Nothing in the post you quoted was an attack on analytics experts, so why are you bringing that up as an argument to ignore his post? Your post is the self righteous drivel that just annoys people on the other side of this argument.

He's correct.
Stats are a useful tool. True.
Stats by themselves will sometimes give you incorrect conclusions (especially if used incorrectly or you use the wrong stats). True.

Both sides need to stop using absolutes and they might make some progress at not only properly evaluating players, but also opening meaningful dialogue with the opposition.
- Thecakeisalie

Ummm piss off? My point about the arrogance of some of the analytics community, a community I am neither part of nor oppose, stands. Who really gets buthurt about what post it was attached to? Move along, Jabroni.

James, I think you're a good blogger, but you ridiculously overrate the analytics we as fans have access to. They just aren't that good. Use them as a tool to help you know what happened, but not much value for rating most players. They just aren't good enough. They should make up about 20% of the conversation.


What you are conflating, and don't understand is that there is a difference between the sound science of analytics, and the practical application and results when put into practice in a particular sport. Until they improve the data collection methods, analytics for a team sport like hockey will never be as valuable as you make them out to be. Now if we're talking about analytics that the teams have access to, that's a different story.
- MJL

Good post

James, I think you're a good blogger, but you ridiculously overrate the analytics we as fans have access to. They just aren't that good. Use them as a tool to help you know what happened, but not much value for rating most players. They just aren't good enough. They should make up about 20% of the conversation.


What you are conflating, and don't understand is that there is a difference between the sound science of analytics, and the practical application and results when put into practice in a particular sport. Until they improve the data collection methods, analytics for a team sport like hockey will never be as valuable as you make them out to be. Now if we're talking about analytics that the teams have access to, that's a different story.
- MJL

Mike Babcock had an interesting comment a week or so ago during a media scrum. A reporter asked him if he/they used analytics. His response was yes we do, but not the ones you look at, we have our own. Analytics is a useful tool that is part of a total package. To draw every conclusion based solely on analytics would be short sighted. Observation and experience are still valuable tools in team building.

Mike Babcock has an interesting comment a week or so ago during a media scrum. A reporter asked him if he/they used analytics. His response was yes we do, but not the ones you look at, we gave our own. Analytics is a useful tool that is part of a total package. To draw every conclusion based solely on analytics would be short sighted. Observation and experience are still valuable tools in team building.
- hagar58

Yea, the analytics we have access to are either archaic, or insufficient due to poor data, but it's all we have. So by all means use them as a tool because they help to paint the picture. The problem is some drastically overvalue them, such as the blogger.

Ummm piss off? My point about the arrogance of some of the analytics community, a community I am neither part of nor oppose, stands. Who really gets buthurt about what post it was attached to? Move along, Jabroni.
- Larsson_fan

Mature response to criticism. Good job.

From the post you made I assumed you were taking a shot at the anti analytics community. Since you mentioned you would stay off the boards if the quoted was true, I assumed you were suggesting it wasn't and were therefore attacking the other side for behaving like they were better than analytics guys (which some do to be fair although I think this is slowly changing):

When the 100% analytics guys start acting or behaving like they are better/truer fans because they pour over stats than the guy that goes to as many games as he can afford and watches the rest on TV then I will simply stay off hockey boards entirely. No time for pompous ass hats.

If you were actually agreeing with the OP, then my bad. Maybe I misinterpreted?

Mike Babcock had an interesting comment a week or so ago during a media scrum. A reporter asked him if he/they used analytics. His response was yes we do, but not the ones you look at, we have our own. Analytics is a useful tool that is part of a total package. To draw every conclusion based solely on analytics would be short sighted. Observation and experience are still valuable tools in team building.
- hagar58

This and MJL's earlier post sum things up perfectly.

It's a bit ironic, Tanner, that you are railing against extremists and people that fail to create proper dialogue and have open discussions about analytics when you appear to fall into that category yourself.

The conclusions you draw purely from hockey statistics are not facts. They are pretty numbers that may have some relevance. They cannot be facts until hockey statistics properly cover every eventuality (which may be impossible until the creation of supercomputers capable of doing all the work instead of our limited human intelligence), the people using them know exactly how to use them, which to use, and how to interpret them properly, and until the collection process to get those original numbers is perfect.

