Wednesday, September 17, 2008

The Most Underrated Goalie in the League

For my money, the most underrated goalie in the NHL is Kari Lehtonen. He is yet another example of a very good goalie getting overlooked because of the poor team around him.

Lehtonen has gone 75-61-14 on Atlanta over the past 4 seasons, with an impressive .913 save percentage and a 2.82 GAA. That is much better than his backups have done (.890, 3.31). His save percentage is trending upwards - .906 in 2005-06, .912 in 2006-07, and .916 in 2007-08 even though the team took a substantial step backwards (although their defensive play probably improved). Hockey Numbers had him a solid 9th in the league this past season shot-quality neutral save percentage, and in 2006-07 Alan Ryder ranked Lehtonen as the 3rd most valuable goalie in the league behind only Luongo and Brodeur.

Lehtonen has posted very consistent even-strength save percentages - .930, .926 and .927 over the past three seasons, according to NHL.com. That level of even-strength performance puts him in a class with Kiprusoff, Giguere, Thomas, Luongo, Lundqvist, Vokoun and Brodeur, i.e. the league's elite goalies.

Given those impressive numbers, it is curious how poorly he is rated around the league. The Hockey News did a ranking of the top 30 goalies last spring, and gave Lehtonen the final spot in their ranking, #30, behind every other starting goalie in the league. There was also a recent series of polls on HFBoards asking fans to rank the starting goalies, and Lehtonen came out in just 20th place. I think Lehtonen is probably a top-10 goalie in the league today.

Atlanta will probably be a weak team again this season, but given his recent history and young age (24), expect Lehtonen to continue his strong play and gradually gain more recognition as one of the top young goalies in the league.

19 comments:

Bruce said...

CG: While I agree there is no way Lehtonen is anywhere near the worst starter in the league, in the eyes of many hockey fans he hasn't yet lived up to the hype of being a #2 overall draft pick, largely due to playing on a crappy, seldom-televised team in a lousy hockey market, exacerbated by his own injury woes and a disastrous first playoff performance in 2006-07 (0-2, 5.59, .849).

Far be it from me to accuse you of cherry-picking, but let me point out an arror in the stats you presented.

Lehtonen has gone 75-61-14 on Atlanta over the past 4 seasons, with an impressive .913 save percentage and a 2.82 GAA. That is much better than his backups have done (.890, 3.31).

Those stats are accurate for his backups of the past 3, post-lockout seasons. The above numbers include Lehtonen's very impressive cup o' coffee in 2003-04, when he went 4-0-0, 1.25, .953, and raised the hype bar considerably further. However the other Atlanta goalies had moderately good numbers in 2003-04; including all four seasons they posted a combined .895, 3.11, lower than Lehtonen but not quite the gap you portray. Coincidentally over those four seasons Lehtonen recorded 164 decisions (.547) and his predecessors/backups -- a motley crew including Pasi Nurminen, Johan Hedberg, Michael Garnett, Byron Dafoe, Mike Dunham, Adam Berkhoel, Ondrej Pavelec,and Steve Shields -- had the other 164 (.486). Let's just say I would hope that Lehtonen is better than those guys. Whatever else on his horrible record, Don Waddell certainly drafted to need when he picked Lehtonen.

But to compare apples and apples, let's just look at the post-lockout era when Lehtonen has been the de facto #1, playing the lion's share of games when he's been healthy.

Lehtonen: 154 GP, .534, 2.86, .912
Backups: 119 GP, .535, 3.31, .890

Lehtonen's GAA and Sv% suggest he's been a little unlucky in posting the same Pts% as the other guys. However on further examination that Sv% appears to be a little misleading:

SoG/60
--------
2005-06
Lehtonen: 31.1
Backups: 29.1

2006-07
Lehtonen: 31.6
Backups: 28.4

2007-08
Lehtonen: 34.6
Backups: 31.9

Totals 2005-08:
Lehtonen: 32.4
Backups: 30.1

... and we see a clear trend where Lehtonen consistently faces 2-3 more shots per game than his backups, every year. How can this be? I thought goaltending was all about stopping the puck, and that the goalie himself had no effect on shot quality or quantity facing him? Hmmm ...

I haven't seen near enough of Lehtonen to judge his peripheral game, but my first guess is rebound control. Likely short rebounds that result in the occasional whack-whack into the goalie's pads, which generally don't hurt his GAA but have the effect of unnaturally inflating his Sv% compared to a more efficient goalie like, say, Johan Hedberg. For this reason I would place less value in his Sv% vs. his teammates and be more willing to accept GAA as a true barometer.

