Friday, May 08, 2009

Carey Price: Thoroughbred In The Gates?

Are These Thoroughbred Trends?

As promised, my second post on Carey Price will deal with the question as to whether he is a thoroughbred or not. This piece is not a debate about Halak or Price (that may come later), just the thoroughbred issue.


Carey Price 2008-2009

This season is touted as a season to forget for Carey Price, but I don't think that on the whole is being very fair. I only have to look at my own writings for reminders, like this from the mid-term analysis:

Carey Price: A+
Top ten in multiple categories in NHL rankings, his contribution to the Habs has been undeniable so far. His value to the fan, as we've seen during Halak's reign, is that he positions himself so well and so far ahead of the play that many saves look too easy. Only a year into his career, to be A+ on a 56-point team midway through the season is an impressive feat. Tough tests to come though, no doubt.


But it wasn't 20 minutes of play after that post before Carey began an atrocious stretch of the season that lasted an unacceptable 14 starts and 6 weeks in all. The beginning of the downfall may have been to injury or, if you buy my pet theory, may have been the How-to-score-on-Price seminar, aka the Youngstars game, in January.

His statistics over that 14-start stretch include a 3.91 GAA, an Aebischeresque 0.862 save percentage and a paltry 5 points from a possible 28 for his team. Conversely, the other 38 games look quite acceptable indeed with a 2.46 GAA, a 0.919 save %, with a contender-like 21 wins and 9 OTLs.

Consider also that in the 14-games of putridity he had a save % under 0.900 on 8 occasions, whereas in his first 25 games that only happened on 6 occasions.

As though letting in nearly 4 goals a game weren't bad enough, it was the style in which he did it in the confidence-sapping stretch of the season that killed the team's morale almost completely by early March. It was goals within the first ten shots ten times and within the first five shots on five occasions. It was goals on consecutive opposition shots (3 times). And it was the untimely way he faltered, such as after the team got a lead or clawed back to tie it up.

All of these things happen to Marin Brodeur once in a while (see playoffs 2009) and to the next tier down a little more. And they happen to average goalies for long stretches.

But isn't that the point? Is he or is he not above average? (A thoroughbred does not conjure thoughts of running in the pack) Perhaps more pertinent, do great goalies start as average/below-average goalies (do thoroughbreds)? To get to the bottom of it, I had a look at some nice stats.


Quality shot stats

Anyone who has been watching Carey Price over the last year (starting specifically at the Philadelphia series) must surely have noticed that good shooters with a good amount of time seem to be able to beat him more than we'd like them to. Anyone who gets sight of the top of the net seems to score. Of course, the defence is heavily to blame for letting so many good shooters get so many good shots away, but there comes a point when you have to wonder -- is Umberger a good shot. If he is, who isn't?

From a subjective point of view, once you get this idea in your mind you start seeing what you want to see. That can be a problem if a goalie lets in high shots game after game. Witness Kostadis' point in the comments to the introductory portion of this series. He notes that Carey Price makes plenty of outstanding saves in some games where we end up un-doming him. Usually because we see the eventual gamewinner as a misplay. So a more objective approach is required from someone so tainted as me, stats...


Quality of shot analysis

For this part of the analysis, I sought some evidence of shot quality from statistics to probe the question properly. The first ones I found were the Behind The Net analysis based on shot quality at 5-on-5.

Here Carey proves himself to be better than average (first pleasant surprise). Based on the average ability to save shots form the positions they came from, we should have expected Carey to let in 93 goals (92.7 actuall) on 1027 shots at even strength. In play, he only let in 88 even strength goals - saving 5 goals more than he should have - good news.

I took it a step further to do some save percentage on these numbers. Basically, I calculated how many goals each goalie saved over and above the expected and divided by the number of goals expected. What this gives us is a statistic that I have called clutch saves; it gives some guide as to what's been happening with "clutch" goaltending this season.

Carey's 4.7 goals saved over 92.7 expected goals means he saves 5.1% of shots that he would have no business saving (if he were average). Very good, I thought.

Furthermore, this percentage puts Price about 32nd among goalies who play enough to warrant consideration. If you look at NHL starters, he's 18th. And if you chuck in Varlamov for good measure, Price is 19th.

