Ongoing Stats and Analytical Discussion Thread: Battle of the Defense

[devilsblood]

The reason Fenwick and Corsi are used is because they give us the largest sample size possible. So sure, there's some nonsense counted in there, but many shots on goal have little to no chance of going in the net either - it's a sport where around 9% of the shots on goal are goals. In addition, people have counted scoring chances for ages and for the most part they have mirrored Fenwick and Corsi.

The danger with something like high-danger scoring chances is A: the definition of them - as you narrow your sample size, NHL scorers are bound to miss more important events (including goals), whereas missing individual Fenwicks and Corsis is less important and B: the repeatability of them. Part of why we like Fenwick and Corsi is that over smaller samples, it's more reliable at predicting future goal rates than goal rates themselves. If high-danger scoring chances show similar randomness to goal rates, they're not particularly useful other than measuring past performance. I haven't really seen anyone take a look at high-danger scoring rates yet.

But was that not his point? Flubbed shots, or shots from non scoring areas do not go into the net that often. Where as a one timer from inside the dots tend to go into the net at a higher then average rate.

How bout shot trackers, like Corsi within dot's, does that exist?

 

[devilsblood]

They're alright, but pretty useless without context. If I were to just pull out Defenseman A from Team 1 and compare him with Defenseman B from Team 2 there's really nothing you can gain from looking at a HERO chart. People think they're the end all of stats, they're really just a quick fix for casual fans who don't understand math.

A lot of times player possession stats end up being driven by team possession stats, not the other way around. On a HERO chart a guy like Henrique has 4th line possession numbers - obviously he's not a 4th liner, but our shots for/shots against numbers are so bad that anyone who gets a lot of minutes on this team is getting bombarded with negative possession. Likewise, a guy like Brandon Dubinsky, who is a pretty average middle six player, has monster 1st line possession numbers because he plays on a team with absurd shot differential. You have to be really careful about comparing guys on different teams.

They're really there for confirmation bias. If you think a guy is good and the HERO chart agrees, then your eye test was probably right. But when people just post them in threads with no context it's really annoying - a lot of things factor into a HERO chart that actually aren't displayed on a HERO chart.

I saw the issue with Rico is he was typically matched up against competition above his level, if we had a true #1, and Zajac then got some tough assignment minutes, his hero, and Corsi for that matter would look much better.

Conversely I remember we had 4th liners, like Gio I think, with pretty good heros, but of coarse he was playing against opposing 4th liners.


Used the hero's a bunch for a bit, but then lost interest.

 

[Triumph]

But was that not his point? Flubbed shots, or shots from non scoring areas do not go into the net that often. Where as a one timer from inside the dots tend to go into the net at a higher then average rate.

How bout shot trackers, like Corsi within dot's, does that exist?

If you read my post - you know, the sentence after the one you bolded - you would see this sentence: "In addition, people have counted scoring chances for ages and for the most part they have mirrored Fenwick and Corsi." Which is true, to the extent that people in the stats community don't count them anymore. So there were shot trackers until they realized that tracking shots is pointless. What you're looking for is something like Ryan Stim.son's passing project, which is yielding some interesting results.

 

[AfroThunder396]

Saw the conversation in the team thread about referees, and since this the analytics thread I figure I would put this here.

Talking about the missed boarding calls on Zajac, Moore, and Zacha, AND getting several majors called against us that were not majors, AND the Price/Palmieri fiasco. I generally look at officiating as the kind of thing that normalizes out over the course of a season (you have favorable nights and unfavorable nights and it tends to even out over 82 games). But then I saw this:



Basically, ever since the Wideman incident the Flames have had dramatically more penalties called on them, outside the norms of game-to-game fluctuation. Maybe the tin foil hats aren't so crazy after all. Just because you're paranoid doesn't mean they aren't out to get you.

 

[Bleedred]

Saw the conversation in the team thread about referees, and since this the analytics thread I figure I would put this here.

Talking about the missed boarding calls on Zajac, Moore, and Zacha, AND getting several majors called against us that were not majors, AND the Price/Palmieri fiasco. I generally look at officiating as the kind of thing that normalizes out over the course of a season (you have favorable nights and unfavorable nights and it tends to even out over 82 games). But then I saw this:



Basically, ever since the Wideman incident the Flames have had dramatically more penalties called on them, outside the norms of game-to-game fluctuation. Maybe the tin foil hats aren't so crazy after all. Just because you're paranoid doesn't mean they aren't out to get you.

There was a thread on the mainboard recently about Flames management meeting with the league over this.

