Why NHL save percentages continue to decline despite goalies getting better

One thing we can be sure of when evaluating the evolution of NHL hockey is: nobody’s getting less talented. Fourth-line “grinders” have hands that rival your average skill player from the '80s, speed keeps ramping up each year, and goalies keep getting savvier and more athletic.

The goalie one is strange, though, because if you solely use what has been used as a major source of evaluation for a long time in our game — save percentage — you might think they’re going the other way.

Going back to the previous 10 seasons, starting in 2015-16, here is the trend in NHL average save percentage:
.915
.913
.912
.910
.910
.908
.907
.904
.903
.900

It just felt like at some point, it would hit bottom and level out. These guys are too good for it to go down in perpetuity, and then last year: .896.

The last time the league average save percentage was below .900 was over 30 years ago in 1996, and it was two seasons prior to that when it was below last season’s .896.

This season, through over 150 league-wide hockey games, that save percentage is .892.

So, what is going on here? How low can it go?

There’s rarely one tidy answer to league-wide trends, and this mystery is no different. It seems to be a convergence of factors, and so I thought it’d be good to just talk it out and compile the theories in one place.

Shooters got better, while goalies got more predictable

This would be the obvious starting point. At some point the goalie community abandoned the plan of “just do whatever, get something in the way of the puck” (“kick save and a beauty!”), and developed some real in-net structure, an actual plan. That plan? Be big and block the middle of the net, and basically say to shooters: If you can make the perfect shot to a top corner, you can have it.

For years, shooters weren’t equipped for this kind of precision. By taking away the freebies, they immediately became harder to solve as big-bodied blockers, rather than savers. They gave away fewer gift goals, and goalies made up ground on the shooters. Save percentage went up.

This “structure and plan” that goalies nearly universally adopted meant they became fancy shooter tutors — you knew where they were going to be and when, to the point that you barely had to look.

At the same time, the league worked to limit the size of their equipment.

As one-piece sticks came along in the early 2000s, people my age tried to learn and adjust to them on the fly. I went from wood sticks (and rebar-like aluminum ones) to two-pieces to one-pieces all during my transition from midget to junior to college. But this generation was born with these things in their hands, they went whippier and whippier until they barely needed to lean on the thing to make the puck go, and they got accurate with these tools.

When goalies said “if you can hit it, you can have it,” well, they started to hit it. Shooters knew they’d be in the same blocking spots, and they barely had to look up to know what they had to do.

The way shots are counted has changed

For a long time, shots were counted by A Guy who sat way off the ice and said, “That looks like a shot,” and recorded it. Different buildings gave out way more shots than others, simply because they had a more trigger-happy shot counter. Some guys rewarded the home team with credited shots more than the visiting teams, to the point where teams knew certain buildings had skewed numbers.

It can be shockingly hard to determine what’s a shot and what isn’t. Saves on pucks going clearly wide aren’t supposed to count. Posts and crossbars don’t either, but did that tick off the goalie first? Tough to say, let’s count it as a shot. A shot goes into a pile with the goalie, a D-man, and a forward, did it get through to the net? Maybe!

But with the introduction of accepted sports betting, you just can’t get these things wrong. People bet real money on the shot over/unders; they bet saves, they bet individual player shot totals. Now, more than ever, the off-ice officials peruse video to make sure each shot was legitimate, lest some bet be paid out where shots weren’t taken, or the opposite. The league has a vested interest in getting it right.

Post-game, the shots are further vetted still, with analysis done with a more fine-toothed comb. A handful of shots that would’ve counted just 10, even five years ago, will get stricken now as they were shown to be blocked, or went clean into the post, or were clearly going wide.

Shots come off the board, a concept that would’ve been foreign not that long ago.

Coaches finally value possession, which kills low-danger shot attempts

I’ve talked about this with Steve Valiquette, our resident goalie expert on Real Kyper and Bourne, and he’s explained through his company, Clear Sight Analytics, that there’s been a clearly defined death of the low-danger shot attempt.

I grew up in the era of low-danger’s "no such thing as a bad shot,” and when I tell you I always hated that concept, I mean I truly have always hated it. I was pretty well-liked by almost all my coaches, but it’s the one thing I’d consistently get in little battles over. If you had the puck coming down the wing out by the boards and were getting closed out, they wanted you to put that thing on net. “No such thing as a bad shot” was gospel, and it sucked.

The only problem is that it’s really freakin’ hard to get the puck, and harder still to get it in the O-zone, and shooting it from 60 feet was simply handing it back to the other team. They just about never went in, and so you’re better to curl up and delay, or to put a little chip behind the D’s heels and make them turn, and to at least get yourself in a 50/50 for the puck that’s located somewhere you chose to put it.

And if you ask Valiquette, he’ll tell you that those low-danger shots were essential to a goalie finding his rhythm, like an in-game warm-up that would keep them sharp, let them feel it, and stay feeling good about themselves (particularly early in games).

These low-danger shots going away have pushed down shot totals. Here’s the average shots per game goalies have faced in the NHL over the past five full seasons:
31.4
31.1
30.1
28.1
27.6

This year, that number: 27.1.

That’s the lowest average shot total that I can find in a season in recorded NHL history: 1998 was the closest, at 27.3.

Analysts figured out that the teams that spent more time in the O-zone tended to win more (provided you account for some score effects). There’s just been a closer evaluation of “possession,” and how just having it in the offensive zone has value. Sure, maybe you weren’t scoring, but the other team sure as heck wasn’t either.
Goalies got so reliable, you didn’t want to just shoot from long range anymore, or it was essentially a turnover. So, hang on for a better look. Don’t fire until you see the whites of their eyes.

Think about NHL overtime. That hanging on to it is what teams want to do now; it’s just very hard when there are 10 men on the ice. Ideally, even just having it gives your team the advantage — as mentioned, they don’t have to defend, the D gets tired, and if there’s a breakdown, they’re in.

So now, you wait for that breakdown, you try to find your way in, and to go with this concept: you want an attempt that comes off a pass. Shooting percentage skyrockets when a shot comes off a pass, particularly when that pass crosses the middle of the ice. This elevated strategy is all brutal on a goalie.

More screens, for two reasons

Now if a team is going to fire one from distance, firing it with nobody in front is usually a waste too. Goalies have middle infield gold glove-level catching abilities; they read plays exceptionally well, and so if they see it from any real distance, they have a skyscraper-high save percentage on those shots.

So, we get bodies in front, and that’s a more conscious decision than ever, knowing how you have to take away the eyes of today’s goaltender.

But the other piece of that is that the game has evolved to value more skilled, skatey, offensively-capable D-men. Your average third pair D used to be some discount version of the ever-effective Derian Hatcher, and now a bunch of them are some Temu version of Scott Niedermeyer. They’re lighter, and if they’re not, they at least have more of a penchant to get up in the play and to want to participate in moving the puck.

I’m generalizing obviously, but today’s D-man is more inclined to that side of the game, and less inclined to straight-up boxing out and defending with snarl. You also can’t just chop guys down with lower-lumbar crosschecks like you could for some time (see: Hatcher), so it’s a less physically intimidating place to go spend time. A strong forward can live there with minimal resistance.

As a result, goalies see fewer pucks, or see them later, and that leads to more rebounds and chaos action in the crease. Those are high-percentage shots being taken too, a rebound found just in front of the crease.

In all, it’s not too hard to see what’s become of goalies’ save percentages. The position is harder to play than ever, as now teams don’t shoot from distance; they hang on to the pucks, and they want a premium scoring opportunity, often with traffic. And when they pull the trigger, guys can really convert.

There are new stats that have emerged to help us better judge goaltending, such as “goals saved above expected.” And yes, there are varied definitions for how they compile these numbers, but in theory, they should account for the environment a goalie is playing in (are they behind a team that gives up a lot of breakaways, shots off passes, and other high danger chances?) and will help you get closer to better understanding how a goalie is really performing.

So as you head into some discussion about your team’s goalie, don’t be too alarmed when you see an .895 save percentage. It may be over 30 years since you’ve seen numbers like that, but as of today, .895 is actually above league average, and you might just be looking at a pretty good keeper.

Today’s mantra of “Don’t fire until you see the whites of their eyes” has left an awful lot of those peepers bloodshot.