Tanner, on another note, your prediction about the Leafs finish this year is starting to look more possible. I think it will depend a lot on whether our rookies can maintain their current level of play or if they hit the rookie wall and fall off. Considering how many games we have left to play and all the back to backs, I'm a bit worried some of them will have trouble with the schedule.

Tanner I don't get it, who are you arguing with? Your arguments all depend on straw men or tiny minorities. This isn't a debate anymore. No one really questions the data. What gets questioned is context and how data relates to it. This is a much more relevant and difficult conversation to have but you never touch on it. You just slam "non believers" and tell everyone that if they don't let the data decide everything, they're morons.

There's nothing wrong with using shooting metrics(the term puck possession is not being used correctly). It is a very useful tool for analyzing the effectiveness of players and teams. Other advanced stats are also very useful.

In reality shooting metrics should only be used as an indicator(a very useful one) to look more deeply into a player or team. To declare a team or player great or terrible based solely on Corsi/Fenwick is wrong. Corsi is a result based stat and not a process based stat as you have declared it to be.

This is why real hockey coaches, managers and players with experience are needed. They absolutely should incorporate all the tools available to them but you must realize that it isn't shot attempts that makes a player or team effective. It's all the "actual hockeying" that players and teams do which eventually lead to those shot attempts which eventually lead to goals which eventually lead to wins.(also please recognize that all shots are not created equal)

Positioning, decision making, reading the play, skating(speed and agility), strength, winning puck races and board battles, maintaining puck possession(the real kind where a team will find a teammate to pass to rather than simply dumping the puck in out of fear), passing, positioning and also most importantly positioning.

In case I forgot to mention...positioning.

As for Gardiner, you were right that he is much better than people give him credit for however this conclusion was dead wrong two seasons ago despite his positive Corsi. Once or twice a game he would make what I'm calling a critical error/giveaway that would lead to a breakaway or odd man rush for the opposition. If he's on the ice for 8 shots for in a game and only 2 shots against but those two shots are breakaways than his 80% Corsi doesn't mean anything in that specific game because those 2 shots against are more likely to lead to goals than the 8 shots for.

Over the past two seasons these critical errors are all but gone. Oct/Nov of last season were still bad but since then Gardiner has been great at limiting thos errors and he is now a reliable two-way defender because of it. Without the critical errors his Corsi numbers are now more meaningful because the danger value of the shots for and shots against are now more closely related to each other.
- Njuice

A couple of things here: 1) shot attempts are a proxy for possession. You just need to get a rough idea of who helps give their team a positive shot differential. What you call it is unimportant.

2) No one bases their entire judgement on a players corsi and nothing else. This is a cheap straw man argument people use, but no one I talk to, follow or learn from does this.

3) The argument that "actual hockeying" goes into to making the stats is true, but I don't think it's a point against what I'm saying. You can't measure leadership, but if leadership helps, you'll see an uptick in things like shot differential when that player does his leading. You can't measure digging in the corners, but if it's an effective strategy it will show up in the shots.

4) all shots are not equal, but it doesn't matter because the bigger sample size you have the more accurate your projections will be. If you have a big enough sample size, you'll have an average shot difficulty, so it won't matter.

For example, in one game you might take fifty shots and only 2 of them were of high quality. This would be an anomaly because in a 10 games, or 100 or a 10000 there would work out to be an average amount (call it X) of good scoring chances in each game.

Admittedly, this is the kind of advanced stuff that limits understanding because your objections are not ridiculous.

5) As for Gardiner, the problem is that with highlights and whatnot showing the same mistakes over and over people have a confirmation bias with him where every time they see him make a mistake, they remember it.

Like, it is absolutely ridiculous to make up the number 3 and say that 3 times per game he used to do a giveaway that lead to a goal or scoring chance. Ironically, Gardiner's best asset is his defensive play.

I go to this site everyday. I read anything Leafy. Then anything Tannery. Then anything Oilery because because misery loves company. I really enjoy your blogs James, even though I don't agree with you on politics. Enjoy your take on analytics. I've learned alot from your blogs. Keep up the good work.
- Procrastinator

Thanks, appreciated.

I think there is issues on both sides.

Analytics bloggers/twitter personalities have done the same thing they accuse others of doing- being close minded.

There is no 1000% correct way to rank players.

If you talk to anyone in Buffalo, Rasmus Ristolainen is an all star, on the cusp of Norris worthy defenseman. If you talk to anyone who uses analytics as a base, he's a below average player.

