finnish players in the last two amateur drafts -- out of junior leagues in those countries. Does anyone know if the real advanced analytics data is being gathered for those leagues?

I would assume that those leagues, esp Russian Jr's, are not really covered yet. So those picks were probably driven by good old fashioned scouts using their eyeballs.

 

[burf]

Leafs have drafted quite a few Russian and Finnish players in the last two amateur drafts -- out of junior leagues in those countries. Does anyone know if the real advanced analytics data is being gathered for those leagues?

I would assume that those leagues, esp Russian Jr's, are not really covered yet. So those picks were probably driven by good old fashioned scouts using their eyeballs.

I think this mindset is driven by a lack of understanding of analytics? "Good old fashioned scouts using their eyeballs" ARE tracking analytics data, especially for teams like the Leafs. They're tracking zone entry/exit data, shot attempts, scoring chances, transition defense.

Many of the "hobby analytics" folks, and every team using analytics, has their analytics department watch as much or more tape than any "good old fashioned scout", because almost every metric you want to track isn't tracked publicly.

 

[Burrowsaurus]

He’s been an analytical guy since he entered the league

https://www.google.ca/amp/s/www.cbc.ca/amp/1.2714556


“The hiring of an analytics-focused 28-year-old assistant GM could signal a significant shift in thinking for the Toronto Maple Leafs.
The NHL club on Tuesday announced Kyle Dubas, a fast-rising hockey executive with a penchant for stats-based analysis, as the latest addition to the management staff. He replaces Claude Loiselle in a front-office shakeup that has also seen vice-president of hockey operations Dave Poulin leave the team.”

That’s why he’s considered an analytical GM

Riiight this doesn’t disprove the bias. A hockey article said he’s a analytics focused guy. Yeah. I don’t doubt that the league and its fans view him that way. His moves don’t seem to back that up tho.

Kyle Dubas Is Not an Analytics GM, and His Failures Don’t Define the Movement

one of many write ups on this.

 

[BondraTime]

Riiight this doesn’t disprove the bias. A hockey article said he’s a analytics focused guy. Yeah. I don’t doubt that the league and its fans view him that way. His moves don’t seem to back that up tho.

Again, I’m not disputing that. I’m saying that he’s considered an analytics guy, because he’s been known as a heavily influenced analytics guy since he was a 24 year old entering the OHL.

The thought he’s considered an analytic guy because he’s young and has glasses is strange, and has no basis in reality.

He’s considered an analytic guy because he’s an analytic guy, and has been since he was a young 20’s year old in the OHL. THAT’S the reason he’s viewed as an analytic guy.

 

[DueDiligence]

Riiight this doesn’t disprove the bias. A hockey article said he’s a analytics focused guy. Yeah. I don’t doubt that the league and its fans view him that way. His moves don’t seem to back that up tho.

Kyle Dubas Is Not an Analytics GM, and His Failures Don’t Define the Movement

one of many write ups on this.

Leaf fans should read this article about Dubas and then wonder why he has got the GM job.

 

[Beech]

I think this mindset is driven by a lack of understanding of analytics? "Good old fashioned scouts using their eyeballs" ARE tracking analytics data, especially for teams like the Leafs. They're tracking zone entry/exit data, shot attempts, scoring chances, transition defense.

Many of the "hobby analytics" folks, and every team using analytics, has their analytics department watch as much or more tape than any "good old fashioned scout", because almost every metric you want to track isn't tracked publicly.

I was at an Ottawa 67's in early 2010. About a week or so after the world juniors. Alex Pietrangelo was there. It took me 10 seconds to see why he was a #4 draft. And it took me 10 seconds to realize he was special.

I was at the 1992 or 1993 (I cannot remember the exact year) allstar game between the OHL and the QMJHL, I saw Daigle and realized, floater. All he did was hang around the blue-line and looked for breakaway passes...12 years before the removal of the red-line and in hockey's dead puck era and in clutch and grab and with some older buildings around, so small ice..I realize; he will not make it. Same game; Chris Pronger. He was excellent.

I would love to know what their respective analytics would have been? Would analytics have supported all 3? Would analytics have flushed Daigle out? Would analytics have swapped Pietrangelo and Bogosian. Or might it have dropped Pietrangelo??? That was the EK year. What of him?

The analytics community loves Josh Ho-Sang. A trouble maker in junior. A trouble maker in the pros. A wasted 1st round pick. I would love to know what got him drafted? The eyeball test of scouts or analytics?

 

[burf]

I was at an Ottawa 67's in early 2010. About a week or so after the world juniors. Alex Pietrangelo was there. It took me 10 seconds to see why he was a #4 draft. And it took me 10 seconds to realize he was special.

I was at the 1992 or 1993 (I cannot remember the exact year) allstar game between the OHL and the QMJHL, I saw Daigle and realized, floater. All he did was hang around the blue-line and looked for breakaway passes...12 years before the removal of the red-line and in hockey's dead puck era and in clutch and grab and with some older buildings around, so small ice..I realize; he will not make it. Same game; Chris Pronger. He was excellent.

I would love to know what their respective analytics would have been? Would analytics have supported all 3? Would analytics have flushed Daigle out? Would analytics have swapped Pietrangelo and Bogosian. Or might it have dropped Pietrangelo??? That was the EK year. What of him?

The analytics community loves Josh Ho-Sang. A trouble maker in junior. A trouble maker in the pros. A wasted 1st round pick. I would love to know what got him drafted? The eyeball test of scouts or analytics?

I think the problem is that you don't really know what you mean by analytics. What analytics are you talking about? And who is making decisions solely on the basis of those analytics?

Like I said, a team like the Leafs will have a scouting department that's tracking analytics for prospects they're looking to draft. That scouting department will be watching a lot of games, and a lot of tape, to track those analytics. Those analytics will be used to confirm the impressions they developed when they watched the games. For example, the impression "Rodion Amirov is a very good transition player" can be supported by looking at "analytics" like zone exit percentage, zone entry percentage, and zone entry carry rate, which will confirm whether the "eye test" impression is actually borne out by the data.

In terms of the "analytics" that we, as fans, have access to for prospects, the only thing we can really look at is NHLe. That tells us how a prospect's production in a given league compares to other prospects in different leagues, by developing a metric that standardizes scoring across leagues, using NHL-equivalent production rates. This is not something that any team uses to evaluate prospects at anything beyond a superficial level, or something that they would base any of their decisions on. It can be a useful tool - for example, Kucherov had among the highest NHLe in his draft class but wasn't picked until the 3rd round - but it's only a starting point that can be used to identify players that you might want to take a deeper look at. It's literally just a reflection of how many points they scored in the year(s) leading up to the draft.

 

[Tnuoc Alucard]

analytics Has proven it’s value in MLB, not so much in the NHL….. glad PM was hired, to bring some common sense to then organization.

 

[Beech]

I think the problem is that you don't really know what you mean by analytics. What analytics are you talking about? And who is making decisions solely on the basis of those analytics?
--------
highest NHLe in his draft class but wasn't picked until the 3rd round - but it's only a starting point that can be used to identify players that you might want to take a deeper look at. It's literally just a reflection of how many points they scored in the year(s) leading up to the draft.

I get you and do follow. I still fear that "analytics", regardless of what, creates one or both of the following problems;
1) it becomes a decision maker. Game 7 of last years world series. My theory has always been, Snell got pulled because the analytics told the Manger to pull him. The Jays use that approach.
2) It creates lazy people. Scouts and Hockey ops people need to go and see. Go and interact. Go and investigate. Amirov may be a good transition player. His numbers may confirm it. Will a scout walk down to the locker room and ask his team-mates about him? and will some issue surface?

I come from the Science/Engineering world. It has incredible similarities. And I can tell you, the above 2 situations arise. Failure to go and spend time in the lab testing or committing to things because the analysis showed it, generality got you into a whole lot of trouble. As did a failure to use sound judgment and experience. Analysis would show me a 0.03" thick sheet of metal will work. Now have someone try and weld it!!!!

The analytics debate began in this thread because McGuire voiced concern. To be fair his concern may be far more related to what I describe versus a dismissal of the usefulness of the info. I tend to agree with him. In my work, I was far more a theoretician than an experimentalist but was always mindful of limits.

 

[Ice-Tray]

I think the problem is that you don't really know what you mean by analytics. What analytics are you talking about? And who is making decisions solely on the basis of those analytics?

Like I said, a team like the Leafs will have a scouting department that's tracking analytics for prospects they're looking to draft. That scouting department will be watching a lot of games, and a lot of tape, to track those analytics. Those analytics will be used to confirm the impressions they developed when they watched the games. For example, the impression "Rodion Amirov is a very good transition player" can be supported by looking at "analytics" like zone exit percentage, zone entry percentage, and zone entry carry rate, which will confirm whether the "eye test" impression is actually borne out by the data.

