Projecting a Goalie’s Save Percentage in Fantasy Hockey

Advanced fantasy hockey managers know that one of the keys to consistently winning at fantasy hockey is having access to accurate player projections. For example, knowing that Corey Perry would score less than 40 goals last season (and not 50 goals) would have prevented you from drafting him at too high a position. We wrote such an article last year (Will Corey Perry Repeat His Incredible Year) and our projections were based on methods that had been tested robustly (not just simple hunches).

Our aim with today’s article is to give you a starting point on how to create your own projections for the category of goalie save percentage. This is generally the category most attributable to a goalie’s skill level and can, in fact, be used to project many of the other fantasy hockey categories used for goalies. At Left Wing Lock, we do this each Summer for every goalie in our Fantasy Hockey Draft Kit which became available to all fantasy hockey managers on August 2.

In two recent articles here at Left Wing Lock (Penalty Kill Save Percentage and The Even-Strength Shutout), we argued that there is very little (if any) difference in skill level for goalies when their team is on the penalty kill. The two most important conclusions from these articles are repeated here: the league average PKSV% (save percentage while on the penalty kill) sits firmly at .875 season after season and no goalie who has faced more than 400 PKSA (shots against while on the penalty kill) has been able to maintain a PKSV% greater than .892.

With this data in hand, we can now begin to develop a method for projecting goalie save percentage. First, you need to understand that a goalie’s save percentage is made up of three distinct units: even-strength save percentage (EVSV%), save percentage while on the penalty kill (PKSV%), and save percentage while on the power play (PPSV%). As a quick warning, be aware that some sites reverse the abbreviations for PPSV% and PKSV%. If you’re grabbing data from a website, be sure you know which category you’re actually looking at. With that said, how much does each unit contribute to a goalie’s overall save percentage? To answer that, you need access to data on all shots taken during the three different types of shifts. It turns out, Left Wing Lock has that data going back many seasons. For this article, we’ll use data that is appropriate to the 2011-2012 season. From this, we know that 82.1% of all shots are taken at even-strength, 15.34% are taken while the goalie’s team is on the penalty kill, and the remaining 2.56% are taken while the goalie’s team is on the powerplay.

As a quick exercise, we’ll examine how to compute the overall save percentage of a goalie with knowledge of these three units. Our internal data tells us that the league average for EVSV% last season was .9207. The PKSV% was .8752 and the PPSV% was .9066. Thus, to compute the league average for overall save percentage, you would perform the following calculation:

= (.8210)*(EVSV%) + (.1534)*(PKSV%) + (.0256)*(PPSV%)

Using the league averages we noted above, you arrive at the following figure: .9134. It turns out, that this figure agrees completely with the simple method of adding up all the league saves last season and dividing them by the total number of shots faced. So, we have an internally consistent method here for computing save percentages. If I asked you to make a best guess for a goalie’s save percentage next season (but I didn’t reveal the goalie’s name to you), you should respond with .9134.

How can we use the above to project a goalie’s save percentage for the 2012-2013 season?

As a first approximation, we can assume that the number of shots faced on the different types of shifts at the league level is a pretty good estimate of what each team will face. Yes, there will be differences, but your odds of guessing which teams will have more powerplays than other teams next season is likely to introduce greater error than this approximation.

With that said, we’ll choose three goalies for our exercise: Cory Schneider, Marc-Andre Fleury, and Antti Niemi. First, here is some data on all three goaltenders:

2011-2012 Save Percentage Data for Three Guinea Pigs
Goalie EVSV% PKSV% PPSV% Total SV%
Cory Schneider .9307 .9588 .9600 .9365
Marc-Andre Fleury .9136 .9053 .9000 .9119
Antti Niemi .9251 .8388 .9211 .9140

The first thing we can do here is make corrections to the PKSV% and PPSV%. We’ll see what each goalie’s overall save percentage would have looked like had they performed at the league average (for PKSV% and PPSV%). We arrive at the following numbers: Schneider – .9216, Fleury – .9075, Niemi – .9170. This is not a bad first step. We simply used the EVSV% from last season, adjusted the PKSV% and PPSV% to league average levels and recomputed the overall expected save percentage. But we can do better than this.

