Defense, Goalies and Perception

If you talk to Avs fans about their lousy season, you'll often hear how awful the defense was. It was obviously a sore spot with the team placing 26th overall in terms of GA.

The strange this is, none of the metrics aside from goals against suggest that Colorado was bad in their own end. The Avalanche actually allowed the 9th fewest shots against per game (29). And it's not like they were giving up break-aways or point blank chances: according to sites that track quality of shots against, Colorado was also in the top third of the league by these metrics: Desjardins has both Avs goalies with expected SV% of .915 (Budaj) and .916 (Raycroft). Nik Backstrom was the only regular starter with a higher expected SV% according to this process.

In addition, Hockey Numbers has COL SQA (shot quality against) at 0.97, good for 7th in the NHL.

The primary culprit, therefore, was the Avs crap-tastic goaltending. Colorado fans can thank the duo of Budaj and Raycroft for the Duchene pick this past entry draft.

It's funny how much blame the COL defense absorbs however. I'm certainly not calling Avs fans thick or unobservant, it's just the nature of perception: on almost every shot on goal, one could probably identify a mistake or missed opportunity to deny the chance. When pucks go in, the "mistakes" are noticed and tallied. If not, then not. For example: Jarome Iginla charges into the zone, one-on-one with Scott Hannan. Jarome makes a strong move left and fires the puck through Hannan's legs at the goal. Outcome one: Budaj makes the save, Hannan is lauded for not allowing Iginla a better chance. Outcome two: the puck sales through Budaj (again!) and Hannan is pilloried for not denying the shot AND providing an unwitting screen. The same play could be perceived differently based solely on the outcome. There's a reason bad goaltenders are coach killers (and good goaltenders make for genius coaches).

Something similar has been happening in Flames circles the last couple of years, although Calgary has been decidedly middling in terms of defensive performance in contrast to Colorado. That said, it'll be instructive to see what happens this year should the projected defensive improvement under Brent Sutter occur but without a concurrent rise in Kipper's SV%.

Evidence mounts

Last summer I wrote a little piece on the value of drafting goaltenders. I was less sure of my conclusions then, but I've become more confident of my position as time has passed. The support the analysis has received from a number of people I consider bright hasn't hurt either.

On that note, Richard Pollock published a piece at Puck Prospectus today that echoes my thoughts on the matter pretty closely:

Half the netminders who were considered their team’s starters in the NHL this season were drafted in rounds many consider to be a crapshoot. In fact, five goalies were drafted in the eighth and ninth rounds (rounds which do not even exist anymore). So, in effect, there are as many goalies starting in the NHL that were drafted in the first round, as there are goalies who (by today’s draft rules) wouldn’t have been drafted at all (a.k.a.—undrafted free agents).

Actually, three goalies who were not even drafted (Backstrom, Roloson and Hiller) had dramatic effects on their team’s play to close this past season. Backstrom was nominated for the Vezina Trophy, Dwayne Roloson played almost every game for the Oilers in the second half of the season and Jonas Hiller has continued his strong regular season play with tremendous playoff play.

...

Considering the cost, or lack thereof, of making such a selection, I wouldn’t be surprised to see teams look for diamonds in the rough more often than drafting a goaltender with a first round selection.

Next time your team is considering drafting a goaltender in the first round, maybe they should instead try to sign Swedish free agent netminder Jonas Gustavsson.

Sounds about right.

good teams, bad goalies, costly decisions?

Heading into the first round here, there's going to be a couple of pretty good teams relying on some pretty bad goaltending to get them through the post season. The three that leap to mind are Calgary (Kipper, .903 SV%), Washington (Theodore, .900 SV%) and Detroit (Osgood, .887 SV%). They were 32nd, 40th and 45th in terms of league ranking by SV% respectively during the regular season.

