Save % Variance and Updated Predictions

Goaltender Game by Game Save %

One of the things I’ve been interested in looking at is the idea of game-by-game goaltender save percentage. One of the problems, in my view, with save percentage as a metric is that it doesn’t take into account games as distinct events. If a goalies stops 48/50 shots in a game, the standard aggregate save percentage metric views that as being twice as important as a goalie who makes 24/25 shots in a game, even though the individual save % for each goalie in each game is identical.

This has the effect of potentially over rewarding good goalies who face a lot of shots, while punishing goalies whose teams are better at preventing shots. To provide an alternate view, I started looking at the game-by-game average save % and game-by-game variance in save % for each goalie this year. The idea behind this was to provide a view as to how a goalie performed on a night-by-night basis, and to see whether the additional variance data might reveal which goalies played more consistently at a higher/lower level and which had large swings in game-by-game save %.

The table below shows the standard save % (Sv%), average save % (Avg Sv%) and save % standard deviation (Sv% StDev) for all goalies who played in at least 24 games for the 2012-2013 season sorted in decreasing order by Sv% StDev. I should note that I excluded games where a goalie’s save % was below 0.500 as it tended to skew the numbers heavily for certain goaltenders.

Player Sv% Avg Sv% Sv% StDev
ELLIOTT, BRIAN 90.68% 90.00% 9.26%
RINNE, PEKKA 91.17% 90.24% 8.61%
BOBROVSKY, SERGEI 93.17% 91.95% 7.53%
SCHNEIDER, CORY 92.92% 92.47% 7.44%
SMITH, MIKE 91.04% 90.86% 7.23%
BRYZGALOV, ILYA 89.96% 88.99% 7.10%
HOWARD, JIMMY 92.29% 91.28% 6.98%
DUBNYK, DEVAN 92.05% 91.20% 6.81%
PAVELEC, ONDREJ 90.49% 89.63% 6.71%
NABOKOV, EVGENI 90.96% 90.52% 6.48%
FASTH, VIKTOR 92.13% 91.88% 6.34%
PRICE, CAREY 90.73% 90.52% 6.33%
VARLAMOV, SEMYON 90.27% 89.56% 6.25%
BRODEUR, MARTIN 90.06% 89.75% 6.15%
HOLTBY, BRADEN 91.99% 91.54% 6.10%
FLEURY, MARC-ANDRE 91.60% 91.46% 6.09%
CRAWFORD, COREY 92.59% 92.53% 6.06%
QUICK, JONATHAN 90.41% 89.78% 5.88%
BACKSTROM, NIKLAS 91.27% 90.80% 5.84%
HILLER, JONAS 91.26% 90.75% 5.50%
LEHTONEN, KARI 91.62% 91.43% 5.46%
MILLER, RYAN 91.50% 91.03% 5.46%
LINDBACK, ANDERS 90.19% 89.39% 5.44%
RASK, TUUKKA 92.86% 92.72% 5.42%
ANDERSON, CRAIG 94.09% 93.52% 5.09%
REIMER, JAMES 92.36% 92.12% 5.03%
NIEMI, ANTTI 92.38% 92.16% 4.89%
LUNDQVIST, HENRIK 92.61% 92.13% 4.67%

At the top of the list you’ll see a few unsurprising names: Brian Elliot didn’t really get it together until later in the year after a terrible start, and the rest of the top 10 (Cory Schneider excepted) is littered with goalies from marginal playoff teams or non-contenders. It’s not necessarily a list of guys you’d want to build your team around.

On the opposite end Henrik Lundqvist has been terrifically consistent all year and is likely the reason the Rangers were able to get into the playoffs in spite of a horrifically low shooting percentage. The same goes for Craig Anderson of the Sens who has been both lights out in terms of save % and variance.

What’s also interesting to note is that for every goalie on the list the average save % is below the standard save %, possibly confirming my initial thought that games where a goalie faces more shots may be driving save % higher. What I haven’t looked into is which is a better measure of a goalie’s “true” ability, although I suspect that neither are going to be particularly good predictors.

I’ve also put together year over year comparisons for each goalie that played at least 16 games in each of 2011-2012 and 2012-2013, shown in the table below.

