The top 5 NCAA men’s hockey goalies halfway through the season

Who’s the best goalie in men’s college hockey?

Ask around and you’ll hear many different answers. With so many league-to-league variations in NCAA men’s hockey, it’s difficult to adequately compare goaltenders. Save percentages, goals against, even records can be heavily skewed by the level of competition within conference play. For example, putting up a .925 in the NCHC is a lot tougher than putting up .925 in Atlantic Hockey; 3.50 goals against carries a different meaning in the Big Ten compared to the CCHA. Because of this, we have to dig a little deeper to find any real meaning behind goaltenders’ statistical performances.

In this article, I’ll take you through goalie stats from the first half, focusing on one particular analytical framework of goalie win shares (GWS). GWS calculates the impact of a goaltender on their team’s success. It accounts for shot rate and goal rate, comparing how goalies stack up to their counterparts with other teams.

While GWS models have notable limitations, such as its inability to account for shot danger, looking at win shares can provide a deeper analysis than just your standard save percentage, goals against and record that dominates college hockey goalie discourse. For those who want an in-depth look at the formulas, click this link to read an article where I break down my methods.

Without further ado, let’s see what these calculations say.

[Link to Spreadsheet]

The Top 5 Goalies by Win Shares

1. Chad Veltri, Robert Morris: 2.32 GWS

Veltri is the exact kind of goaltender that gets overlooked by traditional goalie stats. His .922 save percentage ranks 11th among goalies who played at least eight games, and his 3.01 GAA ranks an abysmal 53rd. Overall, his record is 4-12-3 in 19 games played. However, Veltri sits atop the pack when it comes to goalie win shares with 2.32 wins.

On a Robert Morris team making its return to the NCAA fold, there are a lot of weaknesses in the lineup. Those have kept Veltri busy, yet his performance has given his team a chance to win every night — even if the NCAA’s fourth-worst offense hasn’t seized the opportunity. As a graduate student using his COVID eligibility year, Veltri’s next steps in hockey are unknown. But, playing as fervently as he has this season, I wouldn’t be surprised if there are some pro offers on the table.

2. Ryan Bischel, Notre Dame: 2.26 GWS

Another goaltender whose fifth year has gone swimmingly, Bischel plays on a Notre Dame team that has loved his performance. He’s sitting at second in save percentage (min. 8 GP) and second in GWS.

These figures are benefitted by a team that places such a heavy emphasis on defense, but someone has to make the saves regardless of whether there are 15 or 50. Bischel has proven he can come up with the extra saves. Looking at another general stat, goals saved above average (GSAA), Bischel ranks first with a whopping 16.05 GSAA. Even if his team prevents a lot of scoring chances, Bischel is going above and beyond against the ones that slip through.

3. Justen Close, Minnesota: 1.81 GWS

The first half of this season hasn’t been in tune with Minnesota’s Big Ten dominance the past two years. That’ll happen when the bar is set by two Frozen Four runs and being an overtime goal away from a national championship. However, Justen Close’s performance has been solid to start the year. His 1.81 GWS sits at third in the country, even if his .918 save percentage and 2.57 goals against average — the worst since he became the Gophers’ starter — leave more to be desired.

With a young team in front of him, Close has kept up the kind of championship aspirations that his return made seem possible. He’s a goaltender who had a professional career within his grasp this offseason, yet he came back for one more run with Minnesota. If he keeps up his performance, and the team gels around him, they could be in good hands come tournament time.

4. Trey Augustine, Michigan State: 1.77 GWS

Augustine’s long-awaited college hockey debut has been everything you could ask for. He has led the Spartans to the top of the Big Ten with his 11-3-2 record in 17 games. Considering he’s a freshman, his .916/2.97 stat line is impressive. So far, his performance has amounted to 1.77 GWS.

Much like Close, Augustine has a young team in front of him, with much of its talent concentrated on the offensive side. He faced the most shots per minute of any goalie in the conference in the first half, giving him a heavy workload. So far, that hasn’t been a problem.

5. Jacob Fowler, Boston College: 1.69 GWS

All the attention this offseason seemed to focus on the freshman additions of Gabe Perrault, Will Smith and Ryan Leonard to the Eagles. But so far, Jacob Fowler might be the most important contributor of his class. With the loss of starter Mitch Benson to graduation, Fowler has instantly stabilized the net with his .925 save percentage and 2.16 GAA, and especially his 1.69 GWS.

Notably, those stats include a head-to-head series against Augustine, which he swept. The teams in front of them are notable different, but Fowler has proven he’s a notable prospect despite often being cast in the shadow of Augustine. As the two try to win the net for Team USA at the 2024 World Junior Championship, the comparisons between these two elite goaltenders become even more important.

Nuts and Bolts

The Goalie Win Share Formula

First, I want to properly explain the tool we are using for the stats folks who want a little more detail. If you’re just here for numbers, scroll down because you’ll find this boring. The GWS formula is an adaptation of the NHL model made by Justin Kubatko and used by Hockey Reference. I made one slight change to calculate for total wins since points are not universally distributed to NCAA teams. However, the majority of this formula is Justin’s work. To understand each component, I suggest you read Kubatko’s primer on the Hockey Reference link above.

Here’s the formula, in all its messy glory:

GWS = (2 / 7) × ((19 / 12 × ((shots against per minute) / (league shots against per minute)) × (minutes played) × (league goals against per minute) - (goals against)) / ((league goals) / (league points)))

This formula accounts for two factors I find important in college hockey: frequency of shots faced, and frequency of goals against. Shots matter because some goalies get absolutely peppered and allow more goals because of it, while some goalies might face far fewer shots because their defense loves them their team plays a tight checking style. Likewise, goals matter because these directly lead to wins and losses. The GWS formula just puts numbers behind the existing logic that if two goalies have the same stats, but one faces more shots per minute, the goalie who faces more shots is quantitatively better.

While I didn't include the data in the text of this article, you can find another manipulation of this data on the spreadsheet linked in this article: goalie win shares per game. This comparison compensates for the broader usage of a goalie beyond pure minutes. I calculated GWS per game played for each goalie by dividing their total GWS by games played — named the incredibly creative moniker GWS/GP. This helps show us goalies who have performed well in short stints, better reflecting individual contributions to each win rather than just rewarding goalies who play every game. Under plain GