Tuesday, July 28, 2009

Situational Averages

To further illustrate the importance of special team factors, here are the league average save percentages by game situation for the regular season and playoffs for the last 10 seasons:


YearREG SV%REG EVREG PKREG PPPO SV%PO EVPO PKPO PP
1998-99.907.916.872.918.915.924.872.954
1999-00.905.912.866.917.918.929.872.920
2000-01.903.914.862.912.913.925.857.925
2001-02.908.916.872.919.918.929.873.929
2002-03.909.918.869.922.918.929.870.900
2003-04.911.922.867.913.922.932.877.925
2005-06.901.915.861.913.906.921.865.901
2006-07.906.917.865.912.919.929.881.940
2007-08.909.920.868.916.921.932.885.913
2008-09.908.919.868.905.915.924.873.948
Average.907.917.867.915.916.927.874.923

The difference in save percentage between facing a shot at even strength and a shot on the penalty kill has consistently been about .050. There is also very little difference between expected save percentages at even strength and when a goalie's team is on the power play.

The save percentages give evidence of a more open game after the league resumed in 2005-06, but the overall quality of goaltending in the league today is pushing the EV SV% right back up to where it was in 2003-04.

12 comments:

Bruce said...

Thanks, CG.

The thing that puzzles me about the shot quality debate is the difference between regular and playoff stats.

**** SV% **** EV **** PK **** PP
REG .907 *** .917 ** .867 ** .915
P/O .916 *** .927 ** .874 ** .923
--------------------------------------
** +.009 ** +.010 * +.007 * +.008

That's 10 seasons out of 10 that Sv% was better in the playoffs (by a very consistent +.009 +/- .004), 10 seasons out of 10 that EV Sv% was better, 8.5 out of 10 that PK Sv% was better, and 7 out of 10 that PP Sv% was better.

I always thought that was due to the league as a whole playing tighter hockey in the post-season; in other words, lower shot quality. However, if those effects are to be discounted as an urban myth, then what explanation could there be for that rise in playoff Sv%? Do all the goalies in the league get better in the playoffs, or do all the shooters get worse?

Matt said...

Well, some of it IS that the goalies are better, simply because the (poorer) backups play ~0% of the games, rather than 10-35%.

But I wonder if some of it doesn't have to do with our newish friend, The Score Effect. In the regular season, you still get that 1 point even if you give up the tying goal and then the winning OT goal. In the playoffs, giving up that tying goal is much more disastrous.

I'm not positing the mechanism by which this happens (can't quite get my head around it at the mo), but it's an obvious candidate for investigation.

The Contrarian Goaltender said...

I think Matt has it right. It's mainly because no backup goalies are playing.

Here are the average regular season even strength save percentages for the 16 starting playoff goalies over the last 4 seasons:

2006: .921
2007: .926
2008: .926
2009: .924

Compare that to the playoff averages, and you'll see that mostly explains it. Sometimes the playoff average is a little higher because a few goalies went deep into the playoffs while significantly outperforming their regular season numbers, but on the whole the lack of backups and probably slightly better goaltending overall is what seems to account for the difference.

I do agree that playing to the score effects are even stronger in the playoffs, so that might have a bit to do with it as well. One possibility is that playoff overtime is until somebody scores, and both teams usually play cautiously and put pucks on net whenever they can, which likely means that save percentages are generally higher than normal. Unfortunately I don't have numbers to back that up, but it is something on the to-do list.

Bruce said...

OK, I'll buy "no back-ups" as a contributor to the answer. However, it remains a fact that most goalies have a significantly better individual Sv% in the post season than they do in the regular season. Do they focus more? Face fewer powerplays?

If you are going to adjust for just the 16 starters in the playoffs (as opposed to figuring the difference between all 30 starters and their backups), then you should also adjust for the Sh% of just the 16 playoff teams. Not sure if such data is available on a situational basis.

Vic Ferrari said...

The no-backups thing does seem to eat up much of it.

I'm on board with the 'playing to the score' effect being a touch stronger as well. Though that's just a hunch. I does seem that players get ubercautious late in playoff games, and especially in OT. Beyond what is sensible I think, nobody wants to be the goat, I suppose.

Regular season OT is a different ball of wax imo, the normal risk/reward structure of hockey flies out the window.

I don't know if anyone has ever compared 4v4 stats during the game against 4v4 OT stats ... but I suspect that the latter is way wilder.

Historically in the playoffs there are far fewer 5v3 PPs as well, and a bit fewer 5v4 and 4v3 PPs also.

