There's no average fan on these boards....so I guess he's talking to people who don't read these blogs????
- Garnie

We're all pretty below average, agreed.

Jake Gardiner,...so in hockey when Jake skates wide and fires a harmless shot on net 4 times lets say...but gets caught twice out of position for good scoring chances against ( 2 on 1's )...this shows Jake as a positive player when it's actually the exact opposite. I like when Jake skates 2/10s of second faster than his teammates and leaves them at the blue line (because he held onto it too long ) while he circles the net to try and get back and stop the 2 on 1 he created going the other way...that's my fav of his. You wouldn't see this in a chart and is 1 of the reasons he was benched and the reason he hasn't seen the 1st pair yet.
- Garnie

Since the very first time someone described what Corsi or Fenwick is, people have come up with similar, REAL WORLD examples of how the stats could be flawed as a tool for assessing players.

They have always been ignored by the very same people who turn around call them narrow-minded.

In the NHL EA Sports and things announcers on TV say shape how almost all people see things. The real dumbing down of discourse occurs when people repeat what they hear on TV as fact. The idea that there is two equally valid ways to evaluate players is ridiculous. That each tells "part of the story" is wrong.

Maybe you could make a case between professional scouts vs stats but if we're just talking the average persons hockey watching experience vs stats then it's a nonstarter.

You can watch hockey all you want, but your mind is going to be warped by high-impact plays and highlights + things people say.

No one, whether they're using stats or not, is really "objective". But at least when you use stats you're measuring something - no matter how great of an eye you have, this is always going to have more weight. The idea that it might not is preposterous to a degree it's hard to talk about while still being nice.

We have irrefutable data from every single field of study that says that humans are terrible at drawing conclusions from things they watch. (Observing something in order to write down data points is not the same as watching something and drawing a conclusion).

In nearly every case, measuring something brings out radically different (or at least counter-intuitive) results than what you thought from your experience.


In hockey, you don't have to watch a player and look at his stats and come to a conclusion. It is not about balance. As a fan, when you watch hockey, you should just have fun because nearly all the conclusions you draw from your eye test will be either obvious or so biased as to be scientifically useless. There is simply too much data in a single game to organize with your eyes and mind - and that's before including all the biases (recency, confirmation etc).

It is psychically impossible to watch enough hockey that you could be informed just through what you watch. When people talk about "the eye test" they are really just repeating back some combination of games they've watched, highlights they've seen and things they've heard about the player.

But mostly what authoritative media members have said.

Further bringing the ability of the eye test into question is the idea that, at the professional level, most players have such marginally different skill levels that what we usually argue about is preposterous.

Example: it's pretty clear who the top 3-5 defenseman are, but after that, the next 15-20 guys are practically interchangeable. Anyone who tells you that they can use their eye only to authoritatively say who is better between Pieterangelo and Vlasic is lying or wrong.

Anyways, this idea that there are two equally valid ways to evaluate players is preposterous. The way to evaluate players is with statistics. Traditional scouting is also helpful, but people who watch TV aren't professional scouts.

If a players reputation is different from what his stats say, everything we know about every topic in the history of the world says that the stats should be given priority.

The idea that hockey is the once exception to the entire universe is very dumb.

Oh, and further complicating the idea is that hockey is a game were 98% of the time there are no goals. Tiny little imperceptible plays add up to huge dividends over the course of a season. For every screwup Jake Gardiner makes that results in a goal - he may be skating the puck out his zone 2/10s of a second faster than the league average. Over the course of the year, this is going to be a major fact - but it's not something you can ever really see.

This is why stats are sometimes wildly different from what people observe. Little things make a huge difference in the aggregate, where as you only remember high impact plays like a give away or a goal.
- James_Tanner

This entire post can be summed up by: I am right and you are wrong. when in reality, there ARE 2 sides to every story. there is no such thing are absolutely right or wrong. advanced stats have a role to play, but not the entire job. if you use advanced stats as the entire measuring stick, you will surely miss out on "data" these stats do not take into account. "data" the average person can witness but is not currently considered "relevant".

the main topic of the blog post was to point attention to the fact that people who used advanced stats to prove their point, are missing the point of the discussion. debating is an art form that should be protected. we need discussion, we need people to differ in opinion. without it the world would be VERY boring.

In the NHL EA Sports and things announcers on TV say shape how almost all people see things. The real dumbing down of discourse occurs when people repeat what they hear on TV as fact. The idea that there is two equally valid ways to evaluate players is ridiculous. That each tells "part of the story" is wrong.

Maybe you could make a case between professional scouts vs stats but if we're just talking the average persons hockey watching experience vs stats then it's a nonstarter.

You can watch hockey all you want, but your mind is going to be warped by high-impact plays and highlights + things people say.

No one, whether they're using stats or not, is really "objective". But at least when you use stats you're measuring something - no matter how great of an eye you have, this is always going to have more weight. The idea that it might not is preposterous to a degree it's hard to talk about while still being nice.

We have irrefutable data from every single field of study that says that humans are terrible at drawing conclusions from things they watch. (Observing something in order to write down data points is not the same as watching something and drawing a conclusion).

In nearly every case, measuring something brings out radically different (or at least counter-intuitive) results than what you thought from your experience.


In hockey, you don't have to watch a player and look at his stats and come to a conclusion. It is not about balance. As a fan, when you watch hockey, you should just have fun because nearly all the conclusions you draw from your eye test will be either obvious or so biased as to be scientifically useless. There is simply too much data in a single game to organize with your eyes and mind - and that's before including all the biases (recency, confirmation etc).

It is psychically impossible to watch enough hockey that you could be informed just through what you watch. When people talk about "the eye test" they are really just repeating back some combination of games they've watched, highlights they've seen and things they've heard about the player.

But mostly what authoritative media members have said.

Further bringing the ability of the eye test into question is the idea that, at the professional level, most players have such marginally different skill levels that what we usually argue about is preposterous.

Example: it's pretty clear who the top 3-5 defenseman are, but after that, the next 15-20 guys are practically interchangeable. Anyone who tells you that they can use their eye only to authoritatively say who is better between Pieterangelo and Vlasic is lying or wrong.

Anyways, this idea that there are two equally valid ways to evaluate players is preposterous. The way to evaluate players is with statistics. Traditional scouting is also helpful, but people who watch TV aren't professional scouts.

If a players reputation is different from what his stats say, everything we know about every topic in the history of the world says that the stats should be given priority.

The idea that hockey is the once exception to the entire universe is very dumb.

Oh, and further complicating the idea is that hockey is a game were 98% of the time there are no goals. Tiny little imperceptible plays add up to huge dividends over the course of a season. For every screwup Jake Gardiner makes that results in a goal - he may be skating the puck out his zone 2/10s of a second faster than the league average. Over the course of the year, this is going to be a major fact - but it's not something you can ever really see.

This is why stats are sometimes wildly different from what people observe. Little things make a huge difference in the aggregate, where as you only remember high impact plays like a give away or a goal.
- James_Tanner

Your "example" is ridiculous.
So you're saying that someone can not say who is the better of two players by watching them play. Even if they are very similar. The person will be either "Lying or wrong" which leaves little ability to be "right" or "Telling the truth".
In a 50/50 I'm gonna guess that 'stats' would predict that 50% of the time the person would be right. And... as you say... Stats are never wrong.

Since the very first time someone described what Corsi or Fenwick is, people have come up with similar, REAL WORLD examples of how the stats could be flawed as a tool for assessing players.

They have always been ignored by the very same people who turn around call them narrow-minded.
- Aetherial

You still need possession of the puck to shoot it wide so a missed shot is worth as much as a shot on net when you're only trying to determine who had the puck in a given situation. Corsi isn't trying to determine how dangerous a player is with the puck, it's simply trying to show who has the puck.

With that out of the way corsi on its own is still pretty flawed as a tool to determine how good an individual's possession game is largely because it's affected by many things outside of their control. Linemates, quality of competition, etc. The stats gurus have done an ok job at coming up with new stats to weed that out but then sample size is always an issue.