Both opinions are probably wrong.
- sbroads24

While we need to be more open minded with each other, it's a fine line because you don't need to be open minded to dumb ideas.

There are Republican ideas that I would listen to, but if someone tries to tell me Global Warming is not real, I am going to make fun of them.

Tanner I don't get it, who are you arguing with? Your arguments all depend on straw men or tiny minorities. This isn't a debate anymore. No one really questions the data. What gets questioned is context and how data relates to it. This is a much more relevant and difficult conversation to have but you never touch on it. You just slam "non believers" and tell everyone that if they don't let the data decide everything, they're morons.
- DoubleDown

Great post, I was going to post something similar, but I'll just add my 2 cents. Tanner, while I certainly agree stats have a place in hockey, your problem is that you see them as the be all and end all, which they are certainly not. I don't know if you've taken an actual statstics course, but the correlation factor of a statistic like corsi to team wins is no doubt far from 1, contrary to what you seem to believe, meaning there are undoutbley other factors to consider like shot quality for example. In addition, you cherry pick random statistics to support your arguments without offering the entire picture.

Comparing hockey analytics to politics is an amateurish joke.

Agreed on the post that analytics makes for boring tv.

People have different personal wants and needs and thus have different political preferences. Your self righteous attitude will prevent you from ever understanding this.
- Tumbleweed

I disagree. I am making a point about how we live in an environment where people don't change their mind when they learn a fact. They just turn to someone who won't challenge them and they end up reading and watching whoever will tell them what they want to hear.

I'd love for you to tell me anything I said that was 'self righteous'.

I disagree. I am making a point about how we live in an environment where people don't change their mind when they learn a fact. They just turn to someone who won't challenge them and they end up reading and watching whoever will tell them what they want to hear.

I'd love for you to tell me anything I said that was 'self righteous'.
- James_Tanner

The issue there is what is a fact, and what isn't. So many of your "facts" are really just an opinion. In using politics to make your point, you couldn't have chosen a worse comparison. In my opinion of course.

James, I think you're a good blogger, but you ridiculously overrate the analytics we as fans have access to. They just aren't that good. Use them as a tool to help you know what happened, but not much value for rating most players. They just aren't good enough. They should make up about 20% of the conversation.


What you are conflating, and don't understand is that there is a difference between the sound science of analytics, and the practical application and results when put into practice in a particular sport. Until they improve the data collection methods, analytics for a team sport like hockey will never be as valuable as you make them out to be. Now if we're talking about analytics that the teams have access to, that's a different story.
- MJL

1) Thanks. 2) No I don't. We have access to perfectly fine numbers. Especially when so much is based of shot-attempts which provide huge sample sizes (you score 3 goals in a game, maybe, but you take 75 or so shot attempts). Even if the numbers are off, they aren't off by enough to make a difference. 3) Suggesting an actual percentage of how much people should pay attention to analytics is ridiculous.

There are thousands of things going on in a hockey game. You can't remember them all and you are going to make biased judgements based on size, experience and reputation. Numbers can be blind to this stuff and thus will always give you better answers.

Oh, and the entire bottom paragraph you wrote is so ridiculous and overthetop as to not even be worth arguing about.

The issue there is what is a fact, and what isn't. So many of your "facts" are really just an opinion. In using politics to make your point, you couldn't have chosen a worse comparison. In my opinion of course.
- MJL

Love for you to quote an example, otherwise you're just making stuff up.

1) Thanks. 2) No I don't. We have access to perfectly fine numbers. Especially when so much is based of shot-attempts which provide huge sample sizes (you score 3 goals in a game, maybe, but you take 75 or so shot attempts). Even if the numbers are off, they aren't off by enough to make a difference. 3) Suggesting an actual percentage of how much people should pay attention to analytics is ridiculous.


There are thousands of things going on in a hockey game. You can't remember them all and you are going to make biased judgements based on size, experience and reputation. Numbers can be blind to this stuff and thus will always give you better answers.
- James_Tanner

The numbers we have are perfectly fine for limited use. To help show you what happened in a game, but they cannot tell you why. The problem is you don't use them in the limited manner they are best suited for. Where you go wrong is on the why part of the equation. The huge sample sizes don't remove the problem with data collection, and the team aspect of hockey. The difference between a good player and a bad player is not that much, to state that you can eliminate the difference that the numbers are off. I agree to a certain extent that stating an actual percentage of how much people should pay attention to analytics might be ridiculous, because it's a guesstimate. It is however about the percentage that NHL GM's use analytics in the formula for evaluating players.