The top 5 in the league are all still alive in the playoffs:

Tim Thomas (29.2%)
Simeon Varlamov (26.6%)
Roberto Luongo (26.6%)
Nikolai Khabibulin (26.4%)
Jonas Hiller (22.6%)

Others worth noting:

Henrik Lundqvist (20.8%)
Tomas Vokoun (20.7%)
Martin Brodeur (20.5%)
Pekka Rinne (15.5%)
Steve Mason (13.7%)
Jaroslav Halak (13.4%)
Cristobal Huet (13.2%)
Yann Danis (12.3%)
Cam Ward (12.2%)
Marc-Andre Fluery (3.2%)
Chris Osgood (-6.7%)
Mikka Kiprusoff (-10%)

What does this mean?

Well, it means he's average - or at least he was for the enire season.

The problem here is that averages include all goalies who play in the NHL, not just the 30 starters. So when Pogge (worst in the league, btw) steps in an lets up 9 more goals than expected in 7 games, Carey is compared to that. What's more, for me, where he currently sits is not thoroughbred territory - too many thoroughbreds are ahead of him (some almost as young).

His save percentage, though, at even strength is still a plump 0.914, so no decisive knock-down here. Based on this analysis, his toroughbred status hangs on.


Penalty-kill save %

Behind the Net also has a whole raft of other stats if you're that way inclined. For the second look, I decided to show you goalie performance while his team is a man down. The reason I choose this is because the shot quality analysis isn't done for anything but ES, also it seems logical that shots are better set up on the PP than at ES.

These stats offer a bit of a quick and rude awakening for all those people who support Carey's thorough breeding. If you look across the season for the whole league, Carey Price is in fact the worst starter across all teams in terms of save percentage on the PK. On an average of 38.6 shots per 60 minutes of PP time, Carey Price allowed an average of 7.61 goals. That means his save % was a rather embarrassing 0.803 at 4-on-5.

Now, PP shots are harder to save you say. Of course they are. But the average NHL goalie has a save % of 0.852, and that includes every weak link there is (like Mathieu Garon who saves just over 75% of PP shots in 20 games). Some good goalies have low save percentages here (Brodeur, for example at 0.816), but many don't. Thoroughbreds shouln't, I say.

The top 5 starters in the league here are:

Nicklas Backstrom (0.923)
Henrik Lundqvist (0.918)
Martin Biron (0.894)
Simeon Varlamov (0.893)
Tim Thomas (0.888)

Others to note are:

Craig Anderson (0.910)
Jaroslav Halak (0.890)
Mikka Kiprusoff (0.888)
Pekka Rinne (0.887)
Steve Mason (0.870)
Marc-Andre Fleury (0.867)
Cristobal Huet (0.866)
Jonas Hiller (0.862)
Cam Ward (0.862)
Jose Theodore (0.845)
Yann Danis (0.833)
Roberto Luongo (0.833)


What does this mean?

Well not much on its own. If Carey had been Halak-efficient it would have meant a mere 2 goals less. It's not a case on its own, but as things add up, it certainly seems to follow the pattern. It's another hit against the thoroughbred hypothesis.


Giving a chance to win

In response to Kostadis' case that LIW is wrong about Halak > Price (which is not the point of this piece, rather whether Price = thoroughbred), I thought I could take a similar approach.

He claims, that Carey Price gave the Canadiens a chance to win in 33 games of 49 (so 67.3%), whereas Halak only gave them 52.9% chance. I'll start by pointing out neither goalies stats are based on starts, which I thought they should be. Price goes up and so does Halak, with this tweak. Halak, for his part, gave opportunity for wins in 19/33, or 57.6%.

I thought I would look at things a bit closer. For can we really say that a goalie who loses in OT always gives his team a chance to win. In reality, you say, we can. And I concede on that. But for us at Lions in Winter, we generally look at how many goals a goalie allows to decide whether that would allow a win on most nights (i.e., those where his team doesn't win 6-5). While each goalie loses a couple of games this way, this subjective method actually works in the goalie's favour, since games that are lost 1-0, 2-0 or 2-1 can be counted as games he gave his team a chance to win.

By this criteria, nothing really changes. But if you get generous and say that 3 GAs is a reasonable amount for a goalie in a league where average GAA is parked in between 2 and 3 year-upon-year, then it gets interesting.