I like looking at them to see how much they agree or disagree with what my eyes tell me. You see Bergeron shutting down guys like Crosby while putting up 60+ points and you think he must have monster possession numbers. He does! His relative numbers are through the roof too. Then there are times where the stats disagree with my eyes, could be a combo of me not knowing poop and the stats being flawed.

Since the very first time someone described what Corsi or Fenwick is, people have come up with similar, REAL WORLD examples of how the stats could be flawed as a tool for assessing players.

They have always been ignored by the very same people who turn around call them narrow-minded.
- Aetherial

That is patently, 100% false.

One component of "analytics" is to question everything. Hockey-stats people have and do question each other on everything. There are conferences, websites, 100000 s of papers to read etc.

Like any field of study, it is always evolving. I can guarantee you that every single objection people post on this website when you start talking about stats has been looked at and overcome by the people who are leading the on this stuff.

Just for one example it's been found that score-adjusted corsi for the team is the stat that best predicts future wins. You act like people just came up with some theories and put them out there like they were facts. IN reality all this stuff is constantly being refined.

In the NHL EA Sports and things announcers on TV say shape how almost all people see things. The real dumbing down of discourse occurs when people repeat what they hear on TV as fact. The idea that there is two equally valid ways to evaluate players is ridiculous. That each tells "part of the story" is wrong.

Maybe you could make a case between professional scouts vs stats but if we're just talking the average persons hockey watching experience vs stats then it's a nonstarter.

You can watch hockey all you want, but your mind is going to be warped by high-impact plays and highlights + things people say.

No one, whether they're using stats or not, is really "objective". But at least when you use stats you're measuring something - no matter how great of an eye you have, this is always going to have more weight. The idea that it might not is preposterous to a degree it's hard to talk about while still being nice.

We have irrefutable data from every single field of study that says that humans are terrible at drawing conclusions from things they watch. (Observing something in order to write down data points is not the same as watching something and drawing a conclusion).

In nearly every case, measuring something brings out radically different (or at least counter-intuitive) results than what you thought from your experience.


In hockey, you don't have to watch a player and look at his stats and come to a conclusion. It is not about balance. As a fan, when you watch hockey, you should just have fun because nearly all the conclusions you draw from your eye test will be either obvious or so biased as to be scientifically useless. There is simply too much data in a single game to organize with your eyes and mind - and that's before including all the biases (recency, confirmation etc).

It is psychically impossible to watch enough hockey that you could be informed just through what you watch. When people talk about "the eye test" they are really just repeating back some combination of games they've watched, highlights they've seen and things they've heard about the player.

But mostly what authoritative media members have said.

Further bringing the ability of the eye test into question is the idea that, at the professional level, most players have such marginally different skill levels that what we usually argue about is preposterous.

Example: it's pretty clear who the top 3-5 defenseman are, but after that, the next 15-20 guys are practically interchangeable. Anyone who tells you that they can use their eye only to authoritatively say who is better between Pieterangelo and Vlasic is lying or wrong.

Anyways, this idea that there are two equally valid ways to evaluate players is preposterous. The way to evaluate players is with statistics. Traditional scouting is also helpful, but people who watch TV aren't professional scouts.

If a players reputation is different from what his stats say, everything we know about every topic in the history of the world says that the stats should be given priority.

The idea that hockey is the once exception to the entire universe is very dumb.

Oh, and further complicating the idea is that hockey is a game were 98% of the time there are no goals. Tiny little imperceptible plays add up to huge dividends over the course of a season. For every screwup Jake Gardiner makes that results in a goal - he may be skating the puck out his zone 2/10s of a second faster than the league average. Over the course of the year, this is going to be a major fact - but it's not something you can ever really see.

This is why stats are sometimes wildly different from what people observe. Little things make a huge difference in the aggregate, where as you only remember high impact plays like a give away or a goal.
- James_Tanner

I usually don't mind your posts, respect your options and often read your blogs but after this post I actually feel sorry for you, advanced stats have honestly took away the whole (frank)ing point of watching the sport for you. That sucks and I'm sorry to hear that, there really is no point of you watching the game if what you are seeing with your eyes doesn't mean much to you.

