SCH Mythbusters: Does a Low Shot Against Average Correspond to a Mediocre Save Percentage?

by Killion on Oct 21, 2009 2:20 PM CDT

Mythbusters_basic2_medium

So in the power rankings post, CliffKoroll had an interesting idea - Do teams with a low shots against per game average also have goalies with a so-so Save Percentage? The thought process behind it being... a goalie who is facing more shots beings to feel more comfortable and alert, keeping him on the top of his game, while a goalie facing a low number of shots might not stay as crisp. So I looked at the numbers to see what was what..

Now, these numbers aren't perfect and there are some problems inherent in just comparing Sv% and SA/G. For one, it's way too early in the season to make any final decisions (Huet-haters: please remember that). One good game, like a shutout, can drastically improve a goalies Sv% at this point as opposed to later in the year when putting up a 100% Sv% for a single game won't automatically put you up near the top of the league. Likewise, if a goalie has faced a low number of shots he may be playing teams that pick their spots wisely, meaning they are only taking shots in good situations while a high number of shots against could be played against teams that are just throwing the puck on net and seeing what happens, even if a goal is unlikely. There are plenty of other factors as well that I'm sure you'll all point out in the comments.

Anyways.. click the jump for all the numbers. This was a little rushed so if something doesn't make sense or you see anything else that stands out, again, let me know in the comments

So first up - lets take a look at the goalies so far who have faced the most shots. To save space I've just looked at the top 10 plus Niemi and Huet. I've ignored any goalie who has only played one game because.. well.. it just makes sense. Even still some random numbers get through but I'll point those out if I notice them.

Goalie	GP	SA	Rank	Sv%	Rank
Miikka Kiprusoff	8	265	1	.902	27
Craig Anderson	8	263	2	.935	6
Evgeni Nabokov	9	259	3	.911	17
Henrik Lundqvist	8	232	4	.927	8
Jonathan Quick	8	231	5	.900	28
Nikolai Khabibulin	7	227	6	.912	16
Cam Ward	7	216	7	.926	9
Marc-Andre Fleury	8	215	8	.926	10
Martin Brodeur	7	196	9	.903	24
Roberto Luongo	8	190	10	.879	40
Cristobal Huet	6	109	28	.844	45
Antti Niemi	3	57	38	.912	15

And.. just to look at it from another angle, here are the top 10 in Sv% plus the Blackhawks...

Goalie	GP	Sv%	Rank	SA	Rank
Mathieu Garon	2	.977	1	44	44
Ilya Bryzgalov	6	.953	2	148	23
Andrew Raycroft	2	.941	3	17	56
Ty Conklin	3	.940	4	67	36
Ryan Miller	5	.938	5	130	25
Craig Anderson	8	.935	6	263	2
Ondrej Pavelec	5	.929	7	169	14
Henrik Lundqvist	8	.927	8	232	4
Cam Ward	7	.926	9	216	7
Marc-Andre Fleury	8	.926	10	215	8
Antti Niemi	3	.912	15	57	38
Cristobal Huet	6	.844	45	109	28

Note: Andrew Raycroft is another anomaly. He hasn't played a full game yet, instead he's played 20 minutes in one and 33 in another. I included him anyway though... mostly because he's the replacement for tonight's game if the Hawks manage to make Luongo cry again.

So it appears there's a bit of truth to the idea. Of the goalies who have faced the most shots, 4 of them are also in the top 10 in save%. Granted, we're talking about Ward, Fleury, and Lundqvist here.. they're fairly elite goalies (Question mark on Fleury maybe). We're also including Anderson who is just having a ridiculous year so far. He's the main reason the Avs are doing so well and I don't believe they'll be able to keep it up all year. Regardless, that's only 4 of the top 10.. meaning 60% of the goalies facing the most shots are not in the top 10 in Sv%..

Of course - there's a problem just looking at total SA and calling it a day. If you're facing a lot of shots, it may mean your team isn't doing a lot to help you out which of course would cause a goalies Sv% to drop, which is why it's helpful to also take a look at the top Sv% in the league.

Here the idea hits a few extra bumps... Bryzgalov has arguably been one of the best goalies so far this year - his Sv% is at the top of the league for goalies getting regular time but Phoenix is 6th best in the league at making sure their goalie doesn't face many shots. Likewise, Buffalo is third best in the league for SA/G but Ryan Miller is 5th best in Sv%.

