[Editor’s Note: This is a guest article that was submitted to us here at Defending Big D by Raajan Aery. We appreciate the stats-driven look from a non-partisan hockey fan, and hope you enjoy it too.]
With the trade deadline coming up, the rumor mill has been churning at full speed. One rumor getting some sustained coverage is that the Dallas Stars are interested in acquiring Stanley Cup champion and Pittsburgh netminder Marc-Andre Fleury. Presumably, the Stars’ attraction to Fleury is driven by traditional goaltender analysis which heavily values Sv%. Kari Lehtonen ranks 38th in the league in save percentage with .902 and Antti Niemi clocks in at 40th with .901. Comparatively, Fleury is ranked 35th in save percentage with .906.
To determine an NHL goaltender’s value, most traditional hockey analysts examine a goalie’s wins, GAA and Sv%. These stats, while historically relied upon, are not ideal in many circumstances. Both wins and GAA are heavily reliant on team performance, and are unable to isolate a goaltender’s specific contribution. Sv%, on the other hand, unfairly incorporates goals on the penalty kill, while also not contemplating even-strength shot quality.
Reliance on these statistics can lead to poor decisions. Goaltending is arguably the most valuable position on a hockey team. Unlike skaters, goaltenders play the entire game. Just a few mistakes by a team’s goaltender can easily turn victory into defeat. Goaltenders are disproportionately important, so misevaluating them can be ridiculously devastating.
From an advanced statistics or “fancystats” perspective, it seems that Fleury is not an upgrade over either Lehtonen or Niemi. Indeed, the Dallas goaltending tandem’s advanced stats indicate they are performing quite well, while Fleury’s advanced stats suggest poor performance this season. Both Dallas goaltenders’ traditional Sv% numbers undervalue their contributions to the team at least partially because the Stars ice the 2nd worst penalty kill in the league.
According to hockey statistician Rob Vollman, 4v5 Sv% is not useful for evaluating a goaltender’s particular contribution to his team. He claims this is because “power-play opportunities are so dangerous that even a short hot or cold streak can have a disproportionate impact on a goalie’s year-end save percentage.” Essentially, a lot of the variation seen in traditional Sv% numbers is due to disparities in teams’ total penalty minutes and tactics. Therefore, in order to evaluate goaltenders properly, we’ll need to filter out 4v5 Sv%.
A simple solution to this problem is to just use 5v5 Sv%. This season Dallas is posting a 5v5 Sv% of .929, which is good for 7th in the league, conclusively above average. Lehtonen’s, 5v5 Sv% of .932 is good for 15th in the league, while Neimi’s 5v5 Sv% numbers of .924, though a bit lower than Lehtonen’s, still peg him as an NHL starting goaltender this season. From this perspective, the Stars have iced an effective tandem this season.
Another statistic to bear in mind is shot location. Rob Vollman asserts that shot location statistically corresponds well to the elusive and abstract concept of shot quality. Essentially, when a shot is closer to the goaltender and is square to the net, a shot is more likely to go in. This problem is dealt with by separating shots into HDSv% or high danger save percentage, MDSv% or medium danger save percentage, and LDSv% or low danger save percentage. Emmanuel Perry, the hockey statistician from corsica.hockey, has asserted that HDSv%, of all Sv% statistics, best reflects a goalie’s innate skill or talent. Essentially this means that a goalie should typically be able to repeat his HDSv% numbers. To read more about the location breakdown of these statistics and the methodology behind them, please read Emmanuel Perry’s blog post here.
This season, Lehtonen and Niemi have posted nearly identical HDSv% numbers. Lehtonen ranks 15th in the league in the category with .828, and Neimi is right behind at 17th with .827. These numbers suggest that the tandem is playing at the level of an average NHL starting goaltender. Given that Dallas faces a lot of these types of shots relative to the rest of the league, it seems like Niemi and Lehtonen are pretty much the backbone of the Stars’ defensive system. This means that the Stars’ problems this season must come from somewhere else.
Meanwhile, Marc-Andre Fleury’s 5v5 Sv% this season is .917, which places him at 34th in the league. Additionally, Fleury’s HDSV% is ranked 42nd in the league at .782. Based on these metrics, Fleury has had a season that would position him as a decent backup goaltender, which is how Pittsburgh has been deploying him this season anyway. Ultimately, the Stars would probably not stand to benefit by giving up assets for a goaltender who makes their current options look like brick walls. Based on the numbers, I believe that slotting Fleury in net for the Stars would result in even worse outcomes for the Stars on the scoresheet.
At this point, we might wonder why the Stars have been losing this year despite above average goaltending. There may be a few statistics that we can look at to come up with some ideas. Perhaps the Stars are losing more this season partially because Dallas’ team shooting percentage has predictably regressed from last season’s league leading 10.1% to this season’s 12th place 9.1%. Certainly, this cannot be blamed on either Lehtonen or Niemi. One issue may be the Stars’ power-play. Dallas’ power-play has dropped from 4th in the league in 2015-2016 to 20th in the league this season, contributing to a large decrease in overall S%. The Stars’ 5v5 S% has dropped from 7th in the league last season to 16th this season as well. This might suggest that the floundering offence is to blame in addition to the team’s penalty kill.
While the statistics I have provided have somewhat limited use to predicting future success, they are certainly more useful metrics to determine how a goaltender has performed in a season than traditional statistics. Additionally, they allow hockey nerds to take an approach that is more quantitative than the eye test. This can be useful in finding and exploiting undervalued talent or to avoid making mistakes in asset management.
When looking at Lehtonen and Niemi, the traditional stats don’t paint the whole picture. They’ve both performed well this season, are undervalued, and have even improved from last season in many metrics. The Stars would benefit from looking at advanced statistics to come up with a fresh diagnosis and to avoid worsening their situation at the trade deadline.
Raajan Aery is a law student at Osgoode Hall Law School in Toronto. You can follow him on Twitter here: @raajanaery.