Do you have the data period by period, CG? First 2 periods save% in the playoffs vs first 2 periods regular season save% should weed most of the score effects and all of OT effects out, no?

The Contrarian Goaltender said...

I agree that first 2 periods save % in the season vs. the playoffs would be a good way to compare the relative contexts. Unfortunately I don't have any regular season data by period.

The playoff data supports the theory that save percentages jump late in games and in OT when the score effects get large. In the 2009 playoffs, it was .911 in the first two periods and .926 in the 3rd/OT. I also have a sample of some of the top goalies between 1994-2009, and the average is .919 over periods 1 and 2 and .925 in the 3rd and OT. Again I'm not really sure how that compares to the regular season though.

The Contrarian Goaltender said...

"However, it remains a fact that most goalies have a significantly better individual Sv% in the post season than they do in the regular season. Do they focus more? Face fewer powerplays?"

Is that a fact? I'm not sure what percentage of goalies have a better sv% in the playoffs. I think you're probably right, I just don't have the numbers on it.

It makes sense to me that fewer powerplays, including fewer 5 on 3s, would contribute to slightly higher save percentages in the playoffs than in the regular season.

Bruce said...

Is that a fact? I'm not sure what percentage of goalies have a better sv% in the playoffs. I think you're probably right, I just don't have the numbers on it.

You're right that it's an observation/assumption rather than a fact. So let's check the career numbers of the top 25 active goalies (regular season GP):

Goalie RegSv% PO Sv%

Brodeur .914 .920
Joseph .906 .917
Kolzig .906 .927
Osgood .906 .916
'Bulin .908 .917
Luongo .919 .930
Vokoun .915 .922
Theodore .908 .912
Nabokov .911 .915
Roloson .910 .915
Giguere .914 .925
Turco .911 .914
Biron .911 .908
Lalime .905 .926
Kipper .912 .921
Weekes .903 .927
Legace .912 .888
Fernandez .912 .927
DiPietro .905 .904
Lundquist .917 .907
Miller .910 .915
Boucher .900 .916
Johnson .903 .931
Hedberg .897 .914
Fleury .907 .916

Hmmm, that's 21 out of 25 that are improved in the post-season, with mean Sv% of .909 and .917. That's pretty convincing.

It makes sense to me that fewer powerplays, including fewer 5 on 3s, would contribute to slightly higher save percentages in the playoffs than in the regular season.

Aye, that makes sense. Are there any stats on powerplays/game in the playoffs vs. regular season?

Vic Ferrari said...

Are those playoff save% numbers even strength, CG?

I suspect that there are fewer PPs in the third periods in most playoff games. Though not as much so as 20 years ago, when they really put away the whistle.

Just by memory, but I remember checking for the effect of the 2nd period long change several years ago (02/03 maybe). And the Evsave% was just about spot on for the 1st and 3rd periods, and about .004 lower in the 2nd period. Again just by memory. It seemed to me that overall 4v4 save% was also .004 lower than the overall EVsave%.

We're down to the fine brushstrokes now, I think.

The Contrarian Goaltender said...

No, those were overall save percentages. I think it's probably true that goalies faced fewer PPs late in games and especially in OT.

What was a bit surprising to me is that the percentage of shots that come on the penalty kill are roughly the same in both the playoffs and the regular season since 1999 (about 20% of overall shots in both cases). I was expecting more of a drop in the playoffs. I guess in recent years the refs really haven't put away the whistle that much.

The penalty kill save percentages did increase, which might be from better goaltending similar to the observed EV effect, or maybe it might stem from fewer 5 on 3 situations.

James Benesh said...

are you sure the lack of backups explains the difference in average sv% from the regular season to the playoffs?

When looking at goalies' stats over the years I have come to accept that it is "par for the course" to see your sv% go up by 5-10 points in the playoffs. Not just for the league, but for starting goalies in general.

If this were true, wouldn't it be a good case for eliminating backups altogether? It would appear that the starting goalie can maintain high performance playing every second night for 1-2 months.

The Contrarian Goaltender said...

"are you sure the lack of backups explains the difference in average sv% from the regular season to the playoffs?"

I don't think it explains all of it, but it looks like that is the main effect.

Re: Eliminating backups, I've been arguing for a while here that goalie fatigue is overrated. Not playing the backup at all would be a bit extreme, but teams are probably better off using their starting goalie a lot.

One thing we really don't know, however, is whether fatigue carries over from season to season, or how much high games played numbers increases injury risk, or whether goalies who play a lot are likely to have shorter careers. Any of those might be something to worry about for a team that wants to play its starter for all 82 games. Within the same season, however, I don't think it has much of an effect.