Carey Price, by the 3-goal criteria, actually gave his team a chance to win 73% of the time. And, lo and behold, Jaroslav Halak gave his team a chance to win an incredible 73% of the time. Funny that it would be equal.

By this criteria, Carey has to win. But only against Halak. It doesn't really prove that anyone is or isn't a thoroughbred. After all, no one has ever accused 9th round QMJHL alum Halak of being thoroughly bred. For that we need to look at goalies on other teams - ones that actually win.

Quickly for illustration I looked at Tim Thomas, Roberto Luongo, Cam Ward and Pekka Rinne.

Tim Thomas by the first criteria (W and OTLs) gave his team a chance to win 79.6% of the time. If you guve him the 3 goal benefit of the doubt that's 83.3% of the time.

Roberto Luongo gave 74.1% and 74.1%.

Cam Ward gave 64.7% and 76.5%.

Rinne gave 67.3% and 81.6%.


This is not a random sample, but I think it's a fair group to sample if you want to look at thoroughbreds. While Carey Price and Halak look OK in isolation, in the light of day they don't stack up to Thomas or Rinne. They come close to Luongo, but to be fair to Roberto, he did have 9 shutouts which mean automatic points (not just chance at), both our guys mustered only 1 of those. I think this category shows Carey in a much better light than the other two. What's more, given his age, I think you could argue that giving his team a 73% chance to win is pretty darn goosd. Is it thoroughbred stuff? Let me just say, I'd rather have Rinne based on this season... (is he a thoroughbred?)


One streak to skew all stats

There's a very very solid case to be made that it was all due to one bad streak. But was it just a bad streak? Did 14 games destroy everything? It's a valid argument, I concede.

For me it's about trends, though. If this were the very first slump from Carey Price, then I'm sure it would be water off a ducks back. It's not though, and a responsible scientist like myself has to ask whether the good is the aberration, or the bad. Those of us counting will note that Carey Price has had a few breakdowns since his arrival with the team. To be generous (and a bit facetious) I'll call these non-thoroughbred episodes, or NTEs. The more research-minded among us will note that he also had a pretty long NTE the season following his selection in the first round, not to mention his blow-up in the WHL playoffs that year. So to enumerate the NTEs:

1) 2005-06 Tri-City Americans
2) January 2008 Montreal and Hamilton
3) April 2008 Round 2 Montreal vs. Philadelphia
4) January and February 2009 Montreal (the aforementioned 14 games)
5) April 2009 Round 1 Montreal vs. Boston

That of course allows us to paint every stretch in between as one where the opposite position is held (i.e., a thoroughbred stretch):

1) 2006-07 All teams (Tri-City, Canada, Hamilton)
2) October to December 2007 Montreal
3) February to April 2008 Montreal (including Round 1 vs. Boston)
4) October to December 2008 Montreal
5) March 2009

On balance there are as many good games as bad in there, and most games are probably just adequate or average (less than 3 goals, saving 90% of shots). What we have is a pint glass with half a pint of beer and half a pint of air. If you go for the glass half full approach, you see the Gold medal for Canada the Calder Cup, the end of the 2007-08 season and Round 1 and the very sturdy beginning for this last season. It all adds up for you. The glass half empty perspective notices the floundering in the playoffs, the 14-game stretch to forget and the easy goals, and starts to question whether sub-NHL success will ever be translated.



I've come to the end of this and realise that despite a lot of stats, not much has changed. I haven't proved that Carey Price is not a thoroughbred, nor can we look and say that he is one. I think I would conclude that there is plenty of reason to question whether his progression will be as smoothe as some assure us it will be. I think there's enough evidence to suggest that anyone who says he's not a thoroughbred is at least as sane as someone who purports he is. I think that's all I wanted to do - prove our point of view on this isn't as crazy as some make it out to be.

What I can say with some assurance is that this season he was not a great goalie. On average he was an average goalie. Is that good for 21? I don't know.


I think we'll have to wait and see what happens by next season on this one. Maybe even wait until the next playoff test. Given that Gainey is basing our team around this guy come hell or high water, I'd recommend to anyone that hasn't already dedicated 50 blog pages to this to take the glass half full approach -- it'll be better for your sanity.

No comments: