News Article: "Fun With Numbers" - Advanced Stats Talk Here

[BonkTastic]

Well, it's official: the team has acknowledged the existence of Advanced Stats.

Article can be found here, by Chris Lund:
Fun With Numbers: A Stats Rundown

It's a pretty good read if you're not familiar with the terms, worth the 3 mins it'll take to skim through it.

Click to read more...

 

[Caeldan]

Well, it's official: the team has acknowledged the existence of Advanced Stats.

Article can be found here, by Chris Lund:
Fun With Numbers: A Stats Rundown

It's a pretty good read if you're not familiar with the terms, worth the 3 mins it'll take to skim through it.

Looking forward over the long term, these numbers will likely serve two purposes for you:

1) They'll confirm a lot of suspicions you had about the game. If we went back and figured it out, I'm willing to bet Wayne Gretzky had an excellent Fenwick over the course of the 1985-86 season. That wouldn't be a surprise. Good players are good players. Good teams are good teams.

2) They'll make you raise your eyebrows and change the way you look at the game. If these numbers were discovered to simply reaffirm what we had been told about the sport they would be pointless. Prepare to read pieces that challenge your opinions and see lines of numbers that confound how you perceived a sequence within a game. Good players are good players and good teams are good teams, but how you perceive what qualifies as either could change permanently, ideally for the better.

Did my quick skim but wanted to comments on the concluding remarks...

1) I know people have gone back and done this already - but not in a spot right now to go look up and confirm this. Just curious whether the writer's done basic homework or not before making an 'educated guess'.

2) Really hope that the writer isn't going to try to push these as out of context replacement stats that are the be all and end all of how to quantify a player, as some are apt to do.

 

[BonkTastic]

2) Really hope that the writer isn't going to try to push these as out of context replacement stats that are the be all and end all of how to quantify a player, as some are apt to do.

As a proponent of advanced stats, I agree wholeheartedly.

The key is to add more tools to your analytical toolbox. Pushing out some traditional stats is as bad as ignoring some advanced ones, IMO.

 

[Quo]

Lund is good people, not out to make it the be all, end all. Look forward to this.

Even the most unabashed analytics homer would (or should) acknowledge the need to have additional information beyond what can be derived from hours of labour in Microsoft Excel. At the end of the day, an NHL locker room is a workplace and, like any workplace, there is more to creating a team than assembling a group of people who hit their numbers.

 

[Six Assets]

good read

 

[Sun God Nika]

I wish i could understand advanced stats . ill stick to my plus minus and goals for and against.

 

[Nac Mac Feegle]

When it comes to the advanced stuff, I tone I like the most is the visual showing where shots are taken on the ice, and which of those shots beat the goaltender. To me, that give a really easy visual representation of the game (a lot of shots form the boards or perimeter indicate low scoring opportunities and a team succeeding in making a defensive shell, while a ton of shots form the slot show a real mismatch between teams).

A good read though. More information is always a good thing.

 

[Cosmix]

I doubt that the future articles will be very comprehensive as I don't think the Senators would want to focus too much on the performance of some of their players on some of the advanced stats assessments.

 

[StefanW]

I wish i could understand advanced stats . ill stick to my plus minus and goals for and against.

"Advanced" stats in hockey are not all that advanced from a stats point of reference. It is all about counting things, and then dividing them out. For example, if a player puts 10 shots at the opposing net while the other teams put 8 shots on net while he is on the ice the Corsi works out to 10(his shots)/18(total shots for both side while he is on the ice)=55.56%.

Sometimes you don't count blocked shots, and sometimes you standardize by figuring out what things work out to in a 60 minute game, but it doesn't go too far beyond that at this stage.

The idea is that the other team can't score when you have the puck, and when you have the puck more you end up directing more shots at the other team's net. Over the course of time the teams who have the puck more tend to win more often that teams that don't have the puck as much as their opposition.

