Friday, April 23, 2010

Synergy

Scott Stevens in the playoffs without Martin Brodeur:
42 wins, 51 losses

Martin Brodeur in the playoffs without Scott Stevens:
16 wins, 26 losses

Scott Stevens and Martin Brodeur in the playoffs together:
83 wins, 56 losses

Apart, neither has won a game beyond the second round. Together they have 3 Stanley Cup rings. Just in case anybody needed a reminder that teams win in the playoffs, not individuals.

And no, I'm not criticizing Brodeur. He wasn't the reason New Jersey lost this year, just like Scott Stevens wasn't the reason that Washington never won anything in the 1980s. I feel like I could write the same post I did last summer, where I argued that Brodeur has been the same guy in the playoffs whether the Devils were winning or losing, the difference was the play of the team in front of him. When your team scores eight goals in five games on Brian Boucher, then you're very probably going to lose.

28 comments:

R O said...

I haven't watched too much of NJD vs. PHI but I am guessing that New Jersey held a terrific territorial and chance advantage?

I mean, just by looking at both teams and factoring in their injuries, I would have guessed NJD was by far the better team.

In any case I did hear both praise and condemnation of Brodeur from CBC/TSN panelists but the fact remains that if you're getting goal support of less than two per game (even if your team deserved to score more) then you're not going to win a lot of games (even if you deserved to).

By the way, you are a former/current goalie, right? That makes your stance of goaltending and luck in general very... un-goalie-like. :-)

The Contrarian Goaltender said...

NHL.com has New Jersey with 135 shots, 78 missed shots and 85 blocked by the Flyers while Philadelphia had 126 shots, 47 missed shots and 56 blocked by the Devils. That's a Corsi edge of 298-229 for New Jersey, although that includes special teams and the Devils would also have spent much of the series playing from behind.

It's not surprising that most goalies overestimate their own impact on the game. That's probably true of forwards and defencemen as well. But really anybody who has played net for both really good and really bad teams ought to be aware that it's a team game. Especially a really bad team, that experience tends to really drive the point home in a frustrating fashion. Or at least it should, but even among goalies you'll find people who do nothing but spout conventional wisdom and rate guys based on wins, which of course really baffles me.

Regarding their perspective on luck, I'm not sure if goalies are really a different breed on that one. The population of goalies and the population of people who understand randomness are both fairly low relative to the population at large, so it might just be that the chances are pretty low that you know someone or watch someone on TV who falls into both groups.

On the other hand, I can see why many goalies try to convince themselves or find it preferable to believe that they are the ones who are in control and that they have a chance to stop every shot against. That's probably reinforced by most coaches as well, who don't want their players to use bad luck as an excuse. I guess for some people the correct outlook may not necessarily be the best one in terms of motivation.

nightfly said...

If you like another goalie perspective, R O, I can tell you that I go into a game with only one real goal - stop the next shot. I can't make saves in the last five minutes until I get there. I can't make forty saves on fifteen shots. I certainly can't save the last goal. I can only stop the next one.

That has been a big mental help to me. I used to really drive myself crazy over bad goals, but experience and a lot of watching hockey has since taught me that EVERYONE is prone to let a softie by. Sometimes the softies aren't so soft, as the team has been trapped in its own end for three solid minutes and they're just caught flat-footed. At my level, of course, it's more important because more of the goals are soft - if we were good enough not to let in soft goals somebody would be paying us!

The biggest challenge in net is to realize that you can't win the game - ever. You can only lose it. If my forwards blow a few chances they can often keep possession; if the defense errs, well, that's why I'm there. If I screw up it's in my net. That makes goaltending seem like the most important thing ever, but it's almost all one-way: bad goaltending will cost you, but even elite goaltending may not be enough to save you. You can be perfect but you still need goals from your guys. You can be perfect and give up three on screens, deflections, amazing moves, dumb luck...

The truth is, I'm a goalie primarily because I was always really small for my age. I got picked last for everything. Guys who get picked last get stuck in goal. I found that I enjoyed it, had the reflexes, and was willing to work at improving. Ta-da, a goalie! If I got taller in grade school maybe I'd be a forward now. :) So much for the goalie myth.

Derick said...

I'd been thinking about this as well. Often when you call a player like Brodeur or Crosby overrated because of team success, people say 'the team wouldn't have done as well without them.' That's true, because they're very good players. But the Penguins wouldn't have won the cup without Talbolt, not to mention Malkin. Now Crosby is obviously a better player than Talbot, but the point is, whether or not a player is necessary for a team's success, that doesn't mean team success is a good indication, because players with inferior teams will still have less team success even if they're better.

The Norris trophy nominees were announced today. Who do you think deserves to win? I think Doughty is the clear winner and Keith is no better than Green. Click on my name if you're interested in reading why. I do some complicated math and go beyond the 'is the points leader higher in plus minus higher than the plus minus leader is in points?' and cliches most seem to. For instance, toi compared to team goals against is an underappreciated stat.

R O said...

Yeah Nightfly, I think that's a good philosophy to have when you're tending nets.

It always drives me nuts, the obsession some people have with the timely save or the untimely goal against.

I mean the nature of the position is failure. In what other reasonable profession can a failure rate of 1 in 14 be considered the best that there ever was, is and will be?

--

Norris should be Lidstrom. Seriously.

Green is what he is. A soft-minute scoring defenseman, very skilled at it, but the fortunes of his team don't change nearly as much when he's out of the lineup as one might think.

Doughty, well the kid has a future but he's a young defenceman and those are not to be trusted. There are indications that he's being sheltered and indications that he is not, but the arrows don't all point up like people believe.

Keith is the most deserving of the trio on the nominee, terrific player of course, but Lidstrom is still the man.

Bruce said...

Hey Derick, I read your post but you have comments disabled on your blog so I couldn't comment there.

You should be aware of Gabriel Desjardins' fantastic source, behindthenet.ca where the type of advanced stats you are looking for and many others are freely available. Every player on every team, even strength as well as powerplay and shorthanded, and you can set your own filters as to team, position, minimum GP or TOI ... wonderful stuff.

I used this resource extensively when doing my own recent analysis of the Selke Trophy finalists, which you may find of interest.

Derick said...

Thanks a lot Bruce. I'll check it out.

RO, how can you say Keith deserves it more than Doughty. Keith is on the ice for an average of 22 minutes even strength per es goal against, Doughty for 30! Sheltered by who, his inferior LA Kings team when Keith plays for the Blackhawks? Or do you mean the league?! What possible incentive does the league have to shelter an LA Kings defenseman of all people? That's not exactly their marketing hub.

If you read the post I made about it you'll find that a) Green is not as bad defensively as people make out, and b) Doughty blows both of them out of the water defensively.

Unfortunately a popular mistake can give you a stigma, and a couple bad plays in last year's playoffs did that to Green. But the Norris trophy isn't awarded for the best defenseman in last year's playoffs. That being said, Doughty still deserves it the most.

I haven't looked at Lidstrom much but if you're as lazy about evaluating him as you are the three nominees I'm not hopeful for him.

Jonathan said...

Nice analysis Derick. I think the one major flaw in your analysis is that you did not account for all of the PP time that Duncan Keith is missing. Duncan is arguably the best offensive defenseman in the league after accounting for the fact that Green spent 5:03 of every game on the ice for power plays, while Duncan Keith spends 2:48. That's more that an entire minor penalty every game.

Anyways, your analysis of defensemen vs. goals against is absolutely solid. I decided to re-crunch the numbers to see how many goals per 60 minutes each player would get if they both had 7.5 PP minutes and 7.5 PK minutes per 60. I use 7.5 as my number since they collectively spent 25% of their total ice time on either the PP or the PK.

So using your numbers--thanks for the data :) --

Drew Doughty: ESTOI/GA = 30.38; 45 ES MIN/60 * GA/30.38 ES MIN = 1.481 ESTGA/60
PKTOI/GA = 8.25; 7.5 PK MIN/60 * GA/8.25 PK MIN = 0.909 PKGA/60
I'll be lazy and assume no short handed goals against, so this adds up to 2.390 GAA for Drew Doughty per "normalized" 60 minutes. The calculations above look sorta complicated, but it's not really that bad other than the fact that it has a lot of letters. It's basically just how many goals against he'd average every 60 minutes if he spend 45 of those minutes at even strength, 7.5 on the PK, and 7.5 on the PP.

So onto Duncan Keith: 45/21.05 = 2.138
7.5/11.61 = .646
Total = 2.784 GA/60

Mike Green: 45/22.08 = 2.038
7.5/8.9 = 0.843
Total = 2.881

so adding up all of the numbers, I basically came to the same conclusion as you, just with a different method of quantifying the results:

Doughty = 2.390 GAA per 60 minutes
Keith = 2.784 GAA per 60 minutes
Green = 2.881 GAA per 60 minutes

Again, there aren't actual totals but hypothetical totals; what their numbers would look like if they both spent the same proportion of their respective ice time on special teams. IMO it's a good way to normalize for differing roles.