In the Flames case, they have no choice. In 6 appearances this year back-up Curtis McElhinney didn't fair any better than Kiprusoff. As much as I think the kid got the shaft this season and is potentially better than his results, there aren't any arrows to indicate he could outperform the incumbent.

That isn't really true in the other two cases however. Watching Theodore lose the game for the Capitals last night really brought this home for me - the guy probably isn't even playing at replacement level for that club. Here's his stats line via Hockey Numbers:

Theordore - 2.87 GAA, .900 SV%, .897 SQN% (shot quality neutral save percentage), 28.7 SA/60

Now, here are his back-ups numbers:

Brent Johnson - 2.81 GAA, .908 SV%, .902 SQN%, 30.7 SA/60
Simeon Varlamov - 2.37 GAA, .918 SV%, .917 SQN%, 29 SA/60
Mikael Neuvirth - 3.00 GAA, .892 SV%, .871 SQN%, 27.8 SA/60

Varlamov and Neuvirth are rookies (which were discussed briefly here). They only played 9 games combined, so there's sample issues here. Johnson's number are marginally better than Theodore's results and he played in 19 games this season before being felled by injury.

Still, the only guy Theodore outperformed this season was Neuvirth (whose numbers have fallen steadily since his draft season. I don't think he's the prospect the Caps hoped he'd be). Despite being the "more experienced" - and far more expensive goaltender - there's lots of reasons to think that Theodore is actually the wrong choice to go with here. However, given their investment in him (4.5M through next season) and the general animus towards rookies in the post-season, I'm guessing it's "Jose or bust!" for the Caps. It may prove to be very costly, since I think the Capitals could be major contenders for the cup with merely competent netminding.

The Osgood thing is even less defensible. He's been worse than Theodore on arguably a better team and Detroit has an older, experienced and proven alternative in Conklin. "He's done it before" is the rationale I keep hearing on choosing Osgood as the starter for the play-offs, but Im guessing that will go out the window pretty quickly if he starts (rather, continues) to suck. The same issue cropped up during Hasek's last stint in Motown and the club did the right thing (ran with the better goaltender at the time) in the end.

Course one of these guys might get hot, rendering all this moot. The hockey gods routinely give the touch one or two guys every post season and it's often impossible to predict who it'll be. Personally I hope it Kipper but I'm not going to hold my breath...

Flames expected goal differentials and Kipper

Thanks to a Jlikens post on the mind-bogglingly good fortune of the Boston Bruins so far, I've been working my way through Hockey Numbers recently (with a more Calgary-centric focus).

Some pretty interesting stuff. Most notably, the shot quality and expected goal differential work, which gives us an idea of how well a team is doing at generating and denying scoring chances.

The charts are big and scary, but terms are defined at the bottom. Basically what you're looking at here is a measure of the quality of shots for and against (shots from 12' go in far more often than shots from 30' for example) multiplied by gross shots for and against to yield an expected GD number.

The ES results for Calgary by this method are good. Very good. The Flames share the second best SQF (shot quality for) score in the league. Their SQA is just average, but they also tend to out shoot the bad guys, resulting in the second best EGD in the league (+14) behind the mighty San Jose Sharks. Keep in mind, this isn't actual ES GD (sometimes good quality shots don't go it and sometimes your goalie lets in a floater), but what this indicates is that only the Sharks significantly outchance the opposition more than the Flames so far*.

*(for interest sake, check out Detroit's numbers. They are third behind the Flames in terms of EGD, but they do it with pure out-shooting: they tend to give up high quality chances and generate low ones, but end up in the black because the shot differentials are so high).

Other items of interest:

- Calgary's ES SV% is a 25th ranked 88% - ahead of only Toronto, NYI, St.Louis, Dallas and Buffalo. How does Kipper have a SV% above .900 you ask then? Look no further than the 91.9% save rate on the PK. How ridiculous is that? Also, according to Java Geeks SV% numbers for shot quality, Kipper has struggled to stop average (89%) shots so far. MceLhinney's figure in this regard, for a point of reference, is 94%.