While the idea that a low number of shots against may correspond to a mediocre Sv% could have some truth to it.. there is just as much evidence, if not more, that shows they don't really relate to each other at all. It's probably just too soon in the season to really make any solid claims about this idea yet and it'll be worth checking it out again as the season goes on but for right now we're going to have to say:

Mythbusters_plausible_spray1_medium

via bajalatapa.files.wordpress.com

Tweet Comment 24 comments | 0 recs |

Do you like this story?

Comments

well that was fast.

by WirtzGMnomore on Oct 21, 2009 2:25 PM CDT reply actions

Hmm

It’s a good basic start to an experiment. It might have worked better if you used last season’s stats instead of this seasons, maybe we could see what that looks like.

There’s obviously variables involved in an experiment like this…if a team is only giving up 20 shots per game, but 10 of those are bad turnovers or breakaways (like, ahem, the Hawks) then there’s a good chance the goalies save percentage is going to be worse than a goalie on a team that gives up 40 shots from the point every game.

Also, it depends on the goalie. Some goalies thrive off of a lot of shots, whereas others just can’t handle them. Look at Anderson in Colorado….40 shots per night with Florida last season and he was rock solid. If teams consistently got 40 shots against Osgood he’d be letting in 6 goals per game.

I know when I’m playing goal, this idea does seem to have some merit. If I’m only facing 10 shots in a game, it’s hard to get into the flow. There’s big chunks of time where you never see the puck and you start to get cold and lose your focus. If you’re under constant pressure you have to stay on top of your game, you don’t have another choice.

The sweet spot for me is 20-30 shots. Even a little more, like 40 is fine. But if I’m facing over 40 shots, I’m going to start getting tired. Obviously NHL goalies are in better shape than me but I would imagine the same is more or less true for them.

by HawksFTW on Oct 21, 2009 2:49 PM CDT reply actions

true..

last season’s numbers would have been better – but I really wanted to see where Huet and Niemi both ranked in things as they stand now. Like I said though, this will be better take a look at as the season goes on.

I like the observations from someone who actually plays goalie though. The only time I actually played goalie was in a intramural league of floor hockey in college. We used plastic sticks and I work volleyball kneepads and had a first-baseman’s mitt for a glove… but my fundamentals were pretty strong

Second City Hockey

by Killion on Oct 21, 2009 2:57 PM CDT up reply actions

I like what you’re trying to do, but you really need to switch from “shots” to “shots against per game” or “shots against per 60 minutes” in my opinion. The results might be quite a bit different.

All things Thrashers + stats: www.birdwatchersanonymous.com

by The Falconer on Oct 21, 2009 2:56 PM CDT reply actions

Yeah..

I thought about that.. but I’m a little lazy. Might be something I go back to in the future.

Second City Hockey

by Killion on Oct 21, 2009 2:58 PM CDT up reply actions

This is the spot where you say "welcome to the internet, where I write for free." :)

All things Thrashers + stats: www.birdwatchersanonymous.com

by The Falconer on Oct 21, 2009 3:02 PM CDT up reply actions

"Facts are stupid things." - Ronald Reagan

In the tradition of many SCH posters, I prefer to rely on vague impressions and unsubstantiated conjecture.

Seriously, thanks Matt. I defer to WTF as an actual goalie, but the idea of a sweet spot sounds right.

Also, I third Falcolner’s observation that SA/game is a better measure, but if you really want to go nuts, the right metric would probably compare SA with SV% on a game-by-game basis. Otherwise, based on WTF’s observation, a guy who faced 20 one night and 40 the next would be penalized compared to someone who saw 30 each night.

Anyone out there looking for a topic for a mastesr thesis in statistics? Think NFL QB rating.

I’d do some in-depth analysis here, but I don’t want to step on any toes, and I too am a lazy douchebag. In fact, I’m exhausted from the mental strain of all this theorizing.

by cliffkoroll on Oct 21, 2009 3:18 PM CDT up reply actions

Correlation?

Since I’m currently in statistical hell for a class I would say the only real way to tell if SA/G and SV% are related is to run an actual comparison. Too bad I’m too lazy to help. Just being a douche and throwing it out there for someone with more time (and frankly some clue as to what they’re doing. I didn’t say I was doing well in the class).

by herecomethehawks77 on Oct 21, 2009 3:09 PM CDT reply actions

So I did it anyway...