Now your interpretation of offensive numbers is where I start to disagree. You mentioned that offense should count less for defensemen because the main job of a d-man is to...defend. This is absolutely correct. Here's the thing: by only accounting for goals in which the defenseman scored a goal or an assist, you are already eliminating a large fraction of the goals that he was on the ice for. The defenseman's offensive weighting is already halved to begin with. Then you (correctly) assign less value for an assist than you do for a goal. This is more than enough to halve the weight of a defenseman's offensive prowess. After that, there should be no further need to reduce that particular weighting. After all, a defenseman is directly responsible to any goal that he scored, obviously, and also for any goal which he assisted on.

So onto the number crunch:
Doughty: 362.58 PPTOI/23.67 PPP = 15.318 minutes per PP point
1519.85 ESTOI/21.00ESP = 72.370 minutes per ES point
No short handed points for doughnuts.
It comes out to 1.118 points per 60 minutes.

Keith: 230.40 PPTOI/11.67 PPP = 19.743 minutes per PP point
1705.75 ESTOI/35.33 ESP = 48.280 minutes per ES point
244.417 SHTOI/3.67 = 66.599 minutes per SH point

This comes out to 1.425 points per 60 minutes.

Mike Green:
379.80 PPTOI/26.67 PPP = 14.241 minutes per PP point
1369.72 ESTOI/29.67 ESP = 46.170 minutes per ES point
one shorthanded assist, so one point per 240.91 minutes shorthanded.
Crunch the numbers and Mike Green has 1.532 points per 60 minutes

So I'm just going to sum out the two totals by subtracting points/60 out of GA/60. And now the moment you've all been waiting for...

Drew Doughty: 1.272 GA/60
Mike Green: 1.349 GA/60
Duncan Keith: 1.359 GA/60

Jonathan said...

Over an 82 game season, Doughty is less than 3 goals better than Duncan Keith. This difference can easily come down to a variety of team factors that I have not addressed here. On one hand, Duncan Keith has the worst performing goaltenders of the three. If every team's sv% were to match LA's 91.0% mark, then Keith's stats would improve by 0.28 GA/60, enough would leapfrog both of his rivals. On the other hand, Doughty's team clearly has the weakest skaters. Then again, I saw a stat saying that Keith was competing against the other team's top line far more often than Green or Doughty. I didn't analyze or look into what that stat means, but it seems to favor Keith very strongly.

So I think it should be Duncan Keith.

I have to admit though that I had no clue how to evaluate defensive performance statistically until I read the post on Derick's blog. So mad crazy props on that one.

The Contrarian Goaltender said...

To me, the Norris should go to Duncan Keith in a relatively straightforward decision. I think you forgot to include one major factor in your analysis, a factor that happens to be the focus of this blog: Goaltending.

From Behind the Net, here are the save percentages at even strength behind each of the top 3 guys (plus Lidstrom):

Doughty: .920
Green: .922
Lidstrom: .919
Keith: .899

Keith had the fewest shots against per minute while he was on the ice of the 4 but still ended up with the most goals against, thanks to Niemi and Huet.

Keith's personal EV GAA this year (2.77) was actually much higher than it was in each of the past two seasons. In both of the last two years Keith was very close to Doughty's numbers this year (Keith had a 2.18 last year and a 1.98 the year before, Doughty was at 1.95 this year).

Given that Keith has an established track record at that level, that his shot prevention remained outstanding and that the subjective reports on him this year were very good, I think it's pretty unsupported to suggest that either he declined in effectiveness or that Doughty is clearly superior defensively. The Hawks goaltending was poor this year, but Duncan Keith is still one of the top defensive defencemen in the game.

What clinches it for Keith is his 48 even strength points. That level of even strength offence by a defenceman hasn't been seen in the NHL in at least fifteen years. Not only that, but nobody else has even come close. Mike Green's 40 ESP this year is the second-highest mark since the detailed NHL stats became available in 1997-98. Keith is 20% ahead of him. Nicklas Lidstrom's career-high ESP is 37. Keith this year was 30% ahead of that. Maybe Keith got a bit lucky in shooting percentage or getting second assists or something this year, but you don't get that far ahead of your teammates and the rest of the league unless you're one of the guys driving results.

Keith is probably the biggest 5 on 5 difference maker in the league from the blue line, and he does it playing against top competition. If he got more power play time he'd have even more points. Let's put it this way, on a per minute basis Doughty's rate of power play scoring was 23% higher than Keith's. At even strength, Keith's rate was 53% higher than Doughty's.

Doughty is an impressive young defenceman, but he's still not quite Norris calibre. I'd have to agree with RO that I'd still take Lidstrom over him. There's little doubt that award recognition will be coming for Doughty, just not quite yet.

R O said...

Some players look better than they are because the coach makes their icetime easier. And some players look worse than they are because the coach makes their icetime harder.

This is a pretty simple principle, and it applies to defencemen in a big way. Keith and Lidstrom are given easily the hardest icetime on their teams, and the puck is overwhelmingly in the good end of the ice anyway.

You kind of don't get to win a lot of hockey games unless you can drive the puck 175 feet up ice. That's where scoring chances happen, after all.

Agent Orange said...

CG

"Doughty: .920
Green: .922
Lidstrom: .919
Keith: .899"

I took this a little but further and checked them compared to the rest of their team (using BTN.ca stats).

Below are the save% with them:

On-ice off-ice diff GSin82
0.920 0.915 0.0046 10.15
0.922 0.929 -0.0065 -16.09
0.919 0.916 0.0025 5.90
0.899 0.900 -0.0015 -2.92

I couldn't find (because I am worthless) SA/60 for each player on the site so this is a pretty rough estimate but it shows the following order:

Doughty
Lidstrom
Keith
Green

Yes Keith has the worst goalies behind him but he seems to do less favors for them.

Two other notes.

1) Quality of oppoents.

I would expect Lidstrom/Keith would be playing against the best line every night.

I don't know enough about/haven't seen enough of Green/Doughty to guess. From what I have read on here is that Doughty is "sheltered" but I'm not sure.

2) Quality of teammates.

In terms of defensive ability Lidstrom likely has the greatest assets. Zetterberg/Datsyuk/Rafalski and a host of other 2-way forwards and good defensemen will help with the team defense while he is on the ice.

This however makes his improvement more impressive when compared to his teammates.

Thoughts? Opinions? Did I screw up the math?

Lawrence said...

... and that they have a chance to stop every shot against

...who don't want their players to use bad luck as an excuse

...You can be perfect and give up three on screens, deflections, amazing moves, dumb luck...


Bad "luck" happens to bad goalies. I firmly believe that. Perhaps, it's just a semantics thing, but I've been playing goal for over 20 years and evaluating goalies for another five, and call it what you will, there really is no such thing as "luck" in goaltending. Without this sounding like ridiculous hyperbole, the kids I see who are weak mentally talk about being unlucky, the ones who are strong, say they should have had it, no matter how it went in.

If you don't think you have a chance on every shot, you really shouldn't be playing goal...maybe play something that involves dice.

Goalies benefit by the knowledge that any goal that goes in is 'bad', but that has nothing to do with luck and everything to do with technique, positioning and decision making. So, when one goes in, no matter if it bounces off three guys in the process, you note it and you move on. It's done.

That isn't bad luck, that is a goal.

About the only part of hockey that I would agree with someone if they said that was "bad luck" are goalposts, which most people say are "good luck".

This just goes to show how distorted ideas of 'luck' and goaltending are.

How is it good luck if a shot hits the post and bounces out (potentially into a high-probability for scoring area) and it's a non-event statistically? OR, it bounces in, and it is both a shot and a goal?

This is "good luck" when it's out? and a "bad luck" when it's in? Crazy.

If a puck hits the post when I'm playing and doesn't deflect in, I'm not thinking, "Boy am I lucky", I'm thinking...I should have been in better position. If you asked me about luck, I'd say I was unlucky, if anything, as I'd rather have the additional save. Goaltending isn't as difficult as stats people like to make it. You're just getting hit with a puck after all.

R O said...

As a general rule (and I mean very general, obviously CG and a few other goalies here have a keen eye) goalies are terrible judges of how their games work.

Applies to players too. Seriously.

So going back to goaltending resumes etc. etc., is no defence of one's opinion.

R O said...

If bad luck didn't happen except to bad goalies, Roberto Luongo would have won ten cups by now.

Anonymous said...

How do you explain Huet's collapse over the past two seasons, especially this one?

The Contrarian Goaltender said...

I don't know about Huet, whether it is age catching up to him, whether he was just lucky in Montreal or what the reason is. I'd like to take a more detailed look at him at some point to see if I can find any indicators.

The Contrarian Goaltender said...

I don't see how stats guys are the ones making goaltending complicated. We're not the ones going on and on about intangibles and clutch play, mental toughness, poise, momentum and making the big save. But I have to agree that "you're just getting hit with a puck" is probably the best and most concise way to describe goaltending.

That quote accurately describes the reactive nature of goaltending. As much as us goalies like to think we're in control, really we're just the guy getting hit. And yes, the shooter on the other team does play a role in whether we get hit or not, which is where luck comes in, because luck implies factors beyond your own control. Sometimes it's because they beat you, i.e. shoot hard enough from close enough that you can't react, and sometimes it's because of a bounce or a break.

I agree that a lot of kids blame luck for their failures, but the problem is that they are making excuses, not that they are aware of the existence of randomness in the universe. In my experience, most of those kids are chronic complainers who will attribute their failures to anything external, whether it be bad luck, the defence in front of them, the refs, whatever.

You have to be uncommonly objective to be able to properly assess whether a goal was your fault or just bad luck, especially when emotions are high in an imortant game. For most people it's better to pretend luck doesn't exist. Pretend long enough and you'll start believing that's reality, but in my view it's still a made-up narrative.

Doesn't believing every goal is your fault open you up to poor decision-making? Take goals on unexpected caroms, like the one on Rinne last night or where a guy dumps it in and it bounces off the glass and goes into the net. The only way to guarantee those goals never happen is to never go out to stop dump-ins behind the net in the first place. But in the long run it is almost certainly better for the team to have their goalie retrieve pucks behind the net on routine shoot-ins, even at the cost of a fluke goal or two every season. If you stop retrieving pucks, you are costing your team. Deciding that it was your fault that the goal went in but that you aren't going to do anything differently next time doesn't make an awful lot of sense, and implies that your conclusion was incorrect in the first place.

The other issue is that people tend to see a bit of a false dichotomy. Almost nothing is either pure luck or pure skill. What's that old quote, success comes when talent meets opportunity? Look at Luongo's save on Smyth the other night. He made a great athletic play to get in front of it, but if Smyth makes a better shot the puck still goes in the net. Luongo's skill allowed him to make a save that most goalies probably wouldn't make, but he still got lucky that the shooter permitted him to make it. On any given play there will be elements of luck and elements of skill. In the long run the luck tends to balance out and the residue is skill.

Those last two sentences sounded like I was describing poker or something, but that is my outlook on it. You mentioned that I should be doing something that involves dice instead of goaltending, but to me they're pretty much one and the same. Only instead of one having six sides that can be held in your hand, the goaltending die is invisible and has somewhere around ten sides, nine marked save and one marked goal.

I am not saying that talent doesn't matter, or skill doesn't exist, or that goalies don't need to practice and prepare and work on their technique as much as they can. I'm just saying that luck certainly happens. If it doesn't happen, why doesn't Martin Brodeur have a save percentage of .915 in every game he plays? He just decides to play harder in some and worse in others? That really does not compute.

nightfly said...

As I recall, we've had a similar discussion before, Lawrence... but I am going to take your side this time on one very limited thing: during the game, I DO have the mindset that I am going to stop the next one. I have to have that mindset. I will not make excuses for something I feel like I should have had, and during the game I often feel like I should have had every one. I will very quickly note things I could have done better to improve my chances on getting pieces of shots that I miss.

That does not, however, mean that I can be expected to post a shutout every single time, and that if I don't I am a bad goalie. It's the difference between "feel" and "think." After the game I think it through, take the mental notes I made during the game about whatever was happening, and use it to refine my approach - but again, as you admitted, it's about improving my CHANCES for the next save.

"Luck" is maybe not the most precise word in every context. A lot of it is the skill of the other 11 guys on the rink. Sometimes shooters are going to make the perfect shot or you will be hopelessly screened and never see the shot. That's not making excuses, it's reality. But it's not really precise to say it was my failure so much as the other team's success.

If you don't think you have a chance on every shot, you really shouldn't be playing goal...maybe play something that involves dice.

This is where I reach the exact opposite conclusion. The illusion of 100% control over everything on the rink is the unrealistic, fantasy, video game mindset. If I don't like a dice roll I can always throw again. It's cheating, of course, but it IS 100% control.

Hell, I once steered a rebound to my defenseman, standing along at the side of the net, and he swatted it in with his glove. How the blazes am I supposed to stop that? I'd drive myself crazy if I took the blame for not being in position or leaving a bad rebound on something like that. I was in perfect position for the shot I stopped and sent the rebound to a completely safe area - my guy was the only player from either team within 20 feet.