- Sticking with goaltending because it's a contentious point in Calgary these days, Kipper's contribution to GD this year has apparently been a negative one.

According to the linked graph, the Flames haven't exactly gotten the bounces either. The spike in expected GD seems to coincide with the Flames "turn around" after the Sharks debacle. That passes the sniff test, although I would like to see how the values for this output are generated.

- As I've wrestled with this Kipper @ 5.83M/year thing, I've started to get closer and closer to the conclusion that the Detroit model of spending money on skaters rather than goaltending makes the most sense. Low and behold, Java has a discussion on the very topic and he basically comes to the same conclusion:

It would appear good strategy would be to build a good team and then you can throw any goaltender in front without too much worry so long as the goaltender is at least average...

It would appear that Detroit is playing smart by spending very little on goaltending and spending on defense, Boston and Thomas are following similar logic. Of course this is all in the perspective on wins. What this says about the value of a goaltender is hard to say, due to the fact that a goaltender is worth more on a bad team and less on a good team. The best way to find out how good a goaltender appears to be: throw him on a bad team. Take Lalime, hugely successful in Ottawa (a team that doesn’t need a stellar goaltender), but failed miserably in St. Louis.

The study on which he based his discussion is a couple of years old now, but it looks more and more to me like the conclusions are sound. I've come around quite a bit on this topic since Kipper carried the team to almost-glory back in 03-04. I was a convert for a short time after that - a stone-cold "build the team from the net out" zealot I think. Time, experience and a growing body of data suggests otherwise to me now, however.

The re-structuring of my "goaltending effects" schema really began this summer when I looked at the value of drafting goaltenders and came to the conclusion that it was basically nil. I think the conclusion that "good skaters mostly the drive results" further enhances my thoughts on the issue and reveals why it's so damn hard to reliably find a great tender in the draft - the quality of your goalie is greatly determined by the quality of the surrounding team. Therefore, Gm's should concentrate on drafting and signing the best skaters they can find - then even an average tender can pile up the all-important wins (as we saw with Nabby last year and - yes - Kipper this year) and it won't cost an arm and a leg. Just make sure the guy can make it on time for practices and doesn't fight team mates while he's there...

PS - Im not trying to poke anyone in the eye with this post, as I know it will probably be unpopular to conclude that goaltending isn't all that important to success in general and that Kipper isn't all that good anymore in particular. I welcome discussion on the matter though since my own thoughts are still formative.

Kipper Save Percentage

This proved to be popular last year, so here it is again - Kipper's SV% in scatter-plot form:

click to enlarge

By my eye, there is no real upwards trend here, unlike last year. Rather, the trend is towards less volatility: prior to game 20, Kipper went from terrible to excellent and then back to terrible again. Things seem to have plateaued since, however, with his SV% hovering just above that .900 mark. If you're wondering, the 6-1 loss to SJS that seems to have sparked the team's turn-around was game #17...

So what are we seeing here? Given that the Flames made a marked effort to change strategy and focus after the Sharks shellacking, is the reduction in SV% variability entirely a reflection of the club's adjustment (that's my initial interpretation, but Im open to suggestions)? If so, what does that say about Kiprsuoff and team effects on goaltending as a whole?

Of course, raw SV% is only a small part of the picture. What I haven't included here is shot totals, shot quality and strength of opposition. Still, what the chart suggests to me is we're hovering right around Kipper's ceiling since the turn-around (his SV% has been a respectable .910 from game 21 on). The good news is that that isn't a terrible number. The bad news is it isn't a particularly great number either - a .910 SV% is good for about 20th in the league currently. Peers in that range include: Rinne, Johnson, Sanford, Giguere and Halak. Interestingly, Curtis McElhinney's SV% through his meager 4 games, 100 shots and zero wins has been - that's right - .910...

Tiny, miniscule sample size I know, but a suggestive coincidence nonetheless.