In terms of the "analytics" that we, as fans, have access to for prospects, the only thing we can really look at is NHLe. That tells us how a prospect's production in a given league compares to other prospects in different leagues, by developing a metric that standardizes scoring across leagues, using NHL-equivalent production rates. This is not something that any team uses to evaluate prospects at anything beyond a superficial level, or something that they would base any of their decisions on. It can be a useful tool - for example, Kucherov had among the highest NHLe in his draft class but wasn't picked until the 3rd round - but it's only a starting point that can be used to identify players that you might want to take a deeper look at. It's literally just a reflection of how many points they scored in the year(s) leading up to the draft.

There are companies that track every possible data point on a per game basis. NHL teams subscribe to this service, the Sens are one of them.

You don’t need an analytics department, you simply need a subscription to data already compiled by professionals.

 

[DrEasy]

I was at an Ottawa 67's in early 2010. About a week or so after the world juniors. Alex Pietrangelo was there. It took me 10 seconds to see why he was a #4 draft. And it took me 10 seconds to realize he was special.

I was at the 1992 or 1993 (I cannot remember the exact year) allstar game between the OHL and the QMJHL, I saw Daigle and realized, floater. All he did was hang around the blue-line and looked for breakaway passes...12 years before the removal of the red-line and in hockey's dead puck era and in clutch and grab and with some older buildings around, so small ice..I realize; he will not make it. Same game; Chris Pronger. He was excellent.

I would love to know what their respective analytics would have been? Would analytics have supported all 3? Would analytics have flushed Daigle out? Would analytics have swapped Pietrangelo and Bogosian. Or might it have dropped Pietrangelo??? That was the EK year. What of him?

The analytics community loves Josh Ho-Sang. A trouble maker in junior. A trouble maker in the pros. A wasted 1st round pick. I would love to know what got him drafted? The eyeball test of scouts or analytics?

Well, you can't blame analytics for Daigle going 1OA.

It would be interesting to see a study on the impact of analytics as a whole on the quality of drafts in the last, say, 10 years. Are there more or fewer first round busts? More or fewer late round miracles? Of course , there are probably plenty of confounding factors (more/better scouting, more video to go through, maybe better nutrition and general preparation early on for young prospects reducing the number of good and bad surprises).

 

[AchtzehnBaby]

There are companies that track every possible data point on a per game basis. NHL teams subscribe to this service, the Sens are one of them.

You don’t need an analytics department, you simply need a subscription to data already compiled by professionals.

You still need guys that know what they are looking at/for.

Real GAR experts!

 

[FunkySeeFunkyDoo]

I think this mindset is driven by a lack of understanding of analytics? "Good old fashioned scouts using their eyeballs" ARE tracking analytics data, especially for teams like the Leafs. They're tracking zone entry/exit data, shot attempts, scoring chances, transition defense.

Many of the "hobby analytics" folks, and every team using analytics, has their analytics department watch as much or more tape than any "good old fashioned scout", because almost every metric you want to track isn't tracked publicly.

For the record, I understand the concept of "analytics" quite well -- I have a background in statistics, do time series analysis on a daily basis as part of my job, and followed the analytics revolution in baseball (SABRmetrics) back in the late 80's and early 90's VERY closely. I will admit that I have not looked closely at the latest and greatest hockey analytics, so maybe there are some aspects of it that differ from the SABRmetrics.

I had assumed that the data for junior leagues was provided by some 3rd party -- either the leagues themselves, or a separate company (see @Ice-Tray 's post).

It is possible that the Leafs gather their own data for the Russian junior leagues, although I'd point out a few key things that make me doubt it:

1. Any analytics based decision is only valid if the data is "good". By "good" we mean both that it is collected reliably, consistently and without any kind of bias -- AND that it is a sufficiently large sample size (probably on >200 samples before it starts to become meaningful).
2. To provide valid context, the data needs to be collected not just for an individual player, but for the entire league. IE, you can't just watch Dmitry Ovchinnikov games and compile the data for him -- you need to have that reliable, consistent data for all the 18 year old forwards so that you can assess where he stands relative to the entire population. (IDEALLY this would actually be true over multiple years, so you could analyze how well the data predicts downstream performance correctly -- but I'm guessing the hockey analytics people arent there yet.)

If, as you suggest, the Leafs are collecting their own data for the Russian leagues, then this implies they have a staff member attending the majority of the games played in those leagues.

I also think the premise that their "good old fashioned scouts" are doing the data collection would be considered questionable by most statisticians. For a variety of reasons, I think it's unlikely those guys would collect data "reliable, consistent and without any kind of bias".

NOTE -- I see just looking at Hockey DB that many of the Russian & Finnish players the Leafs have drafted played in the Finnish pro league or the KHL in the year leading into their draft. It seems much more likely to me that those leagues have good coverage -- IE accurate data -- than the junior leagues.