You still need possession of the puck to shoot it wide so a missed shot is worth as much as a shot on net when you're only trying to determine who had the puck in a given situation. Corsi isn't trying to determine how dangerous a player is with the puck, it's simply trying to show who has the puck.


- daeth

It is not even remotely difficult to come up with a scenario where a team can have much more actual possession of the puck and much fewer shots.

Beyond that, even if I accept shots towards the goal as a proxy for possession (and I actually do accept this over a large enough sample size), then I am still not convinced that we are measuring something that is as valuable as some make it out to be and I certainly don't think we can come up with numbers to tenths (I've even seen hundreths !!) of a percent accuracy and use that to compare players.

CF%, 2015-2016, players with over 1000 minutes...

Seguin - 31
Kucherov - 37
Crosby - 39
Benn - 49
Burns - 54
Tarasenko - 55
Ovechkin - 57
P. Kane - 96
Karlsson - 101
Tavares - 103
Toews - 113
Ekman-Larsson - 145


... but yeah, we shouldn't be questioning how useful these stats are right? It would be narrow-minded of us to do so.

How can anyone take a stat seriously that ranks players like that.

This entire post can be summed up by: I am right and you are wrong. when in reality, there ARE 2 sides to every story. there is no such thing are absolutely right or wrong. advanced stats have a role to play, but not the entire job. if you use advanced stats as the entire measuring stick, you will surely miss out on "data" these stats do not take into account. "data" the average person can witness but is not currently considered "relevant".

the main topic of the blog post was to point attention to the fact that people who used advanced stats to prove their point, are missing the point of the discussion. debating is an art form that should be protected. we need discussion, we need people to differ in opinion. without it the world would be VERY boring.
- richardson30

Just because there are two sides to ever story it doesn't automatically mean that both sides are 50% correct. There very much are absolute right and wrongs, that is ridiculous. 2+2=4, just for example.

But no one here - at least I don't think so - is trying to say that people shouldn't argue. It's obviously healthy.

I am not even saying that watching the games has no value - but everyone watches the games. Seriously, think about it: what would anyone's motivation be to follow hockey stats if they weren't into hockey and enjoyed watching and playing it?

Even though I'm an opinionated guy, I don't think very many of my opinions are facts. Like, if I list a top ten whatever position blog, it's not meant as a fact and no one who is being honest thinks otherwise.

But, there are facts.

We are all biased and there isn't a set way to evaluate players - obviously the more information you have about anything, the better your evaluation will be.

That being said, there is actual proof that stats analysts can make better, less biased, draft selections than scouts. That is a fact.

There are also severe problems with the human brain's construction in general that make the "eye test" problematic for anything. We are designed to recognize patterns because this is crucial to survival. This also has the downside of we are, all of us, excellent at filling in gaps of information with our biases and not knowing the difference between real memories and constructed ones. This is a real factual thing and anyone who's had an intro to Psych class can back me up here.

Third fact: When people say "eye test" they are talking as people who watch games while drunk, who watch highlights and who listen to sports radio / read the internet.

This means its more than reasonable to go choose to believe what stats say vs conventional opinion when there are discrepancies (Jake Gardiner, Shea Weber, Drew Doughty etc).

Ignoring scouting completely though isn't a real-world scenario that would ever occur because to do so would be to ignore information, which is the antitheses of an analytical philosophy.

At the end of the day, I think "stats people" just get annoyed because they offer up all this research to make a conclusion and then people just say "well stats don't tell the whole story" because for whatever reason, things you can't measure are the equal of things you can? And anyone who thinks differently is an extremist? Crazy.

100% of world history teaches us that when science or math or measurement tells us our commonly held beliefs are wrong, they are.

I admit I haven't always been the best user of stats. I happen to be in a business that rewards inflammatory statements and I have gone from a "f off with your charts guy" when I started to whatever I am now - learning along the way, in public.

But my abilities are irrelevant, because this whole argument is like arguing over whether it's better to go to a doctor or a naturopath.

When we say fancy stats - we are talking about 1) puck possession 2) shots against 3) shots for. That's just measuring the three basic things that make up hockey. It's not "advanced" and it sure as crap shouldn't be controversial.