 

[Triumph]

Part of this I do think is player-related - people like Alex Chiasson and Matthew Tkachuk probably take more penalties than the guys they replaced. But yeah, more and more things keep indicating that e.g. 'home ice advantage' is largely referee bias in favor of the home team.

 

[devilsblood]

Part of this I do think is player-related - people like Alex Chiasson and Matthew Tkachuk probably take more penalties than the guys they replaced. But yeah, more and more things keep indicating that e.g. 'home ice advantage' is largely referee bias in favor of the home team.

There has to be #'s on this? Do home teams typically enjoy more penalties for?

 

[Zippy316]

There has to be #'s on this? Do home teams typically enjoy more penalties for?

I'd imagine this has a lot to do with the pressure home fans put on the refs at times.

Montreal's crowd practically makes their own penalty calls at times. When Devils have a packed house, they can get on the referees pretty well. I'd imagine the crowd makes light of instances to the refs where they might not have otherwise seen them and then they respond by calling penalties later on.

 

[Jason MacIsaac]

I would love to see analytics of Blandisi's passing in the offensive zone. The amount of HD scoring chances he creates 5 on 5 is quite impressive. I think he and Cammalleri need to be played together.

 

[Zippy316]

I would love to see analytics of Blandisi's passing in the offensive zone. The amount of HD scoring chances he creates 5 on 5 is quite impressive. I think he and Cammalleri need to be played together.

Cammmalleri - Blandisi - Noesen
Wood - Josefson - DSP

Would be a nice bottom-six. Wood would add some offensive dynamic to Josefson and open up space and Blandisi gets some skilled players to play with.

 

[devilsblood]

I'd imagine this has a lot to do with the pressure home fans put on the refs at times.

Montreal's crowd practically makes their own penalty calls at times. When Devils have a packed house, they can get on the referees pretty well. I'd imagine the crowd makes light of instances to the refs where they might not have otherwise seen them and then they respond by calling penalties later on.

But is there an actual difference in penalties called for the home team? And even more so for MTL at home?

 

[Jason MacIsaac]

Since switching to center Blandisi has the following ratio's playing with Wood and DSP.

SC% 55%
HDSC% 50%

This is against NYI x2, Ottawa and NYR. I think we need to start looking at Blandisi as the solution to one of our top 9 center options moving forward.

 

[VoidCreature]

Since switching to center Blandisi has the following ratio's playing with Wood and DSP.

SC% 55%
HDSC% 50%

This is against NYI x2, Ottawa and NYR. I think we need to start looking at Blandisi as the solution to one of our top 9 center options moving forward.

Where did you find that data? I'd love to have a look myself.

 

[Emperoreddy]

Necro bumping this.

 

[Devils731]

In reply to BG

You’re polluting a robust data set of 5 on 5 play over a whole season with man power difference data which isn’t part of the predictor.

The supposition is that Corsi advantaged 5 on 5 play generally leads to wins over the course of a season. We see this borne out by comparing standings to Corsi and it does generally match.

 

[devilsblood]

Here's my logic.

p1) 5v5 corsi works
p2) all situations does not work
p3) I like things that work

Conclusion: I like 5v5 corsi.

 

[Bleedred]

''Who cares about this shit? It's all BS anyway!'' - John Hynes

Just kidding haha. Kind of.

Since we're on the topic of it, guess what forward not named Jimmy Hayes has the worst relative corsi and relative fenwick on the team?

These stats are not real time, by the way, so maybe with tonight's game and heavily out-chancing the Leafs, they may have changed quite a bit for the better in some cases.

I'm actually surprised that 13 of the 22 skaters that have played for us this year have a + rating in relative corsi, 11 of 22 have a + rating in relative fenwick, yet only 2 players have a corsi for rating of 50% or more and only 3 players have a fenwick for rating of 50% or higher. And Dalton Prout (LOL) is one of them in both categories. I'm sure this will get a laugh from the non-believers, especially with Severson and Noesen being the other two. And only 6 players have a relative corsi and relative fenwick rating of -2 or more.

And there's only 2 that have a relative corsi or relative fenwick of below -3 or below. Jimmy Hayes is one of them, at -7.0 relative corsi and -7.1 relative fenwick. The one really sore spot, and I'm gonna say it quick before the molotov's come my way. Well, to be fair, he was a + relative corsi and fenwick last year. against weaker competition. So it's not like he's sucks or that he isn't an NHL defenseman, he just shouldn't be playing as much as he is or against the competition that he's playing or getting 80% defensive zone starts.

