Why are goalie save percentages rising?
Eric T.
October 09 2012 09:30AM
Save percentages have been increasing sharply in recent years. Some of this is because there have been fewer penalties called, but even just at even strength we see a clear increase in save percentage.
How do we explain this?
Better defense?
One possibility is that teams are committing more on defense and pushing shots to slightly less dangerous locations. However, we can calculate average shot distances using Behind the Net's data, and while the trend is unclear, it certainly does not suggest that shots have been pushed farther out in recent years:
Of course, distance isn't the only factor that determines how dangerous a shot is. Angle can also be a key factor. If we saw more bad angle shots from a similar distance, we would expect to see an increase in the x-coordinate (distance from the center of the goal towards the boards), as shots go from being directly in front of the goal out towards the sides. However, that is not what has been observed:
The pattern in the last three years is a bit bizarre, with certain distances (1, 3, 5, 7, 10) much more likely than others (2, 4, 6, 9, 11). It is also interesting that this pattern was not observed in 2008-09; I wonder whether the NHL's official scorers went through some kind of standardization training after that season that led them to converge on this specific recording pattern.
But I'm getting sidetracked; the point is that the x-coordinate has scarcely changed over the years, and if the distance hasn't changed, then that tells us that the angle hasn't changed either. In other words, save percentages aren't going up because the shots are coming from less dangerous spots.
That would seem to suggest that either shooting skill is going down or goaltending skill is going up. I suppose either is possible, but with athlete strength and speed (and composite stick technology) continually improving, it seems unlikely that shooters are getting worse. So I'm going to proceed on the assumption that goaltending is actually getting better, and then the question is why -- are the individual goalies getting better (via improved training, refinements of technique, video analysis of shooters, etc) or is the pool of goaltenders getting deeper?
Can you teach an old goalie new tricks?
To answer this, I looked at the 24 goalies who have played at least 10 games every year since the last lockout to see whether their save percentages rose in parallel with the league average or whether the effect came from new goalies coming into the league who were bringing the average up. I also focused in on the 12 who were starting goalies over this period (never fewer than 25 games played in this sample, averaged over 40), in case the lesser goalies added more noise than value.
It doesn't appear that the rise in the league average came from individual goalies getting better; our sample that played for the whole period ended up almost exactly back where they started (.91949 in 2006, .91944 in 2012), and the trend is very similar for the reduced data set.
There is a potential confounding factor of age here, however. If a goalie was benefiting over time from adopting improving approaches, that might lead to steeper improvements as he entered his prime (his normal development plus new techniques) and less decline as he exited his prime (normal aging offset by improved techniques). So if the goalies in our sample were mostly in the back half of their careers, a real effect might be obscured -- seeing a plateau rather than a decline from those older goalies might actually be a sign of improving methods.
Luckily, the guys we are looking at happen to cover a reasonable spread of ages. There are some like Dwayne Roloson and Martin Brodeur who were on the decline, but there are also some like Cam Ward and Marc-Andre Fleury who are just now hitting their prime. As a result, while the impact of age and development might not be perfectly averaged out, it should at least be muted. The baseline expectation for this group of mixed ages should be something close to flat.
And indeed their aggregate performance was flat, so it's unlikely that the individual goalies were getting better -- collectively, this group ended up right where they started. So if the improvement in league-wide save percentage didn't come from shooters getting pushed to the outside, and if Miikka Kiprusoff and Tomas Vokoun and Roberto Luongo and Ryan Miller and Fleury and Kari Lehtonen and company weren't getting better, then what changed is that the talent pool got deeper. Save percentages went up because the rest of the league got better.
The goalies in this sample combined for 7592 starts over this span, just over 44% of the league's playing time. While their save percentage was exactly level over these years, the rest of the league was improving dramatically -- from an even-strength save percentage of .911 in 2005-06 to .922 in 2011-12.
Out with the bad, in with the good
In 2005-06, Martin Gerber, Alex Auld, Marc Denis, David Aebischer, Robert Esche, Antero Niittymaki, Mathieu Garon, and John Grahame were all starting goalies. Add in a couple of big-name goalies who were well past their expiration date like Ed Belfour and Curtis Joseph and it's fair to say that at least 1/3 of the league had a genuinely bad starting goalie.
In 2011-12, you might put Steve Mason, Garon and Jonas Gustavsson in that category. But the rest of that list has been replaced with the likes of Carey Price, Jaroslav Halak, Jonathan Quick, Jimmy Howard, Jonas Hiller, and Pekka Rinne.
The pool has grown much deeper; facing the same quality shots, there are a lot more goalies capable of holding their own than there were a few years ago. Having a top goaltender used to be a real advantage, but now it is expected -- instead, not having a strong goalie is a disadvantage.
The question for general managers is whether this should be expected to continue. If so, then locking up a goalie long-term means watching the rest of the field continue to improve while your goalie stays the same. The continuing improvement in the goalie pool would make it analogous to the electronics market -- your $1500 television might be really cool for a while, but in a few years you'll be jealous of your buddy who gets a nicer one for $800.
It may still make sense for a contender that needs a quality goalie to take advantage of a window of opportunity, but a team that is still building towards its peak needs to consider this trend before signing a goalie to a long-term deal.
