Photo Credit: CANADIAN PRESS/DARRYL DYCK
The Vancouver Canucks are ninth in the National Hockey League in points, yet fifth in puck-possession. After a couple of years of tracking this sort of stuff, I don’t even feel it’s necessary to link all the math that shows a team’s Fenwick record is more predictive of the team’s future results than its actual win-loss record. A teams’ hockey ability isn’t best measured in how many games it wins, but in how many shots it’s able to direct towards the net versus its opponents.
What determines wins and losses after the shots have been accounted for? Voodoo, really. Sometimes they call voodoo “goaltending” and even though it’s the position played that’s the easiest to track, it’s almost impossible to predict how a goaltender will do in any given year.
There’s a rare breed of goaltender who is both elite and can consistently perform. Henrik Lundqvist is one of those goalies and Pekka Rinne is another. Both ‘tenders have managed to hide the puck-possession deficiencies of their teams for some time now.
Four times since the start of the Behind the Net era the Canucks have managed to be a top-10 team in goaltending. Boston, Florida, Montreal, Phoenix and San Jose have also managed to do this. As a result, the Canucks have generally been able to perform slightly better than their shot record would indicate.
That’s not really the case this season.
Here is the Canucks’ PDO records for each of the last five years. PDO is the simple addition of 5-on-5 shooting percentage and save percentage. It can generally expect to regress to 1.000 over the course of a long period of time:
| Shooting % | Save % | PDO | |
|---|---|---|---|
| 2008 | 8.1% | 0.922 | 1.003 |
| 2009 | 9.6% | 0.924 | 1.020 |
| 2010 | 9.8% | 0.919 | 1.017 |
| 2011 | 8.7% | 0.932 | 1.019 |
| 2012 | 8.6% | 0.929 | 1.015 |
| 2013 | 9.2% | 0.919 | 1.011 |
So how come the Canucks’ PDO has been higher than one for five straight years? Their goaltending, mostly. The Canucks were world-beaters shooting the puck in the 2008-2009 and 2009-2010 seasons but that’s fallen off to a number closer to league average in the last two. The goaltending has been consistently above league average by a fair amount with the exception of 2010.
Here are Luongo’s individual even-strength save percentages in those seasons:
| EV SV% | |
|---|---|
| 2008 | 0.929 |
| 2009 | 0.936 |
| 2010 | 0.925 |
| 2011 | 0.934 |
| 2012 | 0.929 |
| 2013 | 0.932 |
You can see Luongo is generally about eight points above the league average with the exception of the hiccup in 2010 where he only posted a .925. Otherwise, he’s been a top-10 goaltender in each season.
This year, Luongo is not the “starter” and while I hate to subscribe to the narrative that each team needs a No. 1 and a No. 2 goaltender, I think it’s clear that the Canucks’ better goaltender hasn’t been used as often as he ought to be. Schneider’s struggling this season, with a .920 EV SV%, about three points below the current league average, while Luongo has recorded seven quality starts in ten appearances and is right there at 8th among goaltenders with 10-or-more starts.
Unfortunately, the timeonice.com widgets we use to be able to track individual save percentage by game state isn’t up for this season. What I will direct you to is where the Canucks are struggling to win games, despite a very good shot count and shooting percentage:
| 2013 EV SV% | 2012 EV SV% | |
|---|---|---|
| All Situations | 0.919 | 0.929 |
| Score Close | 0.894 | 0.922 |
| Score Tied | 0.883 | 0.914 |
I’ve heard a lot about Vancouver’s defensive “fire drill” in terms of the coverage in their own end of late. The reality is that Vancouver’s goaltending has been letting them down (after Cory Schneider and Roberto Luongo both got off to an exceptionally hot start to this season), with the defenseman taking an undue proportion of the blame.
I don’t want to sound the alarm and suggest that after just 12 games we can write off Cory Schneider. He still looks like a future star to me. But knowing how important Luongo has been to the Canucks’ success, over the last few weeks I’ve been bit by the nagging theory that the Canucks are planning on trading the wrong goaltender. The blame lies mostly on the shoulders of whoever made the decision to start Schneider in Game Three against the LA Kings, which indadvertedly forced management’s hand.
You could get a better return for Schneider. He doesn’t have the long contract and he doesn’t have the no-trade clause. you’re someone who see the Canucks’ “championship window” shutting over the next two to three seasons, they should go for the trade that brings the best return and keeps in one of the most consistent goaltenders over the last five seasons. Of course, it’s pretty clear that the Canucks don’t see their championship window shutting so quickly, otherwise they’d have done more to keep guys like Ehrhoff and Salo, and probably flipped Hodgson a year ago for more immediate veteran help rather than a project…
Stats via Behind the Net’s Fenwick and Team pages, plus NHL.com
While luongo’s numbers are great during the season. In the playoffs, when it counts (assuming you want to win the Stanley Cup), his numbers and success are severely lacking. His past performances in the playoffs do not instill confidence that he can get the job done.
because the Canucks would’ve gotten 15 wins in 2010/11 playoffs with just about any goalie, right?
Whole team blew it IN Boston and in games 1,2 and 5 Luongo was rock solid!
I guess it’d depend on what you could get for Schneider, since he really seems like a very good goalie for years to come so the return would have to be very big. But if Gillis could manage that, I’d generally be on board with trading Schneider!
“because the Canucks would’ve gotten 15 wins in 2010/11 playoffs with just about any goalie, right?”
I didn’t say that, what I said was, his playoff numbers are nowhere near as good as his regular season numbers, he has had blowouts in the playoffs and those numbers do not lie, he is not as good a goalie when it comes to the playoffs than the regular season.
My point is simple, if you are going to put his regular season numbers up as a pro/con to keeping him or Schneider, we need to put up the entire Luongo story up.
Also, we need to remember, the past 4 or 5 seasons the North West have been about 20 gimme points per season for the Canucks, and that crappiness doesn’t get to play Calgary, Edmonton, Colorado or Minnesota in the playoffs, this year might be different of course.
you’re comparing a goalie with a track record to one with a little-to-none track record? the old adage; a bird in hand is worth more than two in a bush, is totally valid when comparing these two goalies. Luongo is a known consistent quanitity, schneider has not proven little as a starter. and i would blame the team in front of luongo more than the luongo for those blow out games
@VC
really? He came one game shy of winning the Stanley Cup. I say he’s fully capable.
While we did make it to the finals that year, let’s look at what actually happened based on the numbers versus Chicago. Luongo had a 3-0 series lead, Chicago came back after Luongo had a 7-2 loss and a 5-0 loss…not stellar numbers, remember the 7uongo jokes? Vancouver had to win the seventh game in overtime to avoid becoming the fourth team in NHL history to lose a series after taking a 3–0 series lead. We will give thanks to Chris Campoli for that gift. Let’s look at the finals vs Boston, people like to suggest that the 2 shutouts were awesome, if you watch those games, Tim Thomas was the standout not Roberto Luongo, the road games are the telling stat when it comes to Luongo’s numbers, 8-1, 4-0 and 5-2…again blowouts compared to the two 1-0 wins, I wont include the 4-0 final numbers, we had too many injured and we can’t hold it against Lu for the rest of the team not scoring for him.
Don’t get me wrong, I like Luongo, I just think we need to compare and show the playoff numbers in with the tally, I am just saying those numbers add up a lot differently compared to his regular season numbers.
Great read, thank you.
My biggest concern last summer when the Canucks were talking about trading Luongo was – will Schneider be able to handle the load? Thus far, my questions been answered and I’m glad they haven’t traded Luongo yet. People are making excuses for Schneiders below avg play this yr, this is professional sports. There are no excuses, it’s about results and he hasn’t delivered. Having said that, long term I think he’s the better option. They just need to keep both this season.
There doesn’t seem to be a market for (either) goalie. Maybe the lightning could use one, but they’ll give Lindback more time. Either of them would be an upgrade for a lot of teams, but not enough to extract the king’s ransom Gillis (and no doubt the fans) expect.
Vancouver looks like a team that is past it’s prime. Vancouver really was unlucky playing 1 of the most talented Stanley cup teams in a decade in 2010. Chicago was loaded. 2011 they had a good chance Kesler got injured and it was over Boston was too good. The window has shut it is time to trade Lou , and move on, they are still a very competitive team, they will need to make changes.
>a team’s Fenwick record is more predictive of the team’s future results than its actual win-loss record.
That’s looking less undeniable than it was in the past. Not past the end of the season, anyway:
http://hockeyanalysis.com/2013/02/27/the-declining-value-of-fenwickcorsi-with-increased-sample-size/
I think I read that post a little differently than you did.
David Johnson wrote:
“As you can see, with increased sample size, the fenwick stats abilitity to predict future fenwick stats diminishes …”
I don’t think he is saying that Fenwick doesn’t do a good job predicting wins but rather Fenwick doesn’t necessarily do a good job predicting future Fenwick stats. Teams with better Fenwick stats should still produce better win-loss records in the “immediate” future but for predicting further into the future other stats might be more reliable.