Player 2011-2012 Avg Sv% 2012-2013 Avg Sv% 2011-2012 Sv% StDev 2012-2013 Sv% StDev
ELLIOTT, BRIAN 93.91% 90.00% 5.24% 9.26%
HALAK, JAROSLAV 92.14% 90.20% 5.89% 8.98%
CLEMMENSEN, SCOTT 90.41% 87.64% 8.29% 8.84%
RINNE, PEKKA 90.83% 90.24% 8.43% 8.61%
VOKOUN, TOMAS 90.46% 91.81% 7.77% 8.47%
MASON, STEVE 88.45% 91.22% 7.57% 8.14%
KIPRUSOFF, MIIKKA 91.84% 87.14% 5.16% 7.71%
HEDBERG, JOHAN 91.03% 87.69% 8.44% 7.55%
WARD, CAM 90.84% 89.77% 5.84% 7.54%
BOBROVSKY, SERGEI 89.50% 91.95% 7.32% 7.53%
SCHNEIDER, CORY 94.39% 92.47% 4.44% 7.44%
EMERY, RAY 90.28% 91.36% 6.43% 7.36%
LUONGO, ROBERTO 90.81% 90.54% 7.64% 7.28%
SMITH, MIKE 91.78% 90.86% 7.39% 7.23%
BRYZGALOV, ILYA 89.74% 88.99% 8.89% 7.10%
HOWARD, JIMMY 91.26% 91.28% 6.40% 6.98%
DUBNYK, DEVAN 90.63% 91.20% 5.76% 6.81%
PAVELEC, ONDREJ 89.78% 89.63% 7.45% 6.71%
GIGUERE, JEAN-SEBASTIEN 91.91% 91.26% 4.94% 6.51%
NABOKOV, EVGENI 90.58% 90.52% 6.99% 6.48%
PRICE, CAREY 91.34% 90.52% 4.98% 6.33%
VARLAMOV, SEMYON 90.89% 89.56% 6.40% 6.25%
BRODEUR, MARTIN 89.89% 89.75% 7.33% 6.15%
FLEURY, MARC-ANDRE 90.34% 91.46% 6.11% 6.09%
CRAWFORD, COREY 88.96% 92.53% 7.94% 6.06%
QUICK, JONATHAN 91.92% 89.78% 7.18% 5.88%
BACKSTROM, NIKLAS 90.12% 90.80% 7.46% 5.84%
GARON, MATHIEU 88.81% 88.80% 8.39% 5.76%
HILLER, JONAS 90.37% 90.75% 5.82% 5.50%
LEHTONEN, KARI 91.88% 91.43% 5.38% 5.46%
MILLER, RYAN 90.50% 91.03% 7.28% 5.46%
LINDBACK, ANDERS 90.99% 89.39% 6.64% 5.44%
RASK, TUUKKA 92.03% 92.72% 7.16% 5.42%
ANDERSON, CRAIG 91.35% 93.52% 5.58% 5.09%
REIMER, JAMES 87.86% 92.12% 9.62% 5.03%
NIEMI, ANTTI 90.54% 92.16% 7.38% 4.89%
LUNDQVIST, HENRIK 92.67% 92.13% 4.67% 4.67%

It’s interesting to note that while Sv% StDev doesn’t seem to have much consistency year over year, Henrik Lundqvist actually posted an identical number in 2011-2012. Jonas Hiller and Kari Lehtonen have also been steady year over year, coming in below 6% Sv% StDev in each of the last 2 years.

One of the bigger changes year over year is James Reimer, who sits with the 3rd lowest Sv% StDev for 2012-2013 after posting the worst number of all goalies in our sample last year. Reimer has also increased his Avg Sv% number by nearly 4.5% this year, and his turnaround is likely the reason the Leafs were able to make it into the playoffs. Leafs fans can probably also breathe a sigh of relief that they didn’t end up picking up Miikka Kiprusoff, who went from being one of the better and more consistent goalies in the league in 2011-2012 to the worst goalie by Avg. Sv% in 2012-2013.

Updated Round 1 Predictions

Updated odds for the first round series are shown below. I haven’t added the updated Cup Win % yet, but I hope to get it up after all teams have finished their 3rd game (this was quickly thrown together with an Excel macro which unfortunately takes a while to run. I’m hoping to optimize it and get these up on a more frequent basis soon).

I should note that while the odds take into account the results of the games played so far, the odds of winning an individual game haven’t been changed in the model. As an example, the model viewed the Bruins as having a 69.3% chance of winning any game at home vs. the Leafs when the series started, and the model still views them as having those odds. The Leafs odds of winning the series, however, have gone up after winning game 2, as the Bruins now have fewer paths to victory.

Team May 2 May 5 Change
Chicago 79.68% 93.23% 13.55%
San Jose 44.54% 80.33% 35.79%
Boston 80.40% 73.69% -6.71%
Washington 41.80% 73.66% 31.86%
Anaheim 44.28% 63.78% 19.50%
St. Louis 41.46% 61.48% 20.02%
Pittsburgh 64.16% 58.11% -6.05%
Montreal 57.36% 52.20% -5.16%
Ottawa 42.64% 47.80% 5.16%
NY Islanders 35.84% 41.89% 6.05%
Los Angeles 58.54% 38.52% -20.02%
Detroit 55.72% 36.22% -19.50%
NY Rangers 58.20% 26.34% -31.86%
Toronto 19.60% 26.31% 6.71%
Vancouver 55.46% 19.67% -35.79%
Minnesota 20.32% 6.77% -13.55%

The Sharks late comeback in game 2 pushed them to have the highest rise over the past 3 days. Having won 2 on the road the Sharks are now roughly 80% favourites to advance. Vancouver is probably in the most trouble of any of the top seeds, and their odds of advancing are now around where the Leafs were to start the playoffs.

Washington, St. Louis and Anaheim have also made big leaps, with each team being more than a 60% favourite after having gone into their respective series as underdogs. Montreal-Ottawa is now the closest series and the model views it as essentially a coin flip. If Montreal can take one of the next 2 they should head home as the favourites going into game 5.

Looking at key games coming up, I’ve included a table below with what I’ll call the “leverage” for the next game for each series. What it shows is the difference in expected series win % for the top seed between a win and a loss occurring in the next game of the series. To give an example, if St. Louis wins game 4, the model would expect them to win the series 86% of the time. If they lose, the model would have them as underdogs, with only a 47% chance to win the series. The difference between 86% and 47% (39%) is the leverage for the game. The higher the number is, the more important the game is in determining the result of the series.

Top Seed Bottom Seed Leverage
St. Louis Los Angeles 39.47%
Anaheim Detroit 38.72%
Montreal Ottawa 38.66%
Pittsburgh NY Islanders 36.95%
Boston Toronto 31.90%
Washington NY Rangers 29.45%
Vancouver San Jose 26.85%
Chicago Minnesota 12.09%

Chicago has the most room for error of any team, as even with a loss they’d still sit as 86% favourites to move on. San Jose is also in a good spot, and would still be favourites even with a loss in game 3.

Anaheim, in spite of taking back home ice advantage in the third game of their series, has the second most important game coming up, and would be in essentially a coin flip situation with a loss, being only 52% favourites.

Advertisements
Tagged with: ,
Posted in Goaltending, Predictions

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: