Arguments Against Advanced Stats?

[idahophilly]

I literally came across this just now by accident. Had no intention of posting to this thread again but this fits in so well with the discussion. Just read it through and draw your own conclusions. I won't comment about it... It's his second point that pertains to analytics but the first point sets a context to it all.

 

 

When Analytics Don't Seem to Matter

December 7, 2014, 5:11 PM ET [15 Comments]  

Jason Lewis
 Los Angeles Kings Blogger • RSS • Archive • CONTACT

Last night's LA Kings 1-2 loss at the hands of the Flyers was frustrating, and it was frustrating for several reasons.

The Kings continue to be a team that is incapable of starting well. Too many times this year we've seen them get behind early and fail to catch up. The Kings have a winning % of .786 when they score first. That's good enough for 7th overall in the league. However, when scored on first they come in at 21st with a .231 winning percentage. Funny thing about the first period scoring, the Kings are a +1 overall in first period scoring differential but have trailed after one frame 11 times. In those 11 games they are 2-8-1. On the flip side they are 8-0-2 if they are leading after one period.

So clearly starts, strong ones, are very important to the Kings.

With that in mind, the team still struggles to find form early on in games. Last night was no different than a number of games already this season where the team has come out looking unprepared and unenergized. Let's round up last night's quotes about this subject.

Here's what Kopitar had to say,

We just didn’t get it going right off the start and we were playing catch up hockey for the most part and that’s hard.

A similar sentiment was echoed by Justin Williams,

We weren’t prepared for the one o’clock game today. We weren’t ready to go right away and got behind.

And finally, Darryl Sutter,

It wasn’t a bad start; it was a penalty late in the first period and two penalties early in the second period. Even though their struggle is their power play, trying to kill three in a row makes the difference between one and two goals. The power-play goal was not a great goal, it was a bad play against young defensemen and it was not a great goal, that’s tough.

Well, if you are behind 0-2 after the first 25 minutes I'd say that isn't a great start. That's just a very Sutter-esque way of looking at things. Bottom line, this is obviously a concern and has been throughout the entire season thus far. The Kings are not winning hockey games when they get behind first, and are trailing after the first period. Nonetheless, they have still had trouble getting off the blocks.

That is frustrating number one.

Frustrating number two comes in the forms of analytics, which has been a polarizing issue this season for both fans and analysts alike.

Does it have merit? How hard should we be looking into these things? 

Sometimes it seems like the world of NHL analytics is a venn diagram with a very small middle sliver who use both traditional and advanced stats to evaluate things. It's either one or the other, and the two sides constantly argue which one is wrong.

Well, if you are in the more traditional category, yesterday was a victory for you against analytics. Why do I say that? Because the Kings destroyed the Flyers. 



Even if you want to factor in score effects the Kings still heavily outplayed the Flyers and it wasn't even close. No offense to any Flyers fans out there who are reading this, but it was pretty clear that the Kings were the better team throughout the entirety of that game.

The Flyers play one of the more unstructured games out there. There seems to be little cohesiveness in their forecheck, neutral zone play, or their penalty kill. Their penalty kill in particular, which entered play yesterday as the worst in the league, is a complete circus. Coverage is pretty much all over the place with an emphasis on high pressure. The Kings were able to spread the Flyers PK on a number of occasions with good puck movement but couldn't find that final blow. The Kings put up eight shots on five powerplays and failed to convert.

That in a microcosm was the afternoon. The Kings were, as they say, all Swedish and no Finnish.

And that's where the analytics person throws their hands up in the air and says "Hockey is stupid sometimes" and walks away.

If "stupid" is the definition you want to use than be my guest. However, that stupidness is a huge part of the NHL and sports in general. Sometimes the better team DOESN'T win. Why do you think gamblers bet on the long shot? If the game were played on paper than no one would ever need to play the game. We could just theorize. That's what analytics are supposed to help us do, however this is still a world where a bit of a chaos theory exists. No statistic could tell you that Quick gave up a bad goal on the Flyers second goal, or that Giroux was uncovered on a center lane drive on the first. They were simply shots on the score sheet. The analytics numbers could tell you the truth in that game, which was that the Kings outplayed the Flyers for about 55 minutes of that 60 minute game. In the end though the two points go into the column for the Flyers. They did what they needed to do. They scored when they got the opportunity, and Mason made saves when he had to. It wasn't pretty but they got the job done.

All the analytics pointed to a Kings victory except for the one that mattered: 2-1 Flyers.

So the analytics are wrong, the Kings should have won. Not quite. As stated before, they told the truth of that game. But in the end did they didn't matter, because the Kings won the corsi game but not the hockey game.

Don't get me wrong, I'm not going off the Steve Simmons end and saying what he said in this now infamous tweeet:




But what I'm saying is that if there was a clear example that advanced stats are not the be-all end-all of hockey analysis, yesterday was it.

You've seen advanced stats quite a bit on this blog if you've followed along, but the moral here is sometimes they just don't matter. A couple of bad penalties and a leaky goal unraveled it all. It's not "Hockey is stupid". It's not "LOL good thing this isn't the Corsi Hockey league." It's just the chaotic world of sports. You can do your best to try and narrow down the good play from the bad, which is what analytics attempt to do. However, everything has its flaws and unpredictabilities. 

Thanks Flyers, you broke corsi. You got the two points in the process though, so who cares right? If you're the Kings, you definitely want to keep playing games like you did yesterday minus the bad start. Eventually a puck or two has to go in right? You'd hope.

The Kings start a Northeast swing through Buffalo and Eastern Canada on Tuesday, and aside from the whole corsi thing they have some work to do. The bad road record will come into focus yet again. Perhaps the team can start to better that 3-4-4 record, and it starts with getting off on the right foot and scoring first.

Follow me on twitter for news and notes about the Kings and the NHL

[BillyInFourC]

so...have we just defeated the people who say +/- is irrelevant because it relies too heavily on the other 9 guys?  +/- FTW?

 

+/- is irrelevant because it's dramatically swamped by noise and has no predictive value from season to season. As the post by Eric T. explains. 

Shot attempt +/- side steps these issues by increasing the sample size of events, giving it predictive value over much smaller samples. 

 

secondary metrics which themselves are based entirely on corsi.  see how the snake eats its tail on this?  could do every bit of the same qualifying with +/-, if you wanted to.  but you don't.

 

 

QoC, OoT, etc can all be measured with or without corsi. This is why TOI% metrics exists, to remove that sampling bias. 

 

your "decade of hockey analytics research" is a concept appropriated from a goalie coach who wanted a measure of how much work his goalies had to do against different teams and different lines.  the thought that it has something to say about the ability or contribution of a particular skater is NOT what it was meant to do, and it has only been used as such for a few years.  and poorly, at that.

 

This a complete non sequitur. What is was developed for and how is used now are completely separate applications with entirely different justifications. If you want to address the actual research that demonstrated the predictive value of these stats, be my guest.

As for your assertion that they are used poorly, that's just like, your opinion man...

 

the rest, the actual "here's what's wrong", you are bringing that to the table by "conventional", non-advanced means.  in your example, a poor corsi was the warning sign that sent you to look at other, more meaningful things.  i guess maybe it is useful like that, as a canary in the coal mine for teams whose management is paying zero attention but has an analytics department.  see how you didn't actually get any particular meaning out of it, though?  see how the actual interpretation came from something very other than shot attempt ratios?  see how shot attempt ratios probably aren't actually required for you to think that maybe something isn't great with the flyers' defensive puck movement, grossmann in particular?  see how the corsi numbers effectively add nothing to the conversation, other than geek cred?

 

 

They add nothing of value to the conversation... except for the front office that acquired the likes of R.J. Umberger, signed Andrew MacDonald to a 30 million dollar contract, and signed Lecavalier only to demote him to the 4th line and bench him a season later. 

As I said before, numbers like these are a check on your assumptions about what makes a hockey player valuable in terms of tangible on ice results. They have predictive value about the how players and teams will perform in the future even if luck runs against them in the short term.

They provide an objective framework for eliminating and countering biases they we develop solely using the eye test. ex. Coach's tend to operate on recency bias when making roster decisions. Player A was on the ice for a lot of goals last game. Player B made just as many critical mistakes, but got bailed out by his goaltender. Player A gets benched or demoted, Player B keeps his ice time. 

[OccamsRazor]

Thanks Flyers, you broke corsi.

 

Basically what i was saying........kind of why i always hate when my Flyers out shoot the opponent by a wide margin more often than not it always seems to end in a loss. I have not stats to back that up stat guys it's in my head. It just seems that way and it seemed fitting that they get a total of 16 shots to 38 and win....so i'll take it.

[flyercanuck]

I didn't see one proponent of the stats say "this is the end all be all, you may not use any other tool to evaluate players" 

The stats are tool to assist video scouting and point out broad trends. The flyers have only played 26 games but we're seeing the continuation of the same trends from last season. Namely that Grossmann, MacDonald, Lecavalier, and Umberger(going back to CBJ numbers) have had dramatic negative impacts on their team's on ice results.

 

 

You don't need advanced stats to see that. Watch A game.

[murraycraven]

There is something to be said for advanced stats...   what it does not count for is the intangible data and also leads to Paralysis by Analysis.

 

 

 

[flyercanuck]

@murraycraven

 

I'd just prefer to enjoy watching hockey instead of making it feel like I'm learning the periodic table of elements.

[brelic]

Does it have merit? How hard should we be looking into these things? 

Sometimes it seems like the world of NHL analytics is a venn diagram with a very small middle sliver who use both traditional and advanced stats to evaluate things. It's either one or the other, and the two sides constantly argue which one is wrong.

Well, if you are in the more traditional category, yesterday was a victory for you against analytics. Why do I say that? Because the Kings destroyed the Flyers. 

 

Wow, that was a poorly written article with an even poorer understanding of statistics and the application of models. 

[murraycraven]

@murraycraven

 

I'd just prefer to enjoy watching hockey instead of making it feel like I'm learning the periodic table of elements.

 

 

Agree FC...   I think they can be useful when taken into context and not used as the gospel.   There is some merit in some of the metrics.  With that being said there are a number of stats that can be flawed in the way they calculate or track the metric.   At the end of the day the eyes dont lie and understanding the actual game is more important IMO.  Stats are great but there is no equation that takes into account the system, strategy and/or tactics a Team plays.   They have there place but should not be the end all be all of decision making...

 

 

 

[Bertmega]

Math can't tell you when a player has the flu, or his dog died, or he drank too much the night before. It is great for projections, but as Podien25 stated, there are too many intangibles involved in this game that stats are just that.... stats.

 

It reminds me of the stock market. Analysts try finding value to beat an index and thus look like rock stars. Statisticians in the sports world are trying to find value in a player, but often fall short due to the intangibles that a human being goes through. 

 

[brelic]

At the end of the day the eyes dont lie and understanding the actual game is more important IMO.  Stats are great but there is no equation that takes into account the system, strategy and/or tactics a Team plays.   They have there place but should not be the end all be all of decision making...

 

MC, I think this is where statistics are important - eyes *do* lie and they lie a lot. Especially when it comes to complex situations. Often statistics are used to confirm what your eye tells you. They act as a second opinion that tells you you're not crazy after all. And other times, they can alert you to something that your eye just can't pick up because there are so many other things happening at once. They prompt you to investigate further.

 

The strength of corsi (as a team measurement of puck possession - which I prefer to the individual corsi number) is that it really doesn't matter what system, strategy, or tactics a team uses. Puck possession is puck possession. That's all it's trying to do - and as a team metric, it shows a strong correlation with playoff berths and playoff performance.

 

What the article posted by @idahophilly showed was a fundamental misunderstanding and misapplication of statistical models - as if it should be able to predict the outcome of a particular game, and then when it fails to do so, well, it must not be any good! It's like saying that a coin toss which comes up heads twice in a row disproves the 50/50 accepted statistical probability of the coin toss.

[brelic]

These are last year's team corsi numbers (pre-playoffs):

 

Rank Team Corsi

1 Los Angeles Kings +630

2 Chicago Blackhawks +495

3 Boston Bruins +450

4 New Jersey Devils +382

5 San Jose Sharks +361

6 New York Rangers +299

7 St Louis Blues +248

8 Ottawa Senators +214

9 Vancouver Canucks +197

10 Tampa Bay Lightning +135

11 Dallas Stars +120

12 Detroit Red Wings +106

13 Columbus Blue Jackets +70

14 Phoenix Coyotes -14

15 Anaheim Ducks -15

16 Florida Panthers -26

17 Pittsburgh Penguins -27

18 New York Islanders -40

19 Nashville Predators -45

20 Winnipeg Jets -57

21 Philadelphia Flyers -65

22 Carolina Hurricanes -70

23 Minnesota Wild -152

24 Washington Capitals -212

25 Colorado Avalanche -238

26 Montreal Canadiens -252

27 Calgary Flames -383

28 Edmonton Oilers -539

29 Toronto Maple Leafs -739

30 Buffalo Sabres -833

 

 

Look at Colorado as a great example of an outlier - they had a great regular season, but a terrible Corsi rating. If you look behind the numbers, I think the first thing you point to is Varlamov's 2.41, .927 as a major contributor to the Av's success. It caught up with him this year, and now the team looks like one would expect with such low possession numbers.

[murraycraven]

These are last year's team corsi numbers (pre-playoffs):

Rank Team Corsi

1 Los Angeles Kings +630

2 Chicago Blackhawks +495

3 Boston Bruins +450

4 New Jersey Devils +382

5 San Jose Sharks +361

6 New York Rangers +299

7 St Louis Blues +248

8 Ottawa Senators +214

9 Vancouver Canucks +197

10 Tampa Bay Lightning +135

11 Dallas Stars +120

12 Detroit Red Wings +106

13 Columbus Blue Jackets +70

14 Phoenix Coyotes -14

15 Anaheim Ducks -15

16 Florida Panthers -26

17 Pittsburgh Penguins -27

18 New York Islanders -40

19 Nashville Predators -45

20 Winnipeg Jets -57

21 Philadelphia Flyers -65

22 Carolina Hurricanes -70

23 Minnesota Wild -152

24 Washington Capitals -212

25 Colorado Avalanche -238

26 Montreal Canadiens -252

27 Calgary Flames -383

28 Edmonton Oilers -539

29 Toronto Maple Leafs -739

30 Buffalo Sabres -833

Look at Colorado as a great example of an outlier - they had a great regular season, but a terrible Corsi rating. If you look behind the numbers, I think the first thing you point to is Varlamov's 2.41, .927 as a major contributor to the Av's success. It caught up with him this year, and now the team looks like one would expect with such low possession numbers.

Like I said before... Advanced stats can be useful but if a team is using advanced stats to make every decision I think it is overkill. Not against them at all but there needs to be balance on both sides of the proverbial statistical line.

If advanced stats are the end all be all we won't need hockey people making decisions or hockey people analyzing the draft prospects.

Possession is a big thing no doubt...

[radoran]

Look at Colorado as a great example of an outlier - they had a great regular season, but a terrible Corsi rating. If you look behind the numbers, I think the first thing you point to is Varlamov's 2.41, .927 as a major contributor to the Av's success. It caught up with him this year, and now the team looks like one would expect with such low possession numbers.

 

Late to the party, but as many have said, statistics are a tool and what you do with that tool is the important thing. Numbers, yes, can be manipulated to say a lot of things "in specific" but "in general" they can be a lot more valuable. You can use a wrench as a hammer, but it is more effective than a hammer at being a wrench.

 

Of the 16 playoff teams last season, four are in the bottom half of the table you posted. That's one in four. Not a perfect correlation, but certainly an indication that there is validity in the stat in general. A "better" Corsi is an indication that you will likely do better than your opponent.

 

Of all the playoff series last year, you could pick the winner by Corsi in all but four of them - Montreal twice (over Tampa and Boston), Pittsburgh over Columbus and Anaheim over Dallas. Again, that's four of 15.

 

In every other series - including 3 of 4 Conference Semis, both Conference Finals and the Cup Final - the "better" Corsi team wins.

 

It definitely shows the effect of "intangibles" (like the Habs/Bruins rivalry factor, for example, or Tampa's relative postseason inexperience), but again correlates fairly well.

 

There will always be statistical anomalies - like the Kings outshooting the Flyers and losing. But over the course of a season - or even a seven game series - the team that has a 20+ shots advantage is more likely to win than the one with the -20 shots.

 

Nothing - nothing - predicts the future 100%, but if you see something that has a 66%-75% correlation, it does tend to indicate a greater degree of predictability than simply flipping a coin.

 

Again, it comes down to how you look at it and how you use it.

 

And, in general, it is better to be more general than more specific, because the greater degree of specificity you seek, the less effective many statistical analyses get.

[mojo1917]

The strength of corsi (as a team measurement of puck possession - which I prefer to the individual corsi number) is that it really doesn't matter what system, strategy, or tactics a team uses. Puck possession is puck possession.

 

My gripe with Corsi is there is no accounting for the quality of the shot, if you wan to know about puck possession , use a stop watch, my guys got the puck in the neutral zone held it for 30 seconds and took a shot at which point they no longer possess the puck. 

 

Corsi makes no allowance for: my guy jumped a pass at the blue line and took 3 strides and let a shot go, unscreened from 50 feet... Corsi says that's a shot and is "good"  when just about anyone watching the game will be have an idea whether it was a good idea to take that shot.  There have been games where i've watched teams have the puck and shoot it from 40 feet and beyond 30 times and maybe 3 of those shots were, difficult for the goalkeeper.  Go ahead good Corsi team keep shooting from there, is what i say.

 

I think Corsi as a puck possession metric is flawed. 

[radoran]

My gripe with Corsi is there is no accounting for the quality of the shot, if you wan to know about puck possession , use a stop watch, my guys got the puck in the neutral zone held it for 30 seconds and took a shot at which point they no longer possess the puck. 

 

Corsi makes no allowance for: my guy jumped a pass at the blue line and took 3 strides and let a shot go, unscreened from 50 feet... Corsi says that's a shot and is "good"  when just about anyone watching the game will be have an idea whether it was a good idea to take that shot.  There have been games where i've watched teams have the puck and shoot it from 40 feet and beyond 30 times and maybe 3 of those shots were, difficult for the goalkeeper.  Go ahead good Corsi team keep shooting from there, is what i say.

 

I think Corsi as a puck possession metric is flawed. 

 

Honestly, Corsi isn't trying to make that specific determination.

 

It's not a specific stat. It's a general stat.

 

And it does seem to break down that in general a team that possesses the puck more and takes more shots has a better chance of winning more games than those with bad puck possession and fewer shots.

 

You can find many specifics that fly in the face of this general tendency, but that doesn't in any way invalidate the general tendency for the stat to be useful.

 

You can have a guy take a shot that goes off the shinpads and into the corner and results in nothing.

 

You can have a guy take a shot off the shinpads and off the back wall/post/stick/skate, back in front, resulting in another scoring chance and, potentially, a goal.

 

Over the course of the whole season those things, statistically, even out. Even if that specific "shot" that was deflected into the corner didn't result in anything, that doesn't mean that the "shot" itself was a bad idea.

[ruxpin]

Corsi makes no allowance for: my guy jumped a pass at the blue line and took 3 strides and let a shot go, unscreened from 50 feet... Corsi says that's a shot and is "good"  when just about anyone watching the game will be have an idea whether it was a good idea to take that shot.

 

I hear you.

 

On the other hand, think of it this way:

 

The stat sheet says Bob went 1 for 3 at the plate last night.  What it doesn't say is whether it was a blooper that fell in shallow center between three players, an infield dribbler in which he beat the 3rd baseman's throw, or a screaming liner down the left field foul line.   But it's 1 for 3 either way.  

 

I'm a little hung up, too, on how corsi compensates for good vs. bad shots, etc., but if Rad is right and it's nearly a 75% proposition between corsi and actual results in terms of series results and even making the playoffs, then somehow you and I might be putting too much weight on that.  

 

So, yes, it's a tool.  Possibly a valuable tool (which is unfortunate, because I cannot get my mind around how to read it!) used in conjunction with other tools available.  

[radoran]

I'm a little hung up, too, on how corsi compensates for good vs. bad shots, etc., but if Rad is right and it's nearly a 75% proposition between corsi and actual results in terms of series results and even making the playoffs, then somehow you and I might be putting too much weight on that.

 

What means you "if"??

 

I haven't done any analysis except looking at last year based on brelic's post.

 

"quick and dirty" to be sure and additional analysis would result in better determination of long term effectiveness.

[brelic]

My gripe with Corsi is there is no accounting for the quality of the shot, if you wan to know about puck possession , use a stop watch, my guys got the puck in the neutral zone held it for 30 seconds and took a shot at which point they no longer possess the puck. 

 

I think I understand what you're saying. 

 

From my understanding of corsi, the game to game variation doesn't matter much. In the long term, the team that consistently outshoots (SOG, shot attempts incl. blocked shots) its opponent will win more often than it will lose, regardless of 'quality' of shot because that kind of stuff evens out. It's not the 'quality' that matters, but the quantity (in the long run) because quantity is one way of determining puck possession. 

[ruxpin]

What means you "if"??

 

Kind of in the same manner of "if ruxpin is smarmy."

 

It's pretty much a given but used for argument purposes.  Poetic license, if you will.

[mojo1917]

@radoran

@ruxpin

 

I love it when you argue with yourself.

 

Keeps that whole two different people illusion going ... good work !

 

Part of me feels very old school and fuddy duddyish at my resistance to embrace the advanced metric, but then there are advanced metrics i do find to be helpful, quality of opponent , zone starts, zone entry, those things i can see being helpful more readily than I do Corsi , which I think has a nebulous feel, maybe while i'm chanting some shantis I will become enlightened by usefulness of Corsi.

[BillyInFourC]

Like I said before... Advanced stats can be useful but if a team is using advanced stats to make every decision I think it is overkill. Not against them at all but there needs to be balance on both sides of the proverbial statistical line.

If advanced stats are the end all be all we won't need hockey people making decisions or hockey people analyzing the draft prospects.

Possession is a big thing no doubt...

 

When looking at the corsi standings, you have to remember that other factors influence W/L record and making the playoffs. Puck possession is a huge part, but it's not the only part. 

1. Goaltending

2. Shooting Talent 

3. Special Teams 

4. Possession

5. Luck 

The Avalanche rode stellar goaltending into the playoffs last season. The Flyers, despite being mediocre at 5v5 last season, had a top-10 special teams unit that helped them win games.

As I've said before, no analyst, and none of the guys hired by NHL teams over the summer would argue that possession metrics are the end all be all, just that they are an important part of the puzzle for predicting success. 

[murraycraven]

When looking at the corsi standings, you have to remember that other factors influence W/L record and making the playoffs. Puck possession is a huge part, but it's not the only part.

1. Goaltending

2. Shooting Talent

3. Special Teams

4. Possession

5. Luck

The Avalanche rode stellar goaltending into the playoffs last season. The Flyers, despite being mediocre at 5v5 last season, had a top-10 special teams unit that helped them win games.

As I've said before, no analyst, and none of the guys hired by NHL teams over the summer would argue that possession metrics are the end all be all, just that they are an important part of the puzzle for predicting success.

Fully agree on all counts.

[BillyInFourC]

My gripe with Corsi is there is no accounting for the quality of the shot, if you wan to know about puck possession , use a stop watch, my guys got the puck in the neutral zone held it for 30 seconds and took a shot at which point they no longer possess the puck. 

 

Corsi is a popular proxy for both puck possession AND scoring chances. The team that gets the most shot attempts also tends to get the most scoring chances over time and out score the opposition. 

 

The reason why people don't just use a stop watch and scoring chances is because the league doesn't automatically generate that data. It takes hundreds of hours of manual tracking just to log scoring chances for a single team. The effort generally isn't worth the marginal improvement in the data. 

Here's a good article from Eric T. explaining the relationship between shot attempts and chances if you have the time.  

 

http://nhlnumbers.com/2012/6/26/shot-quality-revisited-a-look-at-the-correlation-between-scoring-chances-and-shot-totals

[mojo1917]

@BilliyinfourC

thanks for that explanation.

[JackStraw]

And it does seem to break down that in general a team that possesses the puck more and takes more shots has a better chance of winning more games than those with bad puck possession and fewer shots.

 

Or to put it another way, "good teams are good" 

 

I'm not a stat guy but stats can be interesting and any stat is useful if used in the right context. Sometimes the tough part is figuring out what that context is exactly. Grossman is an interesting case study. It's been pointed out that he has lousy Corsi but good +/-. What to do, what to think? Well, if you watch the Flyers you know some things about Grossman:

 

1. He contributes very little offensively. His two goals this season match his career season high, which he has done only once before. He's never hit 15 pts in a season. It stands to reason that he wouldn't contribute much directly to a positive Corsi number. He doesn't get involved much in the offensive zone and he's not a puck mover.

2. He's pretty much the definition of a stay at home defensive defenseman. He's most effective close to his own goal, below the hashmarks, in the corners, along the boards. He blocks a lot of shots. So there's no reason the expect to really significantly suppress shots against attempted (as opposed to actual SOG) either. Stay at home, clear the porch, block shots, tie up guys along the boards. That's basically what he does. None of the things that he's good at really have much influence on shots attempted (for or against) other than in a very indirect way.

 

The fact that his Corsi is "bad" but his +/- is "good" suggests to me that he's doing his job pretty well.

 

http://www.behindthenet.ca/nhl_statistics.php?ds=63&s=35&f1=2014_s&f2=5v5&f4=D&f5=PHI&f7=10-&c=0+1+3+5+8+32+33+34+45+46+11+12+13+14+15+16+29+30+31+35+36+37+38+39+40+47+48+49+50+51+52+53+54+55+56+63+67+57+58+59+60+61+62+64+65+66

 

Not sure how up to date those stats are but, among Flyers defensemen he actually ranks 1st in GF per 60 minutes, and 2nd best in GA per 60 minutes.

 

I mentioned in another thread that last season he was horrid and I would have traded him for a few back issues of the Hockey News. He looks a lot better this year (eye test), and stats (advanced and otherwise) seem to bear that out.