There is a lot of common sense behind it, and it really revolves around just counting stuff that you see on the ice. "Advance stats" as a term really only makes sense in comparison to the older stats, which focus on goals, assists, +/-, save %, and things like that. I think sometimes people see "advanced" and "stats" together and assume that it is something only brainiacs with computers and calculators can figure out, or they think the people who use stats focus on numbers instead of watching the games.

 

[BonkTastic]

There is a lot of common sense behind it, and it really revolves around just counting stuff that you see on the ice. "Advance stats" as a term really only makes sense in comparison to the older stats, which focus on goals, assists, +/-, save %, and things like that. I think sometimes people see "advanced" and "stats" together and assume that it is something only brainiacs with computers and calculators can figure out, or they think the people who use stats focus on numbers instead of watching the games.

Yeah, a stat like Corsi isn't really any more complicated a stat than, say... Save Percentage.

It's the same math, you're just counting different things.

 

[Knave]

Lund is good people, not out to make it the be all, end all. Look forward to this.

That's good.

And as for advanced stats being too complicated - the only complicated bit is the name. Once you read the definition of what the word(s) represent, it's easy to understand.

 

[BigRig4]

So if someone has really rough zone starts (Zack Smith for example) it gets him off the hook for a bad corsi?

 

[Six Assets]

"Advanced" stats in hockey are not all that advanced from a stats point of reference. It is all about counting things, and then dividing them out. For example, if a player puts 10 shots at the opposing net while the other teams put 8 shots on net while he is on the ice the Corsi works out to 10(his shots)/18(total shots for both side while he is on the ice)=55.56%.

Sometimes you don't count blocked shots, and sometimes you standardize by figuring out what things work out to in a 60 minute game, but it doesn't go too far beyond that at this stage.

The idea is that the other team can't score when you have the puck, and when you have the puck more you end up directing more shots at the other team's net. Over the course of time the teams who have the puck more tend to win more often that teams that don't have the puck as much as their opposition.

There is a lot of common sense behind it, and it really revolves around just counting stuff that you see on the ice. "Advance stats" as a term really only makes sense in comparison to the older stats, which focus on goals, assists, +/-, save %, and things like that. I think sometimes people see "advanced" and "stats" together and assume that it is something only brainiacs with computers and calculators can figure out, or they think the people who use stats focus on numbers instead of watching the games.

nice post

I also though Corsi was some compacted stat before I looked into it.

 

[BonkTastic]

So if someone has really rough zone starts (Zack Smith for example) it gets him off the hook for a bad corsi?

Doesn't totally get him off the hook, but it does at the very least bear consideration.

Bad zone starts + good corsi = very likely to be a very good defensive player, or at the very least good at converting a defensive situation into an offensive one.

Bad zone starts + bad corsi = could be a lot of things...

 

[Icelevel]

I like Chris Lund. Good read

 

[StefanW]

So if someone has really rough zone starts (Zack Smith for example) it gets him off the hook for a bad corsi?

If you google zone start adjustments in hockey you will find a few different options for dealing with that sort of thing. There is agreement that a mechanism is needed to adjust for zone start effects, but there is some disagreement about what type of mechanism to use.

Just an FYI, but some of the cool work on this was done by Tyler Dellow, and his blog was pulled down last night after he accepted the position with Edmonton. So there may be some gaps when you start to research this.

This is a good starting point:
http://nhlnumbers.com/2012/11/5/zone-start-adjusted-corsi-corrections-faceoffs

 

[Oil For The Boys]

If we're using Fenwick and Corsi as a stand-in for possession, why don't we just have a statistic for possession? Similar to the way they have it for soccer as well as pass completion %. I understand how the advanced statistics work, I just don't under how it tells me anything differently then if I just watch the players/game. Just seem kinda useless to me.

I like the offensive and defensive zone start percentages though.

 

[Micklebot]

If we're using Fenwick and Corsi as a stand-in for possession, why don't we just have a statistic for possession? Similar to the way they have it for soccer as well as pass completion %.

Tracking possession (without an automated system like sportvu) is very labour intensive. Corsi correlates very strongly with possession and it's component stats are already tracked. It makes doing the legwork to track possession practically useless.

I understand how the advanced statistics work, I just don't under how it tells me anything differently then if I just watch the players/game. Just seem kinda useless to me.

Watching the game is ideal, but subject to bias and not practical when your trying to compare large samples (ie 7 years worth of games where your comparing a player or team to league average, about 9000 hours worth of viewing to get the necessary data).

I like the offensive and defensive zone start percentages though.

Personally, I prefer a version relative to teammates. The effect of offensive zone starts is pretty minor, as has been demonstrated by comparing corsi and goals based metrics for 5vs5 play and and 5 vs5play minus the first ~10 seconds after a faceoff (most research has shown the advantage gained by a faceoff win is negated after ~10 secs of play). A relative zone start stat would at least give a better representation of how the coach views the player.

 

[StefanW]

Not directly related to the OP, but this hockey stats related story by Elliotte Friedman is pretty amazing:
http://www.cbc.ca/sports-content/hockey/opinion/2014/08/in-search-of-vic-ferrari.html

Note to Mods: Is it possible to have a hockey analytics thread here? The topic keeps coming up is different discussions, and it seems important as a Sens fan in particular (or hockey fan in general) to have some type of discussion about analytics and/or venue to post and comment on analytics stories that keep surfacing.

 

[Quo]

Not directly related to the OP, but this hockey stats related story by Elliotte Friedman is pretty amazing:
http://www.cbc.ca/sports-content/hockey/opinion/2014/08/in-search-of-vic-ferrari.html

Note to Mods: Is it possible to have a hockey analytics thread here? The topic keeps coming up is different discussions, and it seems important as a Sens fan in particular (or hockey fan in general) to have some type of discussion about analytics and/or venue to post and comment on analytics stories that keep surfacing.

We have an entire forum for that kind of stuff Stef: http://hfboards.mandatory.com/forumdisplay.php?s=&daysprune=-1&f=241

Hockey Analytics... the Final Frontier. Explore strange new worlds, to seek out new algorithms, to boldly go where no one has gone before.

If you'd like something for just Sens related adv. stat analysis/discussion though, that would be different, wouldn't be opposed, but I'm just one of five. Not sure how an all-purpose thread like that would shake out practically speaking but no harm in trying.

 

[Ouroboros]

In the last 18 months I've given up watching the games entirely and now rely solely on analytics to enjoy and understand the game. I've never enjoyed hockey more.

 

[DrunkUncleDenis]

In the last 18 months I've given up watching the games entirely and now rely solely on analytics to enjoy and understand the game. I've never enjoyed hockey more.

Interesting.

I was going to offer you a , but I feel like you're being serious.

 

[StefanW]

We have an entire forum for that kind of stuff Stef: http://hfboards.mandatory.com/forumdisplay.php?s=&daysprune=-1&f=241

Yeah, I check that out now and again. It is far more fun and interesting to discuss this with other Sens fans.

 

[StefanW]

In the last 18 months I've given up watching the games entirely and now rely solely on analytics to enjoy and understand the game. I've never enjoyed hockey more.

True story for you. During the interwar era it was not uncommon for people to be baseball fans and to track their teams even if they never actually went to a game. And this was before the age of television. Some listened on the radio, while others just went by newspaper stories and tracking stats.

Never underestimate the power of tracking numbers in sports.

 

[Caeldan]

We have an entire forum for that kind of stuff Stef: http://hfboards.mandatory.com/forumdisplay.php?s=&daysprune=-1&f=241



If you'd like something for just Sens related adv. stat analysis/discussion though, that would be different, wouldn't be opposed, but I'm just one of five. Not sure how an all-purpose thread like that would shake out practically speaking but no harm in trying.

Well this thread is as good as any to start - if there's something Ottawa Senators analytics related, I'd say it's within topic based on the original post.