Suggesting Lehtonen is not 20% better than his teammates (Opposition Sh% 8.8% vs.11.0%) but "only" 14% better (2.86 vs. 3.31). He's definitely #1, but the difference is not so great as you might think if you Only look at the Sv% metric.

Put another way, if Lehtonen achieved the exact same GAA, but with the same SoG/60 as his creasemates, his adjusted Sv% would be just .905, a difference of -.007.

Given that all goalies in this study played on the same team, and that there was a significant sample size for both Lehtonen and his backups, I would suggest that the statistics support my ongoing case that goaltenders' style of play is an important factor in determining shots allowed.

An early theory -- with MUCH work to be done -- is that the guys whose method is to take the bottom of the net away are the ones who allow, and don't particularly mind, the short rebounds. As a general rule they are not aggressive or accomplished puckhandlers; they "just stop the puck". Among the names that spring to mind of guys who play this dam-the-bottom-of-the-net style include Dominik Hasek, Roberto Luongo, and J.-S. Giguere, "coincidentally" the top three career Sv% leaders.

Anonymous said...

"Lehtonen consistently faces 2-3 more shots per game than his backups, every year. How can this be? I thought goaltending was all about stopping the puck, and that the goalie himself had no effect on shot quality or quantity facing him? Hmmm ... "

Actually Bruce, backups usually face less shots per game than the starter. I've noticed this trend/fact for years -- check seasonal save pct's for the past 25+ yrs.

The reason? No, not because the backup has special skills that reduce shots... instead, backups face less shots/game due to the coach putting them in net vs. the weaker teams.

You say you haven't seen enough of Lehtonent to judge his peripheral game... exactly... who among us is able to watch most/all of the NHL schedule each year, & then carefully document "rebound control", goalie style of play determining shots allowed, etc.?

The Contrarian Goaltender said...

Interest points, Bruce. Here comes a long post without any answers, but with some points of discussion.

I think I am confident enough in the numbers to dismiss your theory of rebound control explaining the gap. The Forechecker ran the data for 2006-07, looking at every time a shot occurred within 5 seconds of another shot without a stoppage in between. The per-game average of such rebound shots was 3.15 for both teams, i.e. about 1.6 per team per game. So the shot gap between Hedberg and Lehtonen is greater than the average per game total.

Atlanta ranked almost exactly at average in terms of allowing rebounds on shots, giving up 136 rebounds on 2577 shots for a 5.28% rebound allowance percentage (which works out to 1.7 rebounds shots per game). Lehtonen faced over 80% of Atlanta's shots that season, so he was at worst slightly below average in terms of allowing rebounds. That was the season with the biggest gap between Lehtonen and his backups in terms of shots against.

Secondly, one of the things we need to distinguish between is whether the gap is because Lehtonen "creates" shots or his backups are good at preventing them. Johan Hedberg has seen fewer shots per game than his teammates in 6 out of 7 NHL seasons, and his average is nearly 1.5 shots per game lower than the goalies he plays with. Take out Lehtonen, and he's still nearly a full shot per game better than everyone else.

2000-01: Hedberg 27.9, Others 29.2
2001-02: Hedberg 28.6, Others 29.5
2002-03: Hedberg 29.8, Others 31.4
2003-04: Hedberg 25.1, Others 26.6
2005-06: Hedberg 26.2, Others 24.9
2006-07: Hedberg 28.4, Others 31.6
2007-08: Hedberg 31.9, Others 34.4

With respect to the 2007-08 season, I think the gap is somewhat overstated because Atlanta clearly got worse defensively as the year went on, and Lehtonen started proportionately more games late in the season.

Pre All-Star:
Lehtonen: 26 GP, 9-12-2, 3.03, .908, 32.9 SA/G
Hedberg: 25 GP, 11-10-2, 3.30, .896, 31.8 SA/G

Post All-Star:
Lehtonen: 22 GP, 8-10-3, 2.76, .924, 36.5 SA/G
Hedberg: 11 GP, 3-5-1, 3.83, .882, 32.3 SA/G

Behind the Net also posted his measurement of shot quality both midway through the season and at year-end. In his first post, he estimated Atlanta's expected save percentage to be about .913 after 30 or so games played. At season's end, he estimated it to be .904, meaning Atlanta's defensive quality must have nosedived late in the season. Also, he measured the expected save percentage against as .904 for both Lehtonen and Hedberg, which does not imply that the extra shots being given up by Lehtonen were more dangerous chances (which would be the case if the additional shots were all 5-10 foot rebound chances, for example).