Alright, I took the top 38 goalies by SV% and compared it to total number of saves. The result is there is a slightly positive correlation between # of saves and higher SV%, but just barely.

I’d try and post a scatter plot, but then I’d have to kick my own ass and deny myself flaming lemon wings tonight, neither of which is going to happen.

by herecomethehawks77 on Oct 21, 2009 3:28 PM CDT up reply actions

Even worse for SA

Taking the top 20 goalies for SV% and comparing to total SA is an even weaker correlation. Almost no relationship. Again, it’s early in the season.

Here I go to last year.

by herecomethehawks77 on Oct 21, 2009 3:40 PM CDT up reply actions

Wrap it up

To finish this dumb train of thought, there is an even weaker correlation between SA and SV% over a full season. All that means is that there is no statistical correlation between the number of shots faced by a goalie and the percentage he stops.

It just seems like it. For example when Luongo was in Florida he faced 30-40 shots every game and had a high SV%. It just seemed like the SV% was high because of the sheer number of shots, but apparently if he was a crap goalie he would have had a correspondingly bad SV%.

Osgood is an example on the other side. His SV% sucks every year and I used to think it’s because he only faced a few shots per game. No, he just sucked in the regular season last year.

Though I will say I think there is an intangible that some of you mentioned about goalies being “in the game” and active, as opposed to only having to face a few shots and having to come up with just a couple of spectacular saves.

by herecomethehawks77 on Oct 21, 2009 4:01 PM CDT up reply actions

I hate me

I just read my posts and am full of self-loathing. Please pile on.

by herecomethehawks77 on Oct 21, 2009 4:02 PM CDT up reply actions

Don't be hatin'.

Just fear your bad self :P – stats frighten me. My only stats background is in environmental science and I haven’t quite got that degree yet as I’m on the 20 year program. However, when I see how skewed numbers can be made and ponder the variables in mother nature and how these factors relate to a hockey player, I cringe. Don’t get me wrong, it’s interesting, but I always end up wondering what the stat is for GAA of blue-eyed goalies that wear between size 9 – 11 skates hailing from east of the fourth meridian prior to November of a given hockey season (for example). Anyhow, I’m glad there’s people like you out there crunching. Thanks.

"Call Detroit, tell dem... BULLSHIT!"

by Hungryhawk on Oct 21, 2009 5:04 PM CDT up reply actions

So what you're saying is

That you think of stats they same way baseball coaches and advisors do?

wait what?

by pepe126 on Oct 21, 2009 5:09 PM CDT up reply actions

I loved the stats classes I took in college

but it was only 3, and I’m about 7 years removed from them now.

But I like what you added here, and of course Matt for taking the time to get this started.

One of the things that I’d like to see, and more specific to the Burn In Hell Huet!!!! debate, is comparing not the SA/SvPctg correlation, but the percentage of Shots to Prime Scoring Chances.

The part of Cliff’s hypothesis that made the most sense was not just that Huet was facing few shots, but of the few shots he did face, a high percentage of them were difficult saves. If we could take the Prime Scoring Chance % and see the correlation to Sv Pctg, then maybe we’d see something interesting. Unfortunately, even though I see the prime scoring chances stat on tv broadcasts alot, it doesn’t seem to be widely available on the internets. So I guess we’ll be stuck in this neverending circle of hell, where everyone has an opinion, but no one has anything to back it up.

Got more soul than a sock with a hole.

by chrome on Oct 21, 2009 4:12 PM CDT up reply actions

And again like cliff said previously, but I didn't make clear

it would have to be calculated on a game by game basis. So, if I had access to the stats, I would have it set up like this for each game:

Shots Against | Prime Scr Chnce | Prime/Shot % | Sv %
Game 1
Game 2
Game 3
Game n

My guess is, though, that over the course of the season, this would all even out. Huet, or any goalie, will eventually have a stretch where he faces few shots but a high percentage of prime chances and keep a high sv pctg. Other stretches, like now, he won’t. Over the course of the season, a goalie’s number will reflect whether he played well or not.

Which brings us back to the main argument now. It’s early. Stop booing. Relax.