Last year’s EVSV% is not the best place to start when estimating a goalie’s EVSV%. A much better approach is to use his career EVSV%. The career EVSV% numbers for each goalie in this article are: Schneider – .9273, Fleury – .9172, Niemi – .9244. Armed with this data, we’ll plug the career EVSV% numbers into our save percentage formula along with league average values for the PKSV% and PPSV%. We arrive at the following projections:

2012-2013 Save Percentage Projections for Three Guinea Pigs
Goalie SV %
Cory Schneider .919
Marc-Andre Fleury .910
Antti Niemi .916

With some basic research and reasonable approximations, we’ve crafted save percentage projections for three goalies who are likely to come off the draft board early. Will Cory Schneider really fall to .919? Now that you know how his .937 SV% of last season was produced, you have a better idea of how to answer that question.

You can certainly make goalie save percentage projections more complicated than the method outlined above. But, as a tangible and practical strategy heading into your fantasy hockey draft, this method should provide you with defensible and accurate (hopefully) projections.

If this type of player analysis appeals to you, you might be interested in our Fantasy Hockey Draft Kit which provides you with projections on all NHL players (goalies and skaters, alike). Note that our projection method in the draft kit is slightly more involved than the method outlined above, but built up using the same principles.

Comments (6)

  1. cgjoe

    Very good, thorough analysis of goalie ability. Is there any way to factor in the quality of the defense in front of the goalie, or is that more of an intangible? For example, a goalie on team with a poor defense might face higher quality shots or a greater number of shots per game, while a goalie behind a strong defense might face easier shots and fewer shots per game.

    For instance, Brodeur has consistently played behind a tight defense that allowed fewer shots on goal per game on average. Therefore, a goal against Brodeur would count more against him than a goal against a goalie who typically faces 40 shots a game. I don’t have those statistics at my fingertips, so I would be interested to see how shots faced per game impacts a goalie’s fantasy value.

  2. cgjoe

    I agree that the “shot quality” statistic is largely statistically irrelevant. I think it plays into the goalie’s other numbers, which are much more statistically relevant. EVSV% is a very good indicator of a goalie’s worth. Then you can plug in number of games expected to start to get an idea of what their point totals might be.

    I’m in a points league where goalies are rewarded for starts, wins, saves, and shut outs, and penalized for losses and goals against. It’s hard to predict shut outs; typically there will be between 4-8 for any given starting goalie. However, a goalie on a good team (top 4) will get more wins than a goalie on a bad team (bottom 4); thus, team standings projections play an important role as well. I probably wouldn’t take Luongo if he went to a team like Columbus; I think Crawford would be a more attractive pick due to Chicago’s potential this year (and I would likely be able to get Crawford in a later round). Heck, I might even take Lindback over Luongo if Lou goes to a bottom-4 team!

    Thanks for the feedback and reading list!

  3. Rocket

    Mike, this is a deadly article. This is why I follow this site religiously. You guys have new and innovative ways to look at goalie statistics. I will be using this formula to determine my list of goalies (if we have a season). Keep them coming.

  4. Mike

    I really glossed over this fact in the article, but the reason you can safely use the average PPSV% for any NHL goalie is the following:

    PPSV% accounts for approximately 2.56% of the shots faced by a typical goalie. I looked back over several years of data and nearly every goalie falls within .850 and .950 for their PPSV% year after year. If you do the multiplication, you find that at the extremes this adds either .022 or .024 to a goalie’s overall save percentage. At the league average of .9066 (which we suggested using in the article), you’d be adding .023 to a goalie’s overall save percentage. Using the league average then guarantees that you’ll never be more than .001 off. So, there really is no sense in digging up all that data on PPSV%. Just use the league average there.

Leave a Comment

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>