More interesting tidbits from Behind the Net:

For 2006-07, there is a breakdown of shots, misses and saves by game situation. At even strength, Lehtonen faced 1.8 more shots than Hedberg per 60 minutes. On the penalty kill, Lehtonen faced 10.6 more shots per 60 minutes than Hedberg did. That is a huge gap that can probably mostly be described as a statistical fluke (Hedberg played just 116 minutes 4-on-5). However, it had an impact on the shot differential.

In 2007-08, it is similar: Lehtonen faced 45.7 shots per 60 at 4-on-5, while Hedberg faced 39.2 shots per 60. For some reason, opponents fire away at Lehtonen more than they do against Hedberg with the man advantage. I would have thought that a goalie shot prevention effect would be revealed at 5 on 5, not 4 on 5, but maybe we need to look into that a little closer.

And lastly we come to my hypothesis, which is that bad teams play more aggressively with their #1 goalie in the net. From Behind the Net, the Thrashers took 22.4 shots per 60 minutes at evens with Lehtonen and 21.7 shots per 60 with Hedberg in 2007-08. That's a gap of 0.7. The Thrashers are a bad team, so they are going to get outshot by even more in an open game. If we take Hedberg's shot for/against ratio of 1.3, that means that an extra 0.7 shots for by Atlanta would probably result in at least an extra 0.9 shots against per 60 minutes. I intend to check whether starters on bad teams do tend to face more shots or not in the general case, but if so it could explain some of Luongo's gaps as well.

In any event there are several factors at play here - the shot differential can't just be neatly attributed to one thing like rebound control. But the issue of goalie shot prevention is no doubt a rich topic of research and discussion, so it's something that I'm planning to look at in more detail and I know you will as well Bruce, and who knows where we will end up.

Bruce said...

//have the effect of unnaturally inflating his Sv% compared to a more efficient goalie like, say, Johan Hedberg//

Secondly, one of the things we need to distinguish between is whether the gap is because Lehtonen "creates" shots or his backups are good at preventing them.


Thus my use of the term "more efficient". Hedberg, a mobile retriever and excellent puckhandler in my experience, plays the sort of game that I would expect to reduce shots. Comparing Hedberg to Lehtonen would yield some of the same talking points as comparing Brodeur to Luongo; they are from a whole different "family" of goaltender.

Johan Hedberg has seen fewer shots per game than his teammates in 6 out of 7 NHL seasons, and his average is nearly 1.5 shots per game lower than the goalies he plays with.

Interestingly, the one year Hedberg faced more shots, his creasemate was none other than Marty Turco, an extremely efficient sweeper who I consider the best puckhandler in the game. Hedberg (26.2) was good, Turco (24.9) even better.

Take out Lehtonen, and he's still nearly a full shot per game better than everyone else.

Take out Turco, and he's two shots per game better than all those other guys he played with.

In any event there are several factors at play here - the shot differential can't just be neatly attributed to one thing like rebound control.

Yes of course it's not that simple. I chose my words poorly when I said "my first guess is rebound control", I didn't mean that to explain the whole gap but simply to be the primary/an important factor. Your data have not fully dissuaded me from that position, but as always it's more complex than that.

But the issue of goalie shot prevention is no doubt a rich topic of research and discussion, so it's something that I'm planning to look at in more detail and I know you will as well Bruce, and who knows where we will end up.

Who knows, indeed. I look forward to the discussion.

Bruce said...

With respect to the 2007-08 season, I think the gap is somewhat overstated because Atlanta clearly got worse defensively as the year went on, and Lehtonen started proportionately more games late in the season.

Pre All-Star:
Lehtonen: 26 GP, 9-12-2, 3.03, .908, 32.9 SA/G
Hedberg: 25 GP, 11-10-2, 3.30, .896, 31.8 SA/G

Post All-Star:
Lehtonen: 22 GP, 8-10-3, 2.76, .924, 36.5 SA/G
Hedberg: 11 GP, 3-5-1, 3.83, .882, 32.3 SA/G


Well, the Thrashees got "more worse" in front of Lehtonen (+3.6 shots) than Hedberg (+0.5 shots). But to remove the proportional bias, let's just take those splits as "equal" halves of the season for both guys and do a simple average of the averages:

Lehtonen: 32.9 + 36.5 / 2 = 34.7
Hedberg: 31.8 + 32.3 / 2 = 32.0

... which is virtually unchanged from the raw data of 34.6 and 31.9 respectively. There's clearly a difference between the two that is going to be difficult to explain away. Especially given it's not just one season.

As for the PK data, it's interesting, but if it's consistent year-over-year it suggests that the differences in style play a bigger role 4v5 than they do at evens. I must admit that's an area of goalie stats I have never gotten into in the past but between NHL.com and Behind the Net (not to mention this site) I have some good resources going forward.

The stuff from the Forechecker is interesting too. I wonder if there is anything in any of this data that might allow us to pick out "first-shot" strategies for defensive systems; i.e. some play for shot prevention while others don't mind allowing the first shot, count on the goalie to stop it and focus on clearing the rebound. Which wouldn't show up in rebound data but would certainly factor into gross shots against counts. It'll be hard to quantify, but it's hard to imagine that wouldn't have an effect on Sv%. Which is why I keep coming back to GAA as the ultimate measuring stick of team defence, including its goaltending.

I've been having this same discussion more from the skaters' perspective with some of the Statzis of the Oilogosphere who are pushing shots-for-and-against data like Corsi numbers. And I keep coming back to this: shots (and shot prevention) are a means to an end, but that end is the actual goals that go on the scoreboard. There are different strategies to accomplish success, and shot statistics are useful but deceptive. Above and beyond that, the true end that justifies all those means, is Wins. The team that outshoots doesn't necessarily outscore, to put it mildly.

Also, I wonder what else can be gleaned from the shot-aftermath stuff like what the Forechecker strips from the game sheets. What percentage of saves are accompanied by a whistle? For those saves that result in a continuation, what percentage of "next events" (whatever they may be) occur in the defensive zone? Seems logical that the teams that just clear the rebound into the corner wind up hemmed in their zone longer than teams that recover the rebound that the goalie has already directed into the corner with time and space to move it. Would there be sufficient signal in the noise to suggest some goalies put the rebounds in less dangerous places than others? Idle speculation at this point, I'm just saying these might be interesting angles to pursue.

he measured the expected save percentage against as .904 for both Lehtonen and Hedberg, which does not imply that the extra shots being given up by Lehtonen were more dangerous chances (which would be the case if the additional shots were all 5-10 foot rebound chances, for example).

Nor did I mean to imply those were particularly dangerous chances. I'm talking about two-foot rebounds which usually wind up going right back into the goalie because there's no room to get the puck upstairs. It's a style that works; it just tends to inflate Sv% is all. A goalie who has a problem with Bad rebounds will see his Sv% take a hit, for sure.

Bruce said...

backups face less shots/game due to the coach putting them in net vs. the weaker teams.

Anon: That might apply to a coach of a strong team, but that on a weak team tries to get all the points he can, and is more likely to throw his backup as a sacrficial lamb against a powerhouse than to try to play his weaker goalie in a game he can actually win. In any event there's nothing in Atlanta's record that I can see that suggests their coach had any such strategy (or any strategy at all that I can see ... what a brutal team). Certainly Hedberg got a lot of work in late October and November when Lehtonen was hurt (again), and he played all comers. Atlanta was playing a little tighter defence at that point of the season, but whether Hedberg was a passive beneficiary of that or a contributor, well we could argue that 'til the cows come home. All I can count on for sure is whatever side of the discussion I might take, you'll line up on the other side of the fence.

For the record, 18 of the 32 games in which Hedberg got a decision were on the road, while 25 of 44 of Lehtonen's decisions came in the not-so-friendly confines. The other effect I checked for was who got the lion's share of divisional games in the high-flying SouthEast, but Lehtonen got just 16 of the 32 decisions against the SE, Hedberg 12 and Pavelec 4, so pretty much porportionate. Nothing in there that I could see that suggested Lehtonen got the tougher games. On that club, they're all tough games.

Anonymous said...

"...on a weak team tries to get all the points he can, and is more likely to throw his backup as a sacrficial lamb against a powerhouse than to try to play his weaker goalie in a game he can actually win."

Interesting premise. I always thought a team, regardless of strength, would play the best goalie against the best (or, most high-scoring) teams, unless that goalie is very tired or injured... basically, to approach each game as a separate event & to maximize the chance to win each game.

Anyway, the general trend I've noticed from the stats is that the goalie that plays the most minutes on each team tends to face more shots/60 than the other goalies on the team. Not a hard & fast rule, & injuries & other factors affecting goalie availability come into play, but a general trend I've observed.

The Contrarian Goaltender said...

There's clearly a difference between the two that is going to be difficult to explain away.

I agree that there appears to be a difference between those two. I am more interested in Hedberg's performance than Lehtonen's, because he is a journeyman goalie and it looks like his apparent abilities in shot prevention are what are keeping him in the league. If he was facing Lehtonen's shots his GAA would be getting pretty close to 4 and I'm not sure any GMs would be returning his calls (although Colorado did sign Andrew Raycroft so who knows?)