Got more soul than a sock with a hole.

by chrome on Oct 21, 2009 4:22 PM CDT up reply actions

One thing to remember, is a lot of the time, what is considered a “prime scoring chance” doesn’t even end up being a shot on goal. A team has a 2 on 1, someone shoots it and hits the post. That’s considered a scoring chance but not a shot on goal.

What a nightmare. You’d have to actually watch every single game to get this 100% correct.

by HawksFTW on Oct 21, 2009 4:58 PM CDT up reply actions

Yeah.

I thought of that, but, since that would definitely be impossible to track, I figured it would still provide the next best thing.

Got more soul than a sock with a hole.

by chrome on Oct 21, 2009 5:05 PM CDT up reply actions

What's your P?

Just interested if you’ve got that.

If I can't be a good example, I'll just be a horrible warning

by Trixietrx on Oct 21, 2009 8:06 PM CDT up reply actions

Ken Dryden

talks about two types of good goalies, those on good teams and those on bad teams.

A good “bad team” goalie, numbed by the volume of goals he cannot prevent, can focus on brilliant saves and brilliant games, the only things that can make a difference to a poor team. A good “good team” goalie cannot. Allowing few enough goals that he feels every one, he is driven by something else—the penetrating hatred of letting in a goal.

There’s more in The Game about it, but this was all I could find on the internet from it.

twitter.com/kaner88

by Original Six on Oct 21, 2009 3:49 PM CDT reply actions

this is my hastily drawn-up version

Make of it what you will.

The first set of columns compares the best SV% stats to the highest SA/G for goalies who’ve played in a certain minimum number of games (30 for 2007-08 and 2008-09, 3 for 2009-10). The second set compares the worst SV% to the lowest SA/G. Seems to me there’s a slightly bigger correlation between low SA/G and SV%… look at how many shots the goalies with low SV% have to face in general.

Also too lazy to explain more, and I’d hate to be one of those people who’d be so biased they’d try explaining everything away just to support their position, but hey. Food for thought.

Now I’ve got to go— there’s a game tonight, in case you haven’t heard. ;)

by chiblackhawks on Oct 21, 2009 4:09 PM CDT reply actions

PS. Huet is on both columns for comparison's sake.

He didn’t crack the top ten for SA/G or SV% for any other time except in SV% for 07-08.

But you knew that.

by chiblackhawks on Oct 21, 2009 4:10 PM CDT up reply actions

well I'm impressed...

…considering I have yet to master even spel chek.

So, I can’t post my analysis, but I’ll share my findings based on this data.

For the last two years, the coefficient of correlation for SA/G and SV% are 0.57 and 0.55. For this year, it’s only 0.20 so far, but hey, sample size. I’m no statistician, but it seems like there is something here.

I created a rudimentary formula for “Adj SV%” = SV% – SA/G/1000. OK, 1000 is arbitrary (hell, the whole formula is), but, in general, a smaller number gives more credit to goalies who face fewer shots, while a larger number gives greater weight to SV%..

A couple things pop out from this calculation. In 2007-08, Osgood was 9th in SV% but 4th in Adj SV%, while Lehtonen was 8th in SV%, but only 14th in Adj. SV%. (In passing, I also note that Huet was 3rd in SV% and 2nd in Adj SV%. Wouldn’t he look good in a Hawks sweater?)

Last year, Auld (grr) was 9th in SV% but 6th in Adj SV%, while McDonald was 16th in SV% but 21st in Adj SV%.

This year, Halak jumps from 11th to 8th and Khabi drops from 8th to 13th, which I find satisfying. Also, Huet has been urine any way you slice it so far this year (he only moves up one notch, from 24th to 23rd, even with a factor of 500, edging out the Lightning’s Smith.)

Of course, this amounts to so much pud pulling, and, as chrome points out, doesn’t take into accout prime scoring chances and its associated statistical can of worms.

Interesting anomalies aside, though, my conclusion is that there is something to be said here, but not too much. SV% without adjustment is a reasonable apporximation.

Finally, I repeat that I am impressed with the resources deployed on the subject. Together, ladies and gents, we can conquer the world.

Now if you’ll excuse me, I’ve only got two hours to get outside several brew-hahas before the puck drops.

by cliffkoroll on Oct 21, 2009 5:36 PM CDT up reply actions