It is a combination of goaltender technique and equipment evolution.
It isn't talent. If you taught the mediocre of 2000 the same techniques that 6 year olds are learning today you would have the same result.
When the change is gradual the change is hardly noticeable, but if you go watch film of goaltender technique in 10 year intervals you will see a massive difference from 1980 to 1990 to 2000 to 2010.
Stand-up to butterfly. Butterfly to Lock and block. Lock and block to Hybrid goaltenders who move as effortlessly along the ice as they do on their skates.
We are moving to a generation of goaltenders who lack an easily identifiable style for the fan/lack of goaltender technique educated individual. So if they look basically the same, the first variable we look to is stats.
The same stats that convince scouts to look for the flaws in Smith's game when he registers an .899 SV% and the same ones who ignore these same flaws and emphasize what he is good at when he is at .930.
It takes goaltenders years of repetition to learn techniques, one doesn't make a small alteration and change their results in one off-season.
GMs will always misidentify talent because there will always be bad GMs. If you understand what defines the elite goaltenders, then you shouldn't invest heavily in an area that is exploitable.
@Chris Boyle
I'm not sure what you're getting at in the comments about stats or bad GMs, but I'll reply to the rest of what you're saying here...
To me, your hypothesis seems like a subsidiary branch of mine, and I'm not sure whether you also see it that way or whether you think we are in disagreement over some point.
I'm arguing that existing goalies aren't getting any better, that the average save percentage is rising because the incoming goalies are better than the outgoing ones were. Your comment seems to be specifically attributing that to modern goaltending styles, and arguing that style is ingrained and so we don't see an individual goalie improving over time because their styles are long-since established and locked in. I can't disagree with that; from a stats perspective, that's indistinguishable from a situation where the incoming goalies are just superior athletic talents, and while your claims match with my anecdotal understanding, I'm not an expert enough scout to say whether the technique of marginal goalies today is better than it was three or seven years ago.
Either way, I think my point stands -- if you expect those improvements to continue, then a long-term contract to a goalie is a depreciating asset.
@Eric T.
Your hypothesis is correct and it confirms the eye test. The eyes can see the level of improvement in goaltender technique and equipment.
Your graph shows goalie save percentage increased for several years but has flattened in the most recent years. Projecting that trend suggests there will be little significant improvement in the future. Therefore, one could arrive at the opposite conclusion. In contrast to five years ago, perhaps now is a good time to lock in a good goalie to a long term contract.
@Paul J
Perhaps. The overall save percentage is definitely flattening, as the decline in power plays has slowed.
I'm less certain about the even strength save percentage though; I think you could reasonably draw either a straight line or a flattening curve through those data points, which is why I tried to frame things as an open question that GMs should consider.
Actually I'm totally against long-term contracts. I think its hard to find any long term deal (more than 6 years for a goalie, forward, whatever) that, five years in, either the player or the club don't regret. But I would like to see a Luongo trade completed; therefore I don't want his long-term contract perceived any more negatively than it already is. So I couldn't resist an alternative interpretation of the data.
But most important, the depth of your analysis is very interesting, as are Chris Boyle's comments.
Nice piece, Eric. A couple of other things which might be entering in - but would take time to dismiss:
New players coming in may, on average, be worse shooters. The increase in strength and size in the NHL - and the large premium placed by many GM's on size - might also mean we're seeing larger players enter, but ones who are slightly poorer shots than the corresponding small/fast/tricky guys we perhaps saw more of in the past.
We might also be seeing a greater increase in the speed and agility of defencemen, as opposed to forwards. I know Leaf management has discussed the fact that their old model - think: Komisarek and Schenn - is out-of-date, and needs a fast and mobile back line.
While average shot distance or position may not have changed much, higher-percentage shots may well have fallen. Any number of reasons could drive this: a clampdown on hacking at rebounds in close, better shot-blocking or fronting techniques in close, changes in defensive zone strategy. I have no idea if this is the case, but some might be possible.
That said, I'm convinced that the process of sifting out really bad goalies has accelerated, and hope that very soon it'll come to the Leafs. ;-)
@not norm ullman
Yeah, there are some significant assumptions here, and you've highlighted a few.
Still, I can't help but think that if there were some systematic change like less focus on shooting skill in selecting forwards or better defensive efforts to rush shooters' shots (resulting in inferior shots from the same locations), we'd see that as an across-the-board inflation of save percentages and not just impacting the goalies who are entering the league.
There's a distinct possibility that the difference between the long-term goalies and newly-entering goalies is just some combination of noise and age skew, but if that difference is real then I think it becomes overwhelmingly likely that the difference is about the depth of the goalie pool.
To Chris's point that he hasn't expanded on (equipment innovation)... a major, major change in equipment came with the Bauer Vapor sets in the early 2000s. Those pads basically blew the old Brian's, Heaton, and even Vaughn stuff out of the water. The significantly lighter materials that, more importantly, held their form when you dropped to the butterfly, made the butterfly far easier to learn (to the point that it's become part of the textbook goaltending approach). Both the pads and the gloves set standards that are practically universal. The development of lightweight sticks is similarly ubiquitous. And really, that's an extension of the innovations in the late 1980s spurred by Aeroflex.
...so...that, plus what you guys were saying.