Nor did I mean to imply those were particularly dangerous chances. I'm talking about two-foot rebounds which usually wind up going right back into the goalie because there's no room to get the puck upstairs. It's a style that works; it just tends to inflate Sv% is all. A goalie who has a problem with Bad rebounds will see his Sv% take a hit, for sure.

To a computer calculating shot-quality, a two foot rebound shot is recorded as something like a 30-40% chance of a goal (check out Behind the Net's scoring probability chart). So if Lehtonen was giving up, say, 2 extra two foot rebounds per game compared to Hedberg with everything else being the same, that would give him an extra 0.6 or so expected goals per game. That would have a major impact on his shot quality against and would make his shots against look much tougher than Hedberg's. Which is why I think those shots either aren't happening or aren't being counted as shots by the official scorer.

I'm a butterfly, take-away-the-bottom-of-the-net goalie, which you probably have already guessed, and I think the frequency of those chances is overstated. I agree it makes logical sense that they happen more with someone like Lehtonen or Luongo than someone like Brodeur, but Brodeur probably has a fair number of them as well so once again we are talking about margins, and I'm not convinced there is a significant difference. I'm also not sure that the official scorer actually would count it as multiple shots if a guy is just jamming the puck into the pad. Can you find, say, three shots back-to-back-to-back from close range over the span of a couple of seconds anywhere on an NHL gamesheet? Also, in goalmouth scrambles the puck can go anywhere, and I think it is unfair to say that someone like Lehtonen is virtually guaranteed to stop all the very short rebound shots he faces. Sometimes it goes back into the goalie, sometimes the rebound shot creates another more dangerous rebound, and sometimes NHL shooters roof it even with no room to maneuver. Rebound chances still have a very high scoring rate and account for a large portion of goals, so I think the harmless short rebound into the pads is a relative rare event compared to other more dangerous rebound chances.

Do you think the Forechecker's numbers seem out of line (1.5 rebound shots per game)? Those numbers are, again, based on the shot totals that NHL scorers are recording. Now most rebounds likely get cleared and don't lead to an immediate shot against, obviously, but they don't directly impact shots and goals which is what we are discussing here.

Would there be sufficient signal in the noise to suggest some goalies put the rebounds in less dangerous places than others?

I'd guess it would be pretty difficult to pull that out from the gamesheets. If there was a measure of time in the defensive zone, like they have on hockey video games, that might be a better way to differentiate between goalies who are thwarting the opposition through puckhandling or rebound control or the like from others who are keeping the attack going by giving the puck away. There would be a lot of noise in that one as well, and you couldn't isolate whether the root cause was rebound control or puckhandling or whatever, but I'm not sure we'll ever be able to do that with confidence except through extensive game charting.

The team that outshoots doesn't necessarily outscore, to put it mildly.

Yeah, I've seen a few of those discussion threads and I thought you did pretty well in that debate actually. Although I don't completely agree, because I think the missing variable is shot quality. The team that outshoots doesn't necessarily outscore, true, but the team that outchances is actually very likely to outscore, something in the neighbourhood of 75-80% of the time, based on work done by Alan Ryder and Chris Boersma, and the work I've done also supports that range. Chances here would be defined as expected goals from the shots taken during the game.

Bruce said...

//There's clearly a difference between the two that is going to be difficult to explain away.

I agree that there appears to be a difference between those two. I am more interested in Hedberg's performance than Lehtonen's, because he is a journeyman goalie and it looks like his apparent abilities in shot prevention are what are keeping him in the league. If he was facing Lehtonen's shots his GAA would be getting pretty close to 4 and I'm not sure any GMs would be returning his calls (although Colorado did sign Andrew Raycroft so who knows?)


Yeah, that’s the same point in the other direction. Despite the huge gap in Sv% Hedberg’s performance is closer than it appears where it really matters, GA (and I’ll try not to get started on Wins :). I tried to make a similar point earlier this year when I compared Ed Belfour’s season in Florida to those of the star goalies that sandwiched him:

Luongo, 2005-06: 35-30-9, .534, 2.97, .914
Belfour, 2006-07: 27-17-10, .593, 2.77, .902
Vokoun, 2007-08: 30-29-8, .507, 2.68, .919


Something does not compute; how could Belfour be so lucky?

SA/60:
--------
Luongo 34.7
Belfour 28.3
Vokoun 30.3


And maybe you still ask how could Belfour be so lucky to play on such a better version of the Panthers, or maybe you start to wonder what he might have done to affect that total. Funny thing: if I were to rate them as puckhandlers, I would say Luongo not-so-good, Belfour among the best, Vokoun in the middle. Coincidence? Or not?