I usually don't mind your posts, respect your options and often read your blogs but after this post I actually feel sorry for you, advanced stats have honestly took away the whole (frank)ing point of watching the sport for you. That sucks and I'm sorry to hear that, there really is no point of you watching the game if what you are seeing with your eyes doesn't mean much to you.

- ClarksonDavid

Nothing says a good time like sitting at the local sports bar swigging back pints and eating wings while reading the stat lines from the night before. Am I right?!

It is not even remotely difficult to come up with a scenario where a team can have much more actual possession of the puck and much fewer shots.

Beyond that, even if I accept shots towards the goal as a proxy for possession (and I actually do accept this over a large enough sample size), then I am still not convinced that we are measuring something that is as valuable as some make it out to be and I certainly don't think we can come up with numbers to tenths (I've even seen hundreths !!) of a percent accuracy and use that to compare players.

CF%, 2015-2016, players with over 1000 minutes...

Seguin - 31
Kucherov - 37
Crosby - 39
Benn - 49
Burns - 54
Tarasenko - 55
Ovechkin - 57
P. Kane - 96
Karlsson - 101
Tavares - 103
Toewws - 113
Ekman-Larsson - 145


... but yeah, we shouldn't be questioning how useful these stats are right? It would be narrow-minded of us to do so.

How can anyone take a stat seriously that ranks players like that.
- Aetherial

Your question makes perfect sense, but if you do the research you seem unwilling to do, you'll notice that people have dealt with this problem by testing which numbers best correlate to things like winning, which numbers are likely to regress or improve etc.

If you've thought about it this far, there's tons of stuff that deals with exactly what you're talking about.

In the NHL EA Sports and things announcers on TV say shape how almost all people see things. The real dumbing down of discourse occurs when people repeat what they hear on TV as fact. The idea that there is two equally valid ways to evaluate players is ridiculous. That each tells "part of the story" is wrong.

Maybe you could make a case between professional scouts vs stats but if we're just talking the average persons hockey watching experience vs stats then it's a nonstarter.

You can watch hockey all you want, but your mind is going to be warped by high-impact plays and highlights + things people say.

No one, whether they're using stats or not, is really "objective". But at least when you use stats you're measuring something - no matter how great of an eye you have, this is always going to have more weight. The idea that it might not is preposterous to a degree it's hard to talk about while still being nice.

We have irrefutable data from every single field of study that says that humans are terrible at drawing conclusions from things they watch. (Observing something in order to write down data points is not the same as watching something and drawing a conclusion).

In nearly every case, measuring something brings out radically different (or at least counter-intuitive) results than what you thought from your experience.


In hockey, you don't have to watch a player and look at his stats and come to a conclusion. It is not about balance. As a fan, when you watch hockey, you should just have fun because nearly all the conclusions you draw from your eye test will be either obvious or so biased as to be scientifically useless. There is simply too much data in a single game to organize with your eyes and mind - and that's before including all the biases (recency, confirmation etc).

It is psychically impossible to watch enough hockey that you could be informed just through what you watch. When people talk about "the eye test" they are really just repeating back some combination of games they've watched, highlights they've seen and things they've heard about the player.

But mostly what authoritative media members have said.

Further bringing the ability of the eye test into question is the idea that, at the professional level, most players have such marginally different skill levels that what we usually argue about is preposterous.

Example: it's pretty clear who the top 3-5 defenseman are, but after that, the next 15-20 guys are practically interchangeable. Anyone who tells you that they can use their eye only to authoritatively say who is better between Pieterangelo and Vlasic is lying or wrong.

Anyways, this idea that there are two equally valid ways to evaluate players is preposterous. The way to evaluate players is with statistics. Traditional scouting is also helpful, but people who watch TV aren't professional scouts.

If a players reputation is different from what his stats say, everything we know about every topic in the history of the world says that the stats should be given priority.

The idea that hockey is the once exception to the entire universe is very dumb.

Oh, and further complicating the idea is that hockey is a game were 98% of the time there are no goals. Tiny little imperceptible plays add up to huge dividends over the course of a season. For every screwup Jake Gardiner makes that results in a goal - he may be skating the puck out his zone 2/10s of a second faster than the league average. Over the course of the year, this is going to be a major fact - but it's not something you can ever really see.