If you don't think there is human bias in the analysis of analytics, you don't know them very well.

While we need to be more open minded with each other, it's a fine line because you don't need to be open minded to dumb ideas.

There are Republican ideas that I would listen to, but if someone tries to tell me Global Warming is not real, I am going to make fun of them.
- James_Tanner

Sure, but again it's on both sides.

I use stats as a basis for arguments on most occasions.

I have also had arguments with a certain blogger who does analytics for some teams, who basically said this player X wasn't good because of his WAR despite other stats pointing towards him being an effective player, yet then in a different argument, claimed he'd never use WAR as an end all be all.

Keep in mind, the reason for this argument was his clear bias towards a certain player.

Love for you to quote an example, otherwise you're just making stuff up.
- James_Tanner

I need to give you an example of how politics are partisan and corrupt. How a politician will make a decision because it favors his party, or the big money behind him, instead of what is the right thing to do? Seriously James, not going to discuss politics.

As far as analytics go, it's simple. If the numbers pointed to indisputable facts, then everyone looking at the numbers would come to the same conclusions on players. Your conclusions on players are not facts. They're opinions

A couple of things here: 1) shot attempts are a proxy for possession. You just need to get a rough idea of who helps give their team a positive shot differential. What you call it is unimportant.

2) No one bases their entire judgement on a players corsi and nothing else. This is a cheap straw man argument people use, but no one I talk to, follow or learn from does this.

3) The argument that "actual hockeying" goes into to making the stats is true, but I don't think it's a point against what I'm saying. You can't measure leadership, but if leadership helps, you'll see an uptick in things like shot differential when that player does his leading. You can't measure digging in the corners, but if it's an effective strategy it will show up in the shots.

4) all shots are not equal, but it doesn't matter because the bigger sample size you have the more accurate your projections will be. If you have a big enough sample size, you'll have an average shot difficulty, so it won't matter.

For example, in one game you might take fifty shots and only 2 of them were of high quality. This would be an anomaly because in a 10 games, or 100 or a 10000 there would work out to be an average amount (call it X) of good scoring chances in each game.

Admittedly, this is the kind of advanced stuff that limits understanding because your objections are not ridiculous.

5) As for Gardiner, the problem is that with highlights and whatnot showing the same mistakes over and over people have a confirmation bias with him where every time they see him make a mistake, they remember it.

Like, it is absolutely ridiculous to make up the number 3 and say that 3 times per game he used to do a giveaway that lead to a goal or scoring chance. Ironically, Gardiner's best asset is his defensive play.
- James_Tanner

This is a well thought out and explained post. A lot makes complete sense. The idea that a lot of hard to measure things such as leadership will be seen by an uptick in shot differential is a good point.

As for Gardiner, this season he actually seems to be lessening the number of mistakes (brainfarts) he makes from what I have seen. He's been great this year and I think Babcock has probably been a positive influence.

I just have a few thoughts about #4 though.

It seems to be predicated on the idea that teams are going to all have equal amount of quality shots, which might not necessarily be true.
What about the Wilson era Leafs that used to consistently out-shoot their opponents but had a massive amount of peripheral shots and seems to constantly "run into a hot goalie" even though they often faced backups?

Do you actually believe that if enough shots are taken by every player that some players won't have more quality shots than others? Or that some teams won't have more players capable of making better shots? If certain teams do get more quality shots this would be a (admittedly slight) variable missed by most shooting metrics.

Do you know if certain players consistently have higher shooting percentages (than the average)? This would probably answer my previous question, but I really have no idea.

You're so far in the other direction that you're exactly what you hate, but on the other side. You want people to be open and able to have meaningful discussions, but you're clearly unable to accept anything of value from the other side.

Teams use a ton of information and are constantly evaluating it. Teams that are able to do that more effectively are always at the top of the standings. It's why the top of the standings tend to look the same. However, not all information can be represented by an easy to digest number. Why is it so much better to make decisions on one set of information while arbitrarily ignoring other information?

You keep trying to make real world comparisons, but the real world doesn't work that way either. Why are job interviews and references a thing if everyone should just be analyzing data to make their decisions. You learn a ton about someone from a job interviews that can't be learned from looking at data. Why are references a thing if how a person fit in to their last work environment shouldn't matter at all?

Shot attempts are as valuable or useless as +/- because they both are affected by they other 11 players on the ice and have no way to look at the context of the situation. It's not always the best choice to shoot. Face offs aren't always important, but when they are, they're extremely important.

You're so far in the other direction that you're exactly what you hate, but on the other side. You want people to be open and able to have meaningful discussions, but you're clearly unable to accept anything of value from the other side.

Teams use a ton of information and are constantly evaluating it. Teams that are able to do that more effectively are always at the top of the standings. It's why the top of the standings tend to look the same. However, not all information can be represented by an easy to digest number. Why is it so much better to make decisions on one set of information while arbitrarily ignoring other information?

You keep trying to make real world comparisons, but the real world doesn't work that way either. Why are job interviews and references a thing if everyone should just be analyzing data to make their decisions. You learn a ton about someone from a job interviews that can't be learned from looking at data. Why are references a thing if how a person fit in to their last work environment shouldn't matter at all?

Shot attempts are as valuable or useless as +/- because they both are affected by they other 11 players on the ice and have no way to look at the context of the situation. It's not always the best choice to shoot. Face offs aren't always important, but when they are, they're extremely important.
- Hunkulese

Great post. You get it.

This is a well thought out and explained post. A lot makes complete sense. The idea that a lot of hard to measure things such as leadership will be seen by an uptick in shot differential is a good point.

As for Gardiner, this season he actually seems to be lessening the number of mistakes (brainfarts) he makes from what I have seen. He's been great this year and I think Babcock has probably been a positive influence.

I just have a few thoughts about #4 though.

It seems to be predicated on the idea that teams are going to all have equal amount of quality shots, which might not necessarily be true.
What about the Wilson era Leafs that used to consistently out-shoot their opponents but had a massive amount of peripheral shots and seems to constantly "run into a hot goalie" even though they often faced backups?

Do you actually believe that if enough shots are taken by every player that some players won't have more quality shots than others? Or that some teams won't have more players capable of making better shots? If certain teams do get more quality shots this would be a (admittedly slight) variable missed by most shooting metrics.

Do you know if certain players consistently have higher shooting percentages (than the average)? This would probably answer my previous question, but I really have no idea.
- Thecakeisalie

The shot quality argument will probably last as long as the NHL. Even among guys who are both quality analytics analysts (like, guys who might eventually get jobs doing this stuff for the NHL; not me) there are disagreements.

Certain players do take less shots, but more sure-thing shots. (Bozak maybe) and certain guys just take tons of shots (Ovechkin).

I think even for a team like Wilson's Leafs, they probably talked about that more than it actually happened, but that kind of strategy is anomalous if you looked at every team.

There will always be exceptions, and there will always be anomalies. I tend to think you can get a pretty good idea about things without worrying about shot-quality.

You're so far in the other direction that you're exactly what you hate, but on the other side. You want people to be open and able to have meaningful discussions, but you're clearly unable to accept anything of value from the other side.


This is just not true. What side? People who laugh at the idea of using analytics? People who use the Coyotes' current record, or the Panthers, to judge a whole movement? Those people are ridiculous and deserve no respect.

If you want to put forth some objections and have a conversation - well I think this entire comments section is filled with people doing that, and no one seems to be getting mad or making crazy statements ,so I'm not sure what exactly you mean here.



Teams use a ton of information and are constantly evaluating it. Teams that are able to do that more effectively are always at the top of the standings. It's why the top of the standings tend to look the same.

Not because Pittsburgh has Malkin/Crosby/Letang and Chicago has Kane/Hossa/Toews/Keith and LA has Koptar/Carter/Doughty?

However, not all information can be represented by an easy to digest number. Why is it so much better to make decisions on one set of information while arbitrarily ignoring other information?

You keep trying to make real world comparisons, but the real world doesn't work that way either. Why are job interviews and references a thing if everyone should just be analyzing data to make their decisions. You learn a ton about someone from a job interviews that can't be learned from looking at data. Why are references a thing if how a person fit in to their last work environment shouldn't matter at all?

Shot attempts are as valuable or useless as +/-


Objectively Wrong. Shot-attempts give you up to 20 x the sample size in a single game vs plus/minus.


because they both are affected by they other 11 players on the ice and have no way to look at the context of the situation. It's not always the best choice to shoot. Face offs aren't always important, but when they are, they're extremely important.
- Hunkulese

Great post. You get it.
- MJL

Did you honestly read that post? It was anything but great haha.