Belfour had a way better year there than he was given credit for, and the fact he is no longer in the league is very possibly due to people putting too much stock in Sv% as opposed to everything the goalie does out there. But that’s my bias. I know Eddie the Eagle was disappointed to play in the SEL last year, claiming he was still good enough to be an NHL #1. I tend to side with him on that score. Belfour's greatness was underrated his whole career right through to this day. Part of it he brings on himself by being such an unlikeable sumbitch that nobody wants to give him much credit, but it's that combative attitude that made him the great goalie he was. In that way he reminds me of Terry Sawchuk, who for whatever reason is defintely NOT underrated.

Returning to those three equally-drab Jacques Martin coached clubs, Belfour’s better Pts% stands out. It can be attributed somewhat better scoring support as that Florida squad scored 5 goals above the league average, whereas Vokoun and Luongo’s seasons were -12 and -13 respectively. The question of whether the Panthers scored more goals because they spent a little more time in the other team’s end suggests itself, but I quickly dismissed it having bathed in the doctrine that goalies couldn’t possibly affect offensive output. Could they? :)

To a computer calculating shot-quality, a two foot rebound shot is recorded as something like a 30-40% chance of a goal.

Does the computer keep track of where the goalie is? If he's stretched from post to post like a two-foot-high brick wall I don't like the chances of that two-foot shot. I must admit I’m thinking primarily of Hasek, who was the absolute master of the type but many of whose innovations, like Wayne Gretzky’s, will leave the game with him. The guy would go from vertical to horizontal in what seemed like Planck time, between opposition touches of the puck, and they could whack away all day and that puck would never find a hole.

I'm a butterfly, take-away-the-bottom-of-the-net goalie, which you probably have already guessed, and I think the frequency of those chances is overstated.

Light bulb mode [on]. I’ve been expressing this in the wrong terms. Your style of goaltender generally plays deep in the net, often building a dam right across the goal line. In a sense you invite play right in close to the goal and simply try to take the net away. A successfully thwarted scoring opportunity is most likely to be a shot on goal, and a save. Whereas a challenge-style goalie is more likely to thwart the play at the point of attack, to force the missed shot or deflect that extra pass out of harm's way. Success, just no Save. The important count isn't of the extra shots, it's of the first one.

I agree it makes logical sense that they happen more with someone like Lehtonen or Luongo than someone like Brodeur, but Brodeur probably has a fair number of them as well so once again we are talking about margins, and I'm not convinced there is a significant difference.

Yes of course, everything under discussion is in the margins. But when you’re talking about the margin between, say .913 and .919, that’s where you’d expect to look.

Do you think the Forechecker's numbers seem out of line (1.5 rebound shots per game)? Those numbers are, again, based on the shot totals that NHL scorers are recording. Now most rebounds likely get cleared and don't lead to an immediate shot against, obviously, but they don't directly impact shots and goals which is what we are discussing here.

Seems low but again, I don’t know. I can include rebound results in any in-game analysis that I hope to undertake this season. Got a nice new big screen HDTV with a handy-dandy 15-second recall button that can be reset at the next commercial, too. Some nights (when the Oilers aren’t playing) I’m going to try to keep track.

//Would there be sufficient signal in the noise to suggest some goalies put the rebounds in less dangerous places than others?//

I'd guess it would be pretty difficult to pull that out from the gamesheets. If there was a measure of time in the defensive zone, like they have on hockey video games, that might be a better way to differentiate between goalies who are thwarting the opposition through puckhandling or rebound control or the like from others who are keeping the attack going by giving the puck away. There would be a lot of noise in that one as well, and you couldn't isolate whether the root cause was rebound control or puckhandling or whatever, but I'm not sure we'll ever be able to do that with confidence except through extensive game charting.


Now you’re talking my language, CG! I do agree it will be hard to single out. For one thing it depends on whether the goalie is kicking the puck into Nicklas Lidstrom’s corner or Colin White’s. Nowhere ever will you find me arguing that the goalie has the primary effect on the flow of play, but the day I say he has no effect is the day I will have lost my faculties. The truth, as usual, lies somewhere between 0 and 1.

//The team that outshoots doesn't necessarily outscore, to put it mildly.

Yeah, I've seen a few of those discussion threads and I thought you did pretty well in that debate actually. Although I don't completely agree, because I think the missing variable is shot quality. The team that outshoots doesn't necessarily outscore, true, but the team that outchances is actually very likely to outscore, something in the neighbourhood of 75-80% of the time, based on work done by Alan Ryder and Chris Boersma, and the work I've done also supports that range. Chances here would be defined as expected goals from the shots taken during the game.