This is why stats are sometimes wildly different from what people observe. Little things make a huge difference in the aggregate, where as you only remember high impact plays like a give away or a goal.
- James_Tanner

This essay-like post would have been much more impactful if you had taken the time to include a Tears for Fears or Talking Heads video.

This essay-like post would have been much more impactful if you had taken the time to include a Tears for Fears or Talking Heads video.
- 1979AD

Or Pavement... because Indie...

Your question makes perfect sense, but if you do the research you seem unwilling to do, you'll notice that people have dealt with this problem by testing which numbers best correlate to things like winning, which numbers are likely to regress or improve etc.

If you've thought about it this far, there's tons of stuff that deals with exactly what you're talking about.
- James_Tanner

OK... so can we eliminate Corsi from all future discussions and just go with whatever you are talking about.

Honestly, I have seen people say something similar to what you just said and then come right back to Corsi in their arguement the next day.

if it sucks, don't mention it anymore and yeah, it sucks.

While I am addressing you, you constantly have this tone of the stats not lieing and numbers being the ONLY impartial way to judge. I absolutely, 100% agree with you ... IF we have the right, and perfect numbers. We don't. Not only are the current numbers flawed from the age old problem of measurement bias that has always existed in the NHL, but I have yet to see where they solidly and consistently correlate with team or individual performance.

OK... so can we eliminate Corsi from all future discussions and just go with whatever you are talking about.

Honestly, I have seen people say something similar to what you just said and then come right back to Corsi in their arguement the next day.

if it sucks, don't mention it anymore and yeah, it sucks.

While I am addressing you, you constantly have this tone of the stats not lieing and numbers being the ONLY impartial way to judge. I absolutely, 100% agree with you ... IF we have the right, and perfect numbers. We don't. Not only are the current numbers flawed from the age old problem of measurement bias that has always existed in the NHL, but I have yet to see where they solidly and consistently correlate with team or individual performance.
- Aetherial

Why do players who are clearly and obviously better than other players have much worse possession stats? Because there are papers that explain that, numbnutz. Now, back to why possession stats are the only accurate and unbiased way to assess the value of players...

This essay-like post would have been much more impactful if you had taken the time to include a Tears for Fears or Talking Heads video.
- 1979AD

Agreed.

Why do players who are clearly and obviously better than other players have much worse possession stats? Because there are papers that explain that, numbnutz. Now, back to why possession stats are the only accurate and unbiased way to assess the value of players...
- 1979AD

I'm too lazy to do the research.

I can only assume one day people will be discussing this Magic Stat that really seems to assess and rank players in a way that makes at least some sense.

Agreed.
- James_Tanner

Actions are louder than words, Tanner! There is an edit button

Agreed.
- James_Tanner

Not Tears for Fears or Talking Heads though.

How about some Porcupine Tree?

Not Tears for Fears or Talking Heads though.

How about some Porcupine Tree?
- Aetherial

Good call - something from In Absentia preferably.

We're all pretty below average, agreed.
- daeth

When it comes to the Leafs I'd have to disagree...the other 29 teams 100% but I don't need some one telling me a guys elite because of an extra shot/60 when the guy only plays 12 - 20 mins a game against different comp/starts etc. It's insane if you think about it...but hey...it's created some jobs for some math guys I guess.

When it comes to the Leafs I'd have to disagree...the other 29 teams 100% but I don't need some one telling me a guys elite because of an extra shot/60 when the guy only plays 12 - 20 mins a game against different comp/starts etc. It's insane if you think about it...but hey...it's created some jobs for some math guys I guess.
- Garnie

When it comes to the Leafs I'd have to disagree...the other 29 teams 100% but I don't need some one telling me a guys elite because of an extra shot/60 when the guy only plays 12 - 20 mins a game against different comp/starts etc. It's insane if you think about it...but hey...it's created some jobs for some math guys I guess.
- Garnie

That guy who could never play the game is now killing it in the "war room"

- daeth

That guy who could never play the game is now killing it in the "war room"
- mlindsay