 

[Micklebot]

For the record, I understand the concept of "analytics" quite well -- I have a background in statistics, do time series analysis on a daily basis as part of my job, and followed the analytics revolution in baseball (SABRmetrics) back in the late 80's and early 90's VERY closely. I will admit that I have not looked closely at the latest and greatest hockey analytics, so maybe there are some aspects of it that differ from the SABRmetrics.

I had assumed that the data for junior leagues was provided by some 3rd party -- either the leagues themselves, or a separate company (see @Ice-Tray 's post).

It is possible that the Leafs gather their own data for the Russian junior leagues, although I'd point out a few key things that make me doubt it:

1. Any analytics based decision is only valid if the data is "good". By "good" we mean both that it is collected reliably, consistently and without any kind of bias -- AND that it is a sufficiently large sample size (probably on >200 samples before it starts to become meaningful).
2. To provide valid context, the data needs to be collected not just for an individual player, but for the entire league. IE, you can't just watch Dmitry Ovchinnikov games and compile the data for him -- you need to have that reliable, consistent data for all the 18 year old forwards so that you can assess where he stands relative to the entire population. (IDEALLY this would actually be true over multiple years, so you could analyze how well the data predicts downstream performance correctly -- but I'm guessing the hockey analytics people arent there yet.)

If, as you suggest, the Leafs are collecting their own data for the Russian leagues, then this implies they have a staff member attending the majority of the games played in those leagues.

I also think the premise that their "good old fashioned scouts" are doing the data collection would be considered questionable by most statisticians. For a variety of reasons, I think it's unlikely those guys would collect data "reliable, consistent and without any kind of bias".

NOTE -- I see just looking at Hockey DB that many of the Russian & Finnish players the Leafs have drafted played in the Finnish pro league or the KHL in the year leading into their draft. It seems much more likely to me that those leagues have good coverage -- IE accurate data -- than the junior leagues.

I know there's been a lot of work using AI to analyze video to collect data, that might be available from the leagues they want info from

Here's an example of one such company

https://www.icebergsports.com/

 

[FunkySeeFunkyDoo]

I know there's been a lot of work using AI to analyze video to collect data, that might be available from the leagues they want info from

Here's an example of one such company

https://www.icebergsports.com/

That is pretty interesting, I'll take a look at it later today. Assuming that it works, then it would take the human element out of the data collection ... which is huge, as the subjective decision made by humans is a big source of error.

Note that the text in the link refers to Bill James, who authored an annual report titled "The Baseball Abstract". I remember purchasing it in 1983 or 84. Very interesting and groundbreaking stuff.

One thing that is often misunderstood is that people think James introduced the concept of measuring player performance based on statistics.... for baseball that really isnt true. Baseball had almost always been a game where stats played a big role. What James demonstrated is that the stats which were traditionally used to evaluate players (Batting Average, RBIs, ERA) were unreliable or incorrect, that should be replaced by other stats.

 

[pzeeman]

One thing that is often misunderstood is that people think James introduced the concept of measuring player performance based on statistics.... for baseball that really isnt true. Baseball had almost always been a game where stats played a big role. What James demonstrated is that the stats which were traditionally used to evaluate players (Batting Average, RBIs, ERA) were unreliable or incorrect, that should be replaced by other stats.

That's what Moneyball really is. It's not just "using stats". It's not just being cheap. It's using the stats that no one else is using to find undervalued players that help you win. In the A's example, people were undervaluing walks (not making outs) and therefore OBP in favour of batting average, RBI and over rating the negative value of a strike out. Beane was able to build his winning team on the cheap because he was getting players who helped him win that other teams didn't want because they didn't show well in the metrics that they thought would help them win.

I see hockey as much more difficult to track that way and find the inefficiencies. Because baseball can be broken down really easily into well defined, repeated events (a plate appearance) it's easier to measure than the chaotic, constantly flowing game of hockey. It's interesting to see what people are measuring though.

And finally, the true test is to take your measure and see if it predicts the past. Does success in your measure correlate with success in the standings? Or does it end up random?

 

[Masked]

That is pretty interesting, I'll take a look at it later today. Assuming that it works, then it would take the human element out of the data collection ... which is huge, as the subjective decision made by humans is a big source of error.

Note that the text in the link refers to Bill James, who authored an annual report titled "The Baseball Abstract". I remember purchasing it in 1983 or 84. Very interesting and groundbreaking stuff.

One thing that is often misunderstood is that people think James introduced the concept of measuring player performance based on statistics.... for baseball that really isnt true. Baseball had almost always been a game where stats played a big role. What James demonstrated is that the stats which were traditionally used to evaluate players (Batting Average, RBIs, ERA) were unreliable or incorrect, that should be replaced by other stats.

I remember reading Bill James work in the 80's also. It made complete sense to me the first time I read it. What really sets baseball apart from hockey is that every event is independent. What happens when pitcher throws a ball or a batter swings is not dependent upon any other player for the most part.

In hockey how do you break apart the value of a defence pairing that is on the ice with each other 95+% of the time? Every event that happens on the ice for those 2 players factors in both of them.

 

[FunkySeeFunkyDoo]

That's what Moneyball really is. It's not just "using stats". It's not just being cheap. It's using the stats that no one else is using to find undervalued players that help you win. In the A's example, people were undervaluing walks (not making outs) and therefore OBP in favour of batting average, RBI and over rating the negative value of a strike out. Beane was able to build his winning team on the cheap because he was getting players who helped him win that other teams didn't want because they didn't show well in the metrics that they thought would help them win.

I see hockey as much more difficult to track that way and find the inefficiencies. Because baseball can be broken down really easily into well defined, repeated events (a plate appearance) it's easier to measure than the chaotic, constantly flowing game of hockey. It's interesting to see what people are measuring though.

And finally, the true test is to take your measure and see if it predicts the past. Does success in your measure correlate with success in the standings? Or does it end up random?

This is probably the most correct post I've ever seen on this site. Well done.

 

[Ice-Tray]

See, see! All you smart data guys! hehe, love you folks

 

[Ice-Tray]

You still need guys that know what they are looking at/for.

Real GAR experts!

Though according to DJ it’s the coaches that look over the subscription data post game and use it to help evaluate what they watch in video, or at least make things faster so they don’t have to count zone exits for everyone post game. Says it helps them evaluate how positional play and roles are working within their system.

 

[thinkwild]

I know there's been a lot of work using AI to analyze video to collect data, that might be available from the leagues they want info from

Here's an example of one such company

https://www.icebergsports.com/

I've been wondering when this would start. Ive seen some amazing applications of AI lately, in some of the most unexpected places. It really is going to be a game changer. Looks like i would have to get on their mailing list for a demo, but i am also fascinated to see how this turns out.

 

[Knave]

Moneyball was a book about ignoring the As great regular season pitching and some of their big name big money hitters (doing steroids, YEAH BABY!) in favor of fixating on a few of the roleplayers that worked out at Oakland.

This article is always a fun read:
The Many Problems With 'Moneyball'

So while baseball is left to sort out for itself exactly what the true impact of sabermetrics is, it's always good to remember that no set of statistics has ever been invented that predicts the future so well as it predicts the past.

The article is not a denial of analytics or sabermetrics. It just asks people put it into perspective.

And if the impact of sabermetrics is in dispute in something as 'simplistic' as baseball... I don't hold out much hope that analytics in hockey today can have much impact especially on helping front offices pick out players successfully from drafts, UFA and trades.

 

[DJB]

Can we change the title to analytics vs eye ball scouting or something?

 

[pzeeman]

Moneyball was a book about ignoring the As great regular season pitching and some of their big name big money hitters (doing steroids, YEAH BABY!) in favor of fixating on a few of the roleplayers that worked out at Oakland.

This article is always a fun read:
The Many Problems With 'Moneyball'



The article is not a denial of analytics or sabermetrics. It just asks people put it into perspective.

And if the impact of sabermetrics is in dispute in something as 'simplistic' as baseball... I don't hold out much hope that analytics in hockey today can have much impact especially on helping front offices pick out players successfully from drafts, UFA and trades.

That was a really good read. Thanks!

 

[FunkySeeFunkyDoo]

Moneyball was a book about ignoring the As great regular season pitching and some of their big name big money hitters (doing steroids, YEAH BABY!) in favor of fixating on a few of the roleplayers that worked out at Oakland.

This article is always a fun read:
The Many Problems With 'Moneyball'



The article is not a denial of analytics or sabermetrics. It just asks people put it into perspective.

And if the impact of sabermetrics is in dispute in something as 'simplistic' as baseball... I don't hold out much hope that analytics in hockey today can have much impact especially on helping front offices pick out players successfully from drafts, UFA and trades.

Interesting arcticle for sure.

But, as a guy who was a hardcore baseball fan in the 80's, 90's and early 2000's -- I would say that his assertion that there was no competitive imbalance due to spending doesn't match my memory.