Thanks. I agree absolutely on the shot quality point that has been a central plank in my position, although obviously I have made a hash of explaining myself. I would extend it further to consider the shot quality OF the scoring opportunity, not all of which are created equal. Like shots, or attempted shots, a simple count of 0s and 1s for scoring opportunities will only take us so far. But even that would beat the current database, in which scoring opportunities are a closely-guarded state secret.

In the Great Corsi Debate I have stated unequivocally on two or three occasions that shots taken against the flow of play are, by and large, better scoring opportunities. A simple but strong point in favour of this which I have not previously mentioned is the peculiar fact that the league-wide shooting percentage is higher when shorthanded than when at even strength, or so it was in 2007-08; it also seems a trend across the various Sv%-by-game-state examples that you have provided.

I’m interested to hear your thoughts on the above, especially from the perspective of an active goaltender. Don’t feel obliged to be contrarian!

Anonymous said...

Chicago 1992:

GP, Min, Ptspct, GAA, Svpct, SA/60

Belfour - 52-2928-.531-2.70-.894-25.4
Hasek - 20-1014-.700-2.60-.893-24.4
Waite - 17-877-.400-3.69-.844-23.7
Leblanc - 1-60-1.000-.955-22.0

Was Hasek better with the soft skills than Belfour (24.4 < 25.4)? Wasn't Hasek supposed to be terrible at shot prevention, & had an inflated svpct as a result?

Looks to me like the starter (Belfour) played better/more offensive teams, & the backups played the poorer/less offensive teams.


http://www.hockey-reference.com/teams/CHI/1992.html

Bruce said...

Anon: Interesting. More examples welcome (both ways, if you're up to it).

Anonymous said...

http://www.hockey-reference.com/teams/TOR/2006.html

Belfour - 49-2897-.500-3.29-30.6-.892
Tellqvist - 25-1399-.478-3.13-29.9-.895
Aubin - 11-677-.909-2.22-29.2-.924

Anonymous said...

http://www.hockey-reference.com/teams/CHI/1991.html

Belfour - 74-4127-27.4
Hasek - 5-195-28.6 [Keenan; 39 min's per appearance]
Cloutier - 10-403-26.1 [40.3 min's per appearance]

Anonymous said...

http://www.hockey-reference.com/teams/BUF/1999.html

Hasek - 54-3817-30.3
Roloson - 18-911-29.5
Biron - 6-281-25.6

http://www.hockey-reference.com/teams/BUF/1996.html

Hasek: 3417-35.3
Trefilov: 1094-36.2
Blue: 255-32.2
Shields: 75-25.6
Biron: 119-32.3

http://www.hockey-reference.com/teams/BUF/1997.html

Hasek: 4037-32.4
Shields: 789-34.0
Trefilov: 159-37.0

http://www.hockey-reference.com/teams/BUF/1998.html

Hasek: 4220-30.5
Shields: 785-31.2

*Just a few seasons... my general impression has been that usually backups have fewer SA/60... of course, those that play few min's may have some outlier #'s, where the SA/60 are very high or low.

The Contrarian Goaltender said...

Don't worry about the terminology, Bruce, because that isn't a problem at all. I've been a hockey fan since before the Allaire butterfly block took over the NHL, so I'm familiar with both older and newer ways of doing things ("On Goaltending" by Jacques Plante doesn't exactly teach you how to play like Luongo). I've been trying to make up for my lack of explosive athleticism through an efficient position-based game since I first put on a mask, so I prefer to make saves on my knees, but I will intercept passes or throw caution to the wind and try to beat an opposing forward to the puck every now and then. We don't disagree that goal prevention is the objective of a goalie, but rather on whether shot stopping success is a good measure of a goalie's contribution to team goal prevention.

Does the computer keep track of where the goalie is? If he's stretched from post to post like a two-foot-high brick wall I don't like the chances of that two-foot shot. I must admit I’m thinking primarily of Hasek, who was the absolute master of the type

I agree that Hasek closed up shop pretty effectively along the ice, but the point was simply that the shot quality formula would assign a high expected goal number to any short rebound shots listed in the NHL game logs. That Lehtonen and Hedberg have a similar shot quality implies that it is unlikely that one of them allowed a lot more short rebounds than the other, since that would have resulted in a very different shot quality against figure.

In the Great Corsi Debate I have stated unequivocally on two or three occasions that shots taken against the flow of play are, by and large, better scoring opportunities. A simple but strong point in favour of this which I have not previously mentioned is the peculiar fact that the league-wide shooting percentage is higher when shorthanded than when at even strength, or so it was in 2007-08; it also seems a trend across the various Sv%-by-game-state examples that you have provided.

