Hockey Analytics

[That Aud Smell]

Posted on Twitter as a "FWIW" compare/contrast on how (parts of) the NBA and NHL are dealing with analytics.

 

 

[SwampD]

-"Defensive defenseman" is simply a term used by people to describe a bad defenseman in a way that gives value to that player.

 

 

What if you have two defensemen, with identical PDOs, but one has an abormally high oiSV% and the other has a really high oiSH%?

 

Wouldn't you say that the one with the high oiSV% is a defensive defenseman? I have to believe that there are some other comparisons with Corsi or Fenwick #s that could be used to make a case, as well.

 

Using this to make my Franson/McCabe argument, Their numbers are very similar, but I would look at McCabe as more of an offinsive Dman and Franson as more of a defensive Dman and that they both have value in different game situations.

[SwampD]

So, I'm just dipping my toe in the chilly waters of advanced stats.

 

This is where I went looking.

 

http://www.hockey-reference.com/teams/BUF/2017.html

 

Maybe I'm missing something, but I think there needs to be a stat that tracks zone entries against that result in no shots (oiZEASH%?) . It may sound boring, but I think that would be a good measure of a players defense.

Edited January 6, 2017 by SwampD

[pi2000]

So, I'm just dipping my toe in the chilly waters of advanced stats.

 

This is where I went looking.

 

http://www.hockey-reference.com/teams/BUF/2017.html

 

Maybe I'm missing something, but I think there needs to be a stat that tracks zone entries against that result in no shots (oiZEASH%?) . It may sound boring, but I think that would be a good measure of a players defense.

Sounds interesting, how about a stat that tracks net goals for vs goals against at even strength? Goals are what decide games, not shots.

[MattPie]

Sounds interesting, how about a stat that tracks net goals for vs goals against at even strength? Goals are what decide games, not shots.

 

But not an individual player's defensive ability.

[SwampD]

Sounds interesting, how about a stat that tracks net goals for vs goals against at even strength? Goals are what decide games, not shots.

:lol: You mean +/-, don't you?

 

I push back as hard as anyone on AS and even I know that that stat in a vacuum is horribly deficient.

[That Aud Smell]

How do Marner's 32 points in 39 NHL games compare with Eichel's 70 points in 99 games?

 

If we tanked an entire season, then picked the wrong guy at 2, that would be bad, right?

 

And we could blame ownership for that, right?

 

I've not been watching Marner, so I can't say. He does look to be a dymanic winger, if maybe a bit smallish.

 

Eichel was the consensus #2 guy that year, and for good reason, I think.

 

In terms of blaming ownership for picking one player over the other, I'm not sure what the thinking there would be. I don't think anyone disputes that GM TM and his team decided who to pick at #2.

 

People are free to "blame ownership" if the team doesn't evolve into a contender in the next 2-3 years -- if only because, as owners, they're ultimately responsible for the team. It does seem that Terry saves most of his operations meddling for the Bills.

Edited January 9, 2017 by That Aud Smell

[Brawndo]

How do Marner's 32 points in 39 NHL games compare with Eichel's 70 points in 99 games?

 

If we tanked an entire season, then picked the wrong guy at 2, that would be bad, right?

 

And we could blame ownership for that, right?

A few points on that. Marner has consistently had much better linemates having spent time with Matthews, Nylander, Kadri and JVR.

 

Jack's linemates were constantly shuffled last season and he was not placed with top end talent until towards the end of the season

 

He also has a coach who knows how to deploy him more effectively as a rookie, compared to Jack last season.

 

Finally adapting to the NHL while playing wing is much easier then learning the center position.

 

The Sabres had been devoid of centers for years, so GMTM's plan to acquire them was a good one.

 

Marner is an outstanding prospect and may turn out to be more productive then Jack, but not a single draft pundit and probably not a single GM would have advocated taking Marner over Eichel at the second pick.

[WildCard]

Eichel is better than Marner, some as that

Babcock just blows Bylsma out of the damn water

[Randall Flagg]

Given the Sabres' need for a franchise center, how important that position is to playoff success, and what Eichel looks like on the ice at his age, I don't think Marner will ever reach a height that makes us think we should have taken him. If he wins a couple Art Rosses, the gap between the scoring leader and the next few guys (which I expect Eichel to be in) has been so low recently that I will assume Eichel also has a shot any given year. He'd need to start winning scoring races by 30-40% over guy number two for any of us to second-guess, IMO, and that just doesn't happen anymore. 

 

He's a freaky-good player and we will be able to appreciate that without any "what if"s, I'm quite confident about that. 

[Doohicksie]

How do Marner's 32 points in 39 NHL games compare with Eichel's 70 points in 99 games?

 

If we tanked an entire season, then picked the wrong guy at 2, that would be bad, right?

 

And we could blame ownership for that, right?

 

Past performance does not guarantee future results.  You can apply that twice in this situation:  Once when evaluating pre-draft performance to NHL performance for Eich and Marner, and another would be the comparison of their performance to date in the NHL to how they will compare going forward. 

 

There are other considerations as well (as has already been stated):  the Sabres were more in need of center depth; and center is a harder position to adjust to in the NHL than wing is.

 

And in general, you simply can't predict anything related to player performance with 100% accuracy.  You can't look at one player putting up .71 points per game selected ahead of another putting up .82 points per game (in about a third as many games) and say that you clearly should have picked the second guy ahead of the first.

 

Also, consider that Jack was "thrown to the wolves" so to speak, starting in the NHL last year, while Marner spent an extra year in juniors to mature.

[MattPie]

:angel: These are all reasonable points and I appreciate it.  It makes me feel better about being a Sabres fan.  

 

It's always unfortunate to see someone "thrown to the wolves," so to speak, but sometimes it's what has to be done.

 

These are the points, distilled. Clip and save when talking to the Leafs fan in you life:

- Marner is playing wing and with better players.

- Eichel was on a terrible NHL team last year and still scored reasonably well.

- Eichel is on a bad NHL team this year and is scoring reasonably well.

- Marner is playing on a team that emphasizes scoring. There's no way Marner would score as many in Buffalo, and no way Eichel wouldn't score more in Toronto.

- Centers are more valuable than just points.

- Intangibles! :)

[TrueBlueGED]

Marner would not have 32 points in 39 games playing for Bylsma, of that I am confident.

[Taro T]

Marner would not have 32 points in 39 games playing for Bylsma, of that I am confident.

Fairly certain Matthews wouldn't either. ;)

[Drunkard]

Marner would not have 32 points in 39 games playing for Bylsma, of that I am confident.

 

He wouldn't even be in the NHL yet. He'd still be in juniors followed by a year in Rochester where he'd be ready to learn the fine art of grinding them down.

[TrueBlueGED]

What if you have two defensemen, with identical PDOs, but one has an abormally high oiSV% and the other has a really high oiSH%?

 

Wouldn't you say that the one with the high oiSV% is a defensive defenseman? I have to believe that there are some other comparisons with Corsi or Fenwick #s that could be used to make a case, as well.

 

Using this to make my Franson/McCabe argument, Their numbers are very similar, but I would look at McCabe as more of an offinsive Dman and Franson as more of a defensive Dman and that they both have value in different game situations.

Two things.

 

1) My comment was referring to how fans/media label players, not meant to mean that no players add defensive value. I think there's a tendency to label any and all defensemen who don't contribute offensively as defensive defensemen, without any actual evaluation of their defensive game.

 

2) Research has been done on this, and there has been no reputable finding that a defenseman can reliably, consistently, and repeatedly impact the save percentage of his goaltender. There are a few people like Dave Johnson who thinks they can, but they're basically the climate change deniers of the analytics community. This is a major reason there's such a focus on possession*, and before anyone asks, part of this is difficulty in attributing the save percentage of a goalie to one of five skaters on the ice, combined with how naturally variable save percentage itself is and how infrequent goals are.

 

*When analytics types talk about possession, what they really mean is the aggregate of shot generation and shot suppression. There is evidence individual defensemen are better shot suppressors than others, but zilch had been found on the save percentage front. So if we're looking for defensive value, on-ice shot attempts against is the way to go with our current evidence. I think this makes some sense--you'd think a defenseman who has quality gap control and positioning would be on the ice for fewer shots against.

So, I'm just dipping my toe in the chilly waters of advanced stats.

 

This is where I went looking.

 

http://www.hockey-reference.com/teams/BUF/2017.html

 

Maybe I'm missing something, but I think there needs to be a stat that tracks zone entries against that result in no shots (oiZEASH%?) . It may sound boring, but I think that would be a good measure of a players defense.

Stats like this get referred to as "microstats" and there's a lot of it out there, much if it snake oil (the Oilers have used some of it to justify Kris Russell's value, even though the end results are terrible), but some are doing well with it. Ryan Stimson at hockey-graphs.com is a solid writer who has delved into the field.

 

Anyway,even if a player adds defensive value, that doesn't inherently mean they generate actual value if they're a black hole offensively. If a Dman has legitimate in-zone defensive value, but spends 75% of his time there, is that actually a net benefit? I think no. (Hi Gorges!)

[MattPie]

Two things.

 

1) My comment was referring to how fans/media label players, not meant to mean that no players add defensive value. I think there's a tendency to label any and all defensemen who don't contribute offensively as defensive defensemen, without any actual evaluation of their defensive game.

 

2) Research has been done on this, and there has been no reputable finding that a defenseman can reliably, consistently, and repeatedly impact the save percentage of his goaltender. There are a few people like Dave Johnson who thinks they can, but they're basically the climate change deniers of the analytics community. This is a major reason there's such a focus on possession*, and before anyone asks, part of this is difficulty in attributing the save percentage of a goalie to one of five skaters on the ice, combined with how naturally variable save percentage itself is and how infrequent goals are.

 

*When analytics types talk about possession, what they really mean is the aggregate of shot generation and shot suppression. There is evidence individual defensemen are better shot suppressors than others, but zilch had been found on the save percentage front. So if we're looking for defensive value, on-ice shot attempts against is the way to go with our current evidence. I think this makes some sense--you'd think a defenseman who has quality gap control and positioning would be on the ice for fewer shots against.

 

It'd be interesting to bring shot blocks into this discussion too. A D that has lots of shot blocks per opponent shot but also sucks at shot suppression is breaking even at best. Makes me wonder how Jay McKee would have faired.

[TrueBlueGED]

An example of some of the microstat tracking out there, conveniently from the Sabres/Canes game: https://theenergyline.wordpress.com/2017/01/14/car-5-buf-3-1172013/

 

Note the zone entry stats, and how much more likely Carolina was to carry it in compared to us:

 

 

Anyone really want to tell me that Ryan O'Reilly isn't talented enough to carry it in more frequently than Brett Pesce?

[Sabres Fan in NS]

TrueBlue,

 

Maybe I'm doing this wrong, but I counted 35 carry ins for the Canes and 30 for the Sabres.  Not that big a difference in my mind.

 

Also, our top 3 carry iners had more then their top 3 ... 16 vs. 12.

 

What am I missing here Mr. Fancystats?

Edited January 14, 2017 by Sabres Fan In NS

[TrueBlueGED]

TrueBlue,

 

Maybe I'm doing this wrong, but I counted 35 carry ins for the Canes and 30 for the Sabres.  Not that big a difference in my mind.

 

Also, our top 3 carry iners had more then their top 3 ... 16 vs. 12.

 

What am I missing here Mr. Fancystats?

 

The ratio of dump-ins to carry-ins. Carolina carried it in 53% of the time, we only did so 46% of the time. Also, the asymmetric distribution of dumps vs carries down the roster is something I had hoped people would pick out: our carry-in percentage was largely driven by a few players, whereas their's is fairly consistent even among their secondary players. In fact, they have secondary players carrying it in more frequently than some of our top players. Speaks to coaching and system.

[Sabres Fan in NS]

:unsure:

 

No offense, but you are not convincing me that there was that big a difference between these two teams in that game.

[darksabre]

:unsure:

 

No offense, but you are not convincing me that there was that big a difference between these two teams in that game.

Carolina's best players carry the puck into the zone more than our best players. That's a problem. Look at Okposo for instance. There's no way a guy like Okposo should be dumping the puck in that much. He's a talented player giving up possession. 

 

It's better to not enter the zone at all than to enter it with a dump in. A possession team shouldn't be dumping the puck on zone entries more than it carries the puck. Better to abort the zone entry or go offsides than to give the puck back to the other team. 

[TrueBlueGED]

:unsure:

 

No offense, but you are not convincing me that there was that big a difference between these two teams in that game.

 

Well, there wasn't a massive difference last night (it was actually one of our better possession games on the season), but I do think there was a noticeable difference.

 

That wasn't my main objective anyway. I wanted to do two things:

 

1) Show Swamp that the kind of stuff he wants out of "fancystats" does, in fact, exist. Despite what its critics constantly harp on, the field has moved well beyond simply Corsi. The full article even has the exact statistic he was looking for about zone entry attempts against specific defensemen.

 

2) Begin (after all, I would never suggest we should draw firm conclusions based on a single game) an empirical refutation of the notion that we simply don't have the talent to carry the puck in more frequently. Carolina has a bunch of nobodies carrying it in more frequently than some of our best players. We can carry it in more than we do, and we can generate more offensive chances than we do...but there's a systemic problem preventing us from doing so.

Carolina's best players carry the puck into the zone more than our best players. That's a problem. Look at Okposo for instance. There's no way a guy like Okposo should be dumping the puck in that much. He's a talented player giving up possession. 

 

It's better to not enter the zone at all than to enter it with a dump in. A possession team shouldn't be dumping the puck on zone entries more than it carries the puck. Better to abort the zone entry or go offsides than to give the puck back to the other team. 

 

Their mediocre players were doing it too!

[Brawndo]

Interesting that Zemgus was carrying the puck in second most on the team and had fewer dump ins then Jack and Sam. Maybe Zemgus is on to something by not being adaptable to Dreary Dan's System?

[Stoner]

What motivates the analytics bloggers? Looking for a job in the NHL? Trying to change how the game is played? Just a hobby? A fad? Something else?