That’s just how I read it.
I intentionally didn’t really draw any specific conclusions as to what the declining predictive nature of corsi/fenwick with increased sample size means because honestly, I don’t know. It could mean that corsi/fenwick is somewhat systems driven and since some teams seem to change coaches every other year it affects corsi/fenwick over the longer term. It could mean that corsi/fenwick is largely driven by the depth of the team (where as elevated shooting percentage is largely driven by the elite players) and for many teams the second and third lines have a fair bit of turnover over the course of 2-3 years and if GMs aren’t evaluating second-tier players using corsi/fenwick metrics they may not be replacing the players with similarly talented (corsi/fenwick-wise) players (i.e. could be an area smart management could exploit). Or, it could be something completely different that I haven’t thought of.
Thank you for your response.
My position on this topic is definitely evolving and a lot of this has to do with the work you, and others in the advanced hockey analytics community, are doing in this area. I look forward to reading more on this from you in the future.
One question, for a lot of the reasons you mentioned (player turnover and coaching/systems changes, for example) how can you be sure that the additional data (seasons) should still be considered part of the same sample. With all the changes that occur from one season to the next, should two separate seasons be considered part of the same data sample?
Perhaps it’s in the nature of corsi/fenwick metrics to always be a smaller-sample short-term measure.
Why should two seasons be considered part of the same data sample? I’ll turn that around and ask you, if a team makes a trade half way through the season, why should two two half seasons be considered part of the same sample? What about an injury or a coaching change mid-season? At what point are two samples different enough to be considered two completely separate samples? It’s a question worthy of discussion but I personally think it is appropriate to take any size sample and make observations about it and see what those observations tell us. The fact that the percentages and goal rates improve predictability with more seasons of data tells me that it is reasonable to consider multiple seasons part of the same sample, at least for some purposes. And it makes sense too. Most teams keep the core of their team together for multiple years. How much has the core of the Red Wings, Sharks, Predators, Ducks, and Canucks, to name a few teams, changed over the past 5 years? Yes, players have come and gone but the core of those teams, the top end players, have largely remained the same. This is what led me to believe that maybe it is the top end players that drive the percentages (elite players can drive elevated shooting percentages and elite goalies can maintain elevated save percentages), but it is the depth of the team, the second and third liners, that drive corsi and second and third liners change far more frequently. It may also explain why corsi is so highly correlated with winning. Teams with good depth win, teams without depth don’t.
Oh, OK. I see what you’ve done. I was looking at it from a different viewpoint.
In addition to the ‘why do corsi/fenwick behave differently than percentages and goal rates in larger sample sizes’ question you present, I guess one could also ask if there an optimal sample size to maximize the reliability of corsi and fenwick?
It looks like data you used in your analysis was an aggregate for the entire league. I don’t know if you have stats for individual teams but there could be an interesting comparison between teams that are more stable and those that are more volatile.
Excellent article Cam.
Cory’s regression is not completely surprising
there have been only 3 elite goalies in last 10 yrs. Lui IS one of them Also @VC Lui’s ES Sve % is .931 in the playoffs so contrary to MSM & public opinion he DOES not have difficulty performing in playoffs.
Cory is most likely a good goalie like Quick Price ~.924/.925 ES But at the beginning of yr he still had `17% he was only ave. (A scary thought) Both Rask & Raycroft had ‘better’ numbers in same games then Cory? One looks great the other is out of the league?
Luongos has never even had a single year where he had a .931 sv % in the playoffs. Im not sure where you are getting that number from?
His average is .916 since he arrived with the Canucks and .904 over the last 4 years.
I think dan is talking about even strength sv%. I haven’t checked but are you using the same numbers?
I guess I prefer Schnieds but Gillis in his wisdom[?] gave Lou a contract that will prove to be untradeable. Burke sweet talked NHL head office into putting in the Loungo clause in the CBA so it will be tough to trade Lou for any real assets. Package Cory, Booth and what ever else and get a pure goal scorer and a big mean stay at home Dman that can fight and clear the front of the net. Not Florida please Gillis took enough of their castoff’s. Maybe while they are at it see if they can throw in AV and the rest of the coaching staff. Every game they get blown out in instead of calling a time out to settle things down AV lets it continue until it’s embarassing. Not only that but the teams PP and breakout are almost non existent. They don’t seem to have any set plays on the PP. Some of the Dmen[Garisson] seem not to know the positioning he should be and the D has both go in the corner leaving opposition in front of the net, Lindy Ruff is available>