Here's the other thing, though - I would drive everyone else crazy if I blamed him, too. It was literally "just one of those things," same as Rinne's carom, or Boyle's own-goal to lose game 3. Not "luck" per se, but he certainly didn't score an own-goal on purpose. So - no blame. He cusses a bit and I tell him to forget it. We move on. Otherwise we'll be miserable instead of enjoy the game - and incidentally, we'll also play like crap if we're at each other's throats over mistakes.

Lawrence said...

"shoot hard enough from close enough that you can't react"

"because luck implies factors beyond your own control"

According to Mama google "Luck: is something happening against the odds or by accident."

I still don't know how we debate this idea that luck has something significant to do with the game. The odds are 1 in 10 will go in, so when one it ten goes in, how is that bad luck?

If the puck is directly at the net, there was intention behind it, again, I'm not going to say that luck cannot exist, I just think it's so immeasurable that you might as well not consider it.

Sure when Steve Smith bounces the puck off Grant Fuhr that's accidental or against the odds that it bounces back into the net, but how often does that happen? Once in 5000 shots? That's nearly immeasurable.

"Take goals on unexpected caroms"
so you're telling me that a goal against on an unexpected carom is bad luck yes?

Then by your definition a save off an unexpected carom must be good luck correct? Or is it 'just another save?'

The way I look at it, if you believe in bad luck, untimely bounces, negatives against the odds, then by default you have to believe in good luck, timely saves and positives against the odds.

The difference in my mind is that when you move the puck toward the goal, you have the intention of it going in the goal, so if it does, no matter how it does, it's success with no measurable "by accident" involved because it's intended.

However, if you have a goalie play the way Jaroslav Halak did Monday night, he defied the numbers 52/53 - .981 and Montreal won a game they had no business winning. If you can't say that the saves he made on the 5 on 3 were more key events in that win than the dump in high glove which was likely going over the net with 30 seconds left, then there is a lack of some serious perspective. You're trying to tell me to ignore intention on "bad luck" goals, but at the same time discount the score, the time and the events in a game on timely saves? Are you serious? A goalie is required to have the same mental focus in a 12 shot against 10-1 winning scenario as a 1-1 tie in OT? Yeah right, think again. Brodeur may not consciously decide to try harder in games, but it's likely he does play harder in certain scenarios, such as those above.

Back to Jaroslav, even HAD Halak not made the save and the puck went over and behind the net, it's very very unlikely that event would have any change on the outcome. Further, if he had caught the puck, turned around and thrown it in the net, it's probably, not certainly, too late for Washington with 20 seconds to go.

I don't think you could say the same things would be true on the five on three. So, there is a hierarchy of importance to the saves, therefore one is "key" the other is not.

R O said...

This...

If the puck is directly at the net, there was intention behind it, again, I'm not going to say that luck cannot exist, I just think it's so immeasurable that you might as well not consider it.

... is just insane.

Luck is kind of one of the most measurable concepts there is. You can construct models based on "weighted coin-flips" (or more complicated probability distributions if you like) and compare to reality.

You can look at repeatability over time, which suggests a persistent effect. By correlating the performance of the population from one interval in time to another.

You can look at the effect of sample size, if there is "noise" (as it is classically defined, an unbiased and non-persistent effect) in the data then larger sample sizes will provide more information. A number of ways to do this, you can look at the distribution of single data points, vs. "smoothed" (by rolling averages or other filters if you prefer) data.

All of that is there, in goaltending, in huge quantities.

It's just a disingenious thing to say that goaltending is not affected by luck in a significant matter. To do so is to betray a fundamental misunderstanding about the game.

The Contrarian Goaltender said...

I think we have a different definition of luck. To me luck encompasses a lot more than just accidental or rare happenings. My definition is more in line with the #1 listing on freedictionary.com: "The chance happening of fortunate or adverse events." There is of course good luck as well as bad luck.

Luck does not have to be an unlikely or rare event. If you flip a coin 10 times, you will often get 6 or 7 of one side of the coin, either heads or tails. That's not an unusual outcome because the variance is pretty high in that small of a sample, but it was still entirely because of luck.

If a goalie is a .900 goalie and one in ten go in, then his luck is probably balancing out as he is just performing up to his skill level. But any .900 goalie can have a .920 save percentage over a playoff series or a few week stretch, and during that period he was simply getting lucky. Screen shots were hitting him, opposing shooters were missing the net, his team was defending scoring chances against unusually well, etc. Then maybe the next series he's at .875. His skill hasn't changed, but his results have.

I don't know how to explain those differences without invoking luck. A goalie doesn't turn into Dominik Hasek for two weeks and then turn into Dan Cloutier for the next two. There is variability in athletic performance, goalies have good games and bad games and so does the defence in front of them, but surely the results we see are much greater than the actual spread in individual performance. The only other possible explanation is that there are mental factors or a goalie is varying his effort level or something like that, things that we would never know conclusively and even if they were true (which most of the time I don't think they are), for purposes of analysis it's no different than attributing it to luck because you don't know when those factors will align the next time.

Regarding timely saves, I'm not saying there are no important saves. Goals that tie the game or give a team the lead have a significant impact on win probability. My argument is that nearly every save is a key save, and that therefore a goalie's performance on "key saves" closely reflects his overall performance on all saves by simple mathematics (since it makes up most of the sample). As a result, there is no reason to focus on them. I agree that Halak's last minute save didn't have a huge impact on Montreal's win probability. But nearly all of his prior 51 did. Even if you throw out the last save, the stats still show that Halak was terrific in game 6 (and, despite that, also a little lucky, because we know he doesn't have the skill level of a .981 goalie).

Bruce said...

A goalie doesn't turn into Dominik Hasek for two weeks and then turn into Dan Cloutier for the next two.

I dunno ... did you see Luongo in the 2009 playoffs?

(Kidding!)

As for Duncan Keith, I've had this guy in my fantasy league pool for the last three years, watch him pretty closely and know his game fairly well. His is a (phenomenal!) skating game, always moving, circling, stepping up into the play to cut off attacks before they develop. I have likened his style to "playing a shallow centre field". He cuts out a lot of singles, but every so often somebody drills an extra base hit over his head. In Keith's case that is breakaways, which he seems to get burned on more frequently than most defencemen, certainly most elite defencemen.

Surely an expectation from that style of play might be that the guy should have a very good shots ratio, but perhaps a less flattering Sv% ON. Which is what we do see in his case, at least in 2009-10.

All of which of course is anecdotal, and not enough information to say much more than "hmm, that's interesting".

Lawrence said...

"To do so is to betray a fundamental misunderstanding about the game."

"despite that, also a little lucky, because we know he doesn't have the skill level of a .981 goalie"

I clearly don't understand hockey then and need it explained to me in layman's terms, because attributing any variance on a performance to luck is thievery in my opinion.

What I hear is a bunch of statistics people talking about a concept that they cannot make the relationship to in real terms. Certainly, you can build a mathematical model to prove something and then assign a term to describe it, but how is it possible to prove that term is correct?

Say I am a 915% skill goaltender, I head into a game well rested, clear minded, I've eaten well, I feel great and I go in and let in one out of 30 - .967.

The next game is the next night, I couldn't sleep, I'm a bit tired, I don't feel well, add all the hyperbole you want...or don't, make it realistic if you wish, and I let in 5 on 25 shots. - .800

You're telling me the variable there is luck? Again, then this is semantic, and luck is not the best word for it. Maybe "Life" is. That's not luck, it's life. AND more importantly...it's poor decision making. If I'm not feeling well, I shouldn't be playing, and those poor decisions will affect my statistics.

Furthermore, I don't see how these factors would not only compound over a larger sample size. If you take a goalie like Kiprusoff who plays nearly every game, plays back to back, plays and puts up a .920sv%...how can his career not be more subject to "life" than Halak or Rask or Vernon in the 80's when goalies split duties nearly 50/50.

Aren't teams with two "arguable" #1 goalies at the advantage of choosing the one who feels the best prior to the game. (now, it's arguable that we don't know the selection process, but, we cannot assume that either way.)

I played every single game for my team this year, less one, because we don't HAVE a backup. There were certainly games where I would have sat on the bench if I could. If Kipper pays through games when he is sick, when he's fighting the puck, and plays poorly, that isn't LUCK. It's lack of having a back-up goalie.

It doesn't surprise me at all that Rask's sv% was unusually high, because he only played 45 games. So if you wait to see his sv% after his sample size is 5000 shots, and he only plays 45 games per year, you just have to wait longer than with a goalie who plays 73 games. Likely, the 45 game goalie is going to have better numbers.

Anderson is a good example. Is it any surprise that he has had worse stats this year (.917) playing 71 of 82 games than his last 4 years combined (.924) playing 68 of 332? Not for me. And that is a big difference in base skill. At the half-way point of the season he was first or second in sv%, playing a normal number of games, but then as he went on, the numbers dropped.

Attributing all of these factors to a catch all term such as 'luck' is blindness.

Lawrence said...

"The chance happening of fortunate or adverse events." There is of course good luck as well as bad luck."

Getting sick may be a chance happening, but playing when you are sick is a choice.

Being tired and playing back to back games is not a chance happening, it's a result, and a choice.

Being presented with a potential race for the puck out to the blueline is chance, going out and sliding on your belly is a choice.

It could go on and on...

Retaliation is a good one. Or flopping when bumped in traffic... all choices which are not repeatable with consistent frequency, but have lasting results.

The Contrarian Goaltender said...

Lawrence, I get what you are saying. I agree that we're mostly arguing semantics. I think you are correct that there are other variables at play here that we are lumping in under the banner of luck.

From my perspective, I don't see that as a problem, for the most part. Things like the amount of sleep a goalie gets the night before are variables that are out of his control anyway, just as much as it is out of his control whether that screen shot from the slot picks the corner or hits the glass. I think it can be reasonably argued that whether he sleeps well or not is a matter of luck. If we're trying to predict how a goalie is going to perform tomorrow, then it falls in with a bunch of other unknown factors that can't possibly be accounted for ahead of time.

My perspective is typically forward-looking, i.e. I'm trying to be able to predict future performance rather than explain past performance. That's why I tend not to get caught up in trying to analyze micro factors that may have had a real impact on a game here or there, but nobody can tell when they show up next.

Then again, sometimes we should look at those other factors to interpret performance. If a family member passes away, or a goalie is battling injury, or there are fatigue factors in play like you mentioned, then they should be considered in the analysis. In that case it might be a mistake to attribute all variance to luck alone.

I think you're a little quick to simply assume that a low number of games played equals a high save percentage, and vice versa. Rask's teammate Tim Thomas only played in 43 games, how come his save percentage wasn't so high?

I don't see fatigue effects for Craig Anderson, I see unsustainably high rates in Florida for the last two seasons. Anderson this year was pretty much what I expected. If fatigue was such a huge factor for Anderson, how come he stood on his head against San Jose in the playoffs after playing nearly all the games down the stretch?

I think more work needs to be done on studying goalie fatigue. For some guys in some situations it might be a factor, but assuming that it applies to everyone without even checking the details of their usage or their numbers might be as big of a mistake as not taking it into account at all.

Lawrence said...

Well, if we are arguing semantics I would propose a change in the term, because 'luck' seems to be less appropriate than 'intangibles'.

I agree it is otherwise impossible to measure the degree of how a goalie feels/prepares/eats/hydrates/dopes/stretches etc into the equation, but other than doping, I can tell you all those 'intangibles' affect my game, and I'm as human (but not as skilled) as the guys in the NHL.

I think if you look at the vernacular use of luck...it's attributed to aspects at such a low rate of chance that it seems outer-worldly.

Like winning the lottery (1 in 76,275,360.

or guessing correctly in trivia when you say "I have no idea the answer, but I'll guess, 38216."

or, never kicked a football in your life and kicking a 50-yarder to win 1 million dollars.

Then when someone says "Boy, Jaroslav Halak has been so lucky these last, now three games, to have stopped 130 of 134 shots." It, to me, takes away from the simple fact that although he has been playing at an unsustainable rate, he has also been very very good.

I know that there are intangibles at play (confidence, playing to the score, superstition, rest, voodoo..who knows the specifics) but let's not just discount it and say he's been lucky.

He's not playing with his eyes closed, or blind-folded, or trying a new technique where he faces the opposite direction.

For me, I feel like we say "luck" because it takes away from the player. Ie. we know the vernacular usage and use it because it says...don't kid yourself, you've just been lucky, we've measured and you'll come back to Earth.

as opposed to using "intangibles" because this term takes away from the person doing the measuring. It forces statisticians to say "I have no idea how that is happening, but it's happening, and there is no explanations." And I think that bugs the hell of of people who like to measure things.

You said it yourself, the numbers show that the best thing would have been for Washington to have gotten less shots on Halak, which makes no sense at all, except by the numbers.

Lawrence said...

One last aside.

Seeing the Avs eliminated from the playoffs filled Calgary fans, like myself, with endless glee.

But unlike myself, the sentiment was "HA! we KNEW your luck was going to run out!"

but we had been saying that since game 10. Oh, wait till your luck runs out...we'll be in the playoffs. Then game 30, then 40, then at game 76, we're all saying "COME ON, they CANNOT be lucky like this forever, we have to be in the playoffs."

Never happened.

So now, they lose to the top ranked team in the WC and we say...HA HA, I KNEW it, they were lucky all along.

Nope. They played the best team in the West, they were out skilled, not unlucky. In fact, they were still pretty damn lucky.

But hey, whatever heals the wounds I guess.