I think this seems reasonable, however I'd caution against overestimating the impact. I'd guess chances against the flow of play are more dangerous, but probably not by a whole lot. I looked at the 9 available years of even-strength and power-play save percentages, and the average shooting percentage was 8.27% at even-strength and 8.34% while shorthanded.

Even when a team is getting outplayed and outshot, that doesn't mean all of their chances are going to be against the flow of play. They could have a line or two with a favourable matchup that is holding their own and getting chances, or they could scoring goals on the power play, etc.

We've seen (and played in) games where one team controls the play, takes a lot of outside shots without much success, and gives up a few odd-man rushes that end up in goals in a 2-1 loss despite a 40-20 shot margin. In that game, there is no question that the shots against were more dangerous. I've seen expected goal numbers or calculated them myself for games where the shots were something like 35-20 but the team that was outshot actually had more expected goals. Still you have to be careful with generalizing, because there are so many different team situations.

It's like shots on goal - most people think it is logical that either more shots on goal means that every shot will be easier to save (i.e. it is easier to post a higher save percentage), or that more shots on goal is tougher and it will lead to a lower save percentage. There are subjective arguments to be made for either position, and lots of specific game examples that could support your viewpoint, but when you plot the numbers it turns out to be way more scatterplot than straight line, and there is really no overall relationship between the two things(*). For some teams there is a relationship, for example they take a lot of low-quality shots, for instance, and give up few shots but of higher-quality against. But another team is exactly the opposite, and another team could take a lot of high-quality shots and give up the same against, etc., etc., so you can't generalize.

I would suspect that the shots against the flow of play effect would show up most often in severe mismatches, where one team has a sizable territorial advantage, not just a shot advantage (like on the power play, for example, as you alluded to). And in that case, the quantity of the scoring chances of the better team could still easily make up for the better quality chances that they allow, in which case it may not have such a decisive effect on winning or losing. But I haven't looked at that specific issue in much detail, so I'd need to see some good evidence before I would fully support you in that claim.

By the way, I have a couple of posts in the works that take a closer look at the general issue of shot prevention/shots against. It's mostly just spitting out numbers, so if you spot anything that I missed please let me know.

(*) I am talking about today's NHL with the current state of parity. The strength of the correlation between shots against and save percentage has fluctuated somewhat. However, I've never seen it come out that more shots makes for significantly easier saves - in periods of low parity (e.g. the 1970s) the opposite is actually true, that more shots means it is actually more likely the goalie has a low save percentage, because the reason for the extra shots against is not from style of play but because the team has poor defensive talent. However, in general there is no relationship between shots against and save percentage.

Bruce said...

Florida 2005-06
------------------
Luongo: 4305 MP, 34.7 SoG/60
McLennan: 678, 31.9

Florida 2006-07
-------------------
Belfour: 3289, 28.3
Auld: 1471, 29.7
Anderson: 217, 32.1

Florida 2007-08
-------------------
Vokoun: 4031, 32.9*
Anderson, 935, 34.3


Note: In my post well up the page I screwed up Vokoun's SoG/60 rate last year, mistakenly using "Saves" instead of "Shots" in my calculation. Thus the difference between Belfour and the guys who sandwiched him was significantly more stark than I represented it.

So in accordance with your theory, Anon, Luongo's backups faced 2.8 shots per game fewer than BobbiLu did, but Belfour's faced 1.7 more and Vokoun's 1.4 more than the starter.

Anonymous said...

FLA Shots against:

2005-06: 2848
2006-07: 2395
2007-08: 2748

Belfour was largely responsible for a 400-500 swing in SA? (Keep in mind he 'only' played 66% of total team goalie min's)

FLA PP opportunities against:

2005-06: 514
2006-07: 443
2007-08: 344

FLA Winpct:

2005-06: .518
2006-07: .524
2007-08: .518

FLA SRS ranking (see hockey-reference.com):

2005-06: -.28
2006-07: -.27
2007-08: -.29

There are seasons where Belfour's backups faced fewer SA/60 than he did... & some where they faced more.

Did FLA's skater personnel change during those years? Do the skaters tend to fallback/rely on younger workhorse goalies lke Vokoun & Luongo, vs. tightening things up for a 40 yr old? Who coached the team during those yrs, what was the coaching philosophy?

Bruce said...

Who coached the team during those yrs, what was the coaching philosophy?

2005-06: Jacques Martin
2006-07: Jacques Martin
2007-08: Jacques Martin

Goalie Pete said...

Lehtonen primarily suffers from the fact that he's a fat, lazy, piece of shit. If you get paid a fat salary to be a professional athlete, there's no excuse for being out of shape.