Steven Santini

*duck's and tries to run away from the bullets and molotov's coming my way*

 

[Bleedred]

In reply to BG

You’re polluting a robust data set of 5 on 5 play over a whole season with man power difference data which isn’t part of the predictor.

The supposition is that Corsi advantaged 5 on 5 play generally leads to wins over the course of a season. We see this borne out by comparing standings to Corsi and it does generally match.

Unless you have an incredibly talented team, it's very hard to consistently make the playoffs while getting outshot badly every night. Ottawa did it a couple times recently, mostly because Hamburger guy had an extremely fluke run of goaltending. Anaheim was able to do it a lot during Carlyle 1.0 and Washington may have done it under Boudreau. There are a lot of instances where in a playoff series, the team that outshot the other team loses, like Washington last year against Pittsburgh. I mean, it's Washington, all bets are off with them. Anyway, my point is that's in a small sample size of games. Conversely, Pittsburgh of 2016 was the most dominant playoff team that I've seen in a while. They outshot some other really good teams by quite a bit. Last year though, not so much. It was more solid goaltending/opportunistic goal scoring. They basically rode their talent last year, whereas they rode a really perfect system + their talent the year before.

I think of the un-believers will always point to those couple of Pete years where we didn't make the playoffs, but I still don't think we were that great in those years. In 12-13, we were usually sunk pretty early by a soft goal, and were almost always chasing and pushing to get back into the game. Same thing with the half season under Johnny Mac. And even in all of the Pete years, our shots for weren't anything spectacular. You can argue the shot counter undercounts the shots, but if you argue that, you also have to consider that he undercounted the shots against also. We had a positive shot differential for the first few Pete years, but were always near the bottom in total shots for. If you struggle to get past the 25 shot mark every game, you ain't gonna score a lot of goals.

 

[Bleedred]

It's still early in the season and the stats are still easily to fluctuate a bit because of that, but we don't have as many black holes with a -3 relative corsi and below, like we did last year.

Coleman (he only played 23 games though) and Wood were both black holes last year, who have massively improved this year.

Coleman, Wood, Kalinin, Lovejoy, DSP, Gazdic (only played 11 games but was horrific, a special kind of suck, especially considering the limited ice time against probably limited competition). All of these guys were at least a -3.3 relative corsi and some even quite a bit worse than that. And the first 5 of them played 23 games or more and 4 of them played 43 games or more.

 

[Zajacs Bowl Cut]

can't believe how out of touch BG is on this whole Corsi debate.

"I think this is true, and if you don't, YOU'RE AN IDIOT!"

whew.

like I said yesterday, find me ONE PERSON who has ever said that Corsi is the "be all, end all" statistic. There isn't one. Its strawman nonsense.

 

[Feed Me A Stray Cat]

Benedict Gomez used the 2012-13 New Jersey Devils as another one of his anecdotal anti-Corsi examples, replete with unnecessary caps and bold. A team that played a 48-game season. I'm curious what type of statistical background he actually has.

 

[AfroThunder396]

Some people literally believe that stats guys go to sleep, wake up, look at the possession report of last night's game, and then do their analysis without ever watching hockey.

 

[Zippy316]

There's an interesting thread there where he analyzes the Devils up to now. In short, Devils are 29th/30th in unblocked and regular shot rates adjusted for score, allowing more shots in high danger areas (shown below), rank 5th in 5-on-5 save percentage, and have the ~26th worst penalty differential.

 

[Wingman77]

More would be accepting of advanced stats if those that follow them more than others didn't come across with the similar holier than thou attitude as a beer/wine snob. The forcefulness sometimes just becomes a bit over the top from some of those in support of it (not speaking necessarily on this site, more so in general). It is understandable why some would get turned off from even giving such stats a chance.

An objective and reasonable attitude that there are positives and negatives on either side would work just fine, but people are more concerned with being right on the internet

 

[Zajacs Bowl Cut]

More would be accepting of advanced stats if those that follow them more than others didn't come across with the similar holier than thou attitude as a beer/wine snob. The forcefulness sometimes just becomes a bit over the top from some of those in support of it (not speaking necessarily on this site, more so in general). It is understandable why some would get turned off from even giving such stats a chance.

An objective and reasonable attitude that there are positives and negatives on either side would work just fine, but people are more concerned with being right on the internet

its the total opposite in this case. Go look at some of BG's posts from late last night.

He decided that ALL situations should be used, not only 5 on 5....because he said so. He also demeaned people and said they don't understand statistics like he does.