There are a lot of problems with goaltending in the NHL.
Problems may be a misleading term, but it’s the best that I can come up with. The traditional thought had been “goaltending wins championships,” and then the 2013-2014 season happened; Jonathan Quick backstopped the Los Angeles Kings to a Stanley Cup, and finished the playoffs with a .911 save percentage. The NHL average in the regular season was .914. Quick was technically below-average, and still added a second Stanley Cup ring to his collection.
Problems, indeed.
This article is just an outline of the misconceptions that surround goaltending, and in turn, has significant fantasy impacts. Some might seem counterintuitive, some might not. All the same, there are truths and falsities that surround the position that require some clearing up.
Problem #1 – Goals Against Average (GAA) is a team stat, not a goalie stat.
Most fantasy hockey leagues have at least three roto stats for goalies: wins, save percentage, and GAA. Some also have total saves or shutouts as an additional category. GAA is a staple, though, and will be found in almost every roto fantasy hockey league.
Goals against average is also flashed on broadcasts alongside save percentage, as if goalies have control over their GAA. Here’s the thing: they largely don’t.
Here’s what I mean:
- In the 2013-2014 season, Ben Scrivens and Brian Elliott both finished the year with a .922 save percentage, both saving 92.2-percent of the shots they faced. Scrivens had a 2.56 GAA while Elliott finished with a 1.96.
- The year before, James Reimer and Antti Niemi both finished the year with a .924 save percentage, both saving 92.4-percent of the shots they faced. Reimer had a 2.46 GAA while Niemi had a 2.16.
- In 2011-2012, Josh Harding finished the year with a .917 save percentage, and Marc-Andre Fleury finished with a .913 save percentage. Harding saved 0.3-percent more shots than Fleury, but had a 2.62 GAA while Fleury’s was 2.36.
Despite saving shots at the same rate, or better rate, than their counterparts, Scrivens, Reimer, and Harding all had considerably worse goals against averages. Despite this, those promoting GAA as a goaltender stat would lead readers/listeners to believe that Elliott, Niemi, and Fleury are the superior goalies. This is a failure to understand basic mathematics. Here’s an example:
Goalie A faces 30 shots in a game, saving 90-percent of them, for a save percentage of .900. His goals against, though, would be 3.00.
Goalie B faces 20 shots in that same game, saving 90-percent of them, for a save percentage of .900. His goals against, though, would be 2.00.
Despite both goalies saving pucks at the exact same rate, Goalie B gets the win, and has a much more impressive GAA. The rate of shots faced, though, dictated the goals against, not the rate at which the goalies saved the puck.
While good save percentages will tend to give a lower GAA (saving more pucks, after all, means fewer goals against), the rate of shots faced is a huge factor as well. That’s why targeting goalies from Ottawa, Toronto, or Buffalo will not help your fantasy team’s GAA.
Problem #2 – Goaltending is extremely volatile
A goaltender’s save percentage can take a long, long time to normalize. In fact, there’s work done by Eric Tulsky, formerly of Broad Street Hockey, that shows goaltenders can take upwards of 5000 even-strength shots faced before a goalie’s true talent level shows through. That’s nearly three full seasons worth of shots. That’s why some people (myself included) were fooled by Jonathan Quick’s 2011-2012 season, when he finished 3rd among regular NHL goalies (min. 41 games played) in save percentage with a .929. Since that season, he’s 24th out of 31 goalies with at least 60 games played in save percentage at .910, but 5th out of 31 in goals against average. He should thank his team defense.
There are lots of examples like Quick (Corey Crawford and Mike Smith are a couple others that come to mind), and that’s why looking at just one year worth of stats is, essentially, useless. Only one goalie cracked 1900 even strength shots faced last year (Semyon Varlamov), which means most other regular goalies had one-third of a reliable sample last year.
Here’s how volatile save percentage is in practice:
- Over the last three seasons, just two goalies have played 30 games and had a save percentage over .920 in each season, Henrik Lundqvist and Cory Schneider.
- Over the last three seasons, just seven goalies have played 30 games and had a save percentage over 920 in each season, the two mentioned, plus Sergei Bobrovksy, Brian Elliott, Jaroslav Halak, Jimmy Howard, and Tuukka Rask.
Seven goalies have been able to post what would be an elite fantasy save percentage in two of the last three seasons. Seven.
Problem #3 – Backups can post elite seasons
Goalies are able to outperform their regression in small samples, and a small sample can extend as far as a full regular season (this is why some of this stuff is counterintuitive).
Think about this:
- In a standard 12-team roto ESPN league, three of the top-10 fantasy hockey goalies last year were backups (Cam Talbot from New York, Martin Jones from Los Angeles, and Alex Stalock from San Jose). Those standard leagues include wins, goals against average, and save percentage.
- The reason for this is that two-thirds of the goalie categories are ratio stats (GAA and save percentage). The three goalies mentioned only had 12 wins each, and none played more than 24 games. That said, each of them had a save percentage above .930 and a goals against average under 2.00.
- These aren’t their true talent levels, but the extremely good ratios posted made them elite fantasy goalies, and more valuable than names like Fleury, Niemi, Luongo, and Quick.
Understanding goaltenders is an ever-evolving subject that requires a lot more education than simply staring at a box score. The problem lies in that goaltending categories make up at least 30-percent of overall roto stats in fantasy hockey.
Remember, there are very few goalies that are elite year in and year out in fantasy hockey. Heck, even Henrik Lundqvist finished outside the top-10 last year. Knowing and accepting that a fair amount of their production is random, though, will help fantasy hockey owners sleep better at night.
*As always, thanks to Behind The Net, Hockey Reference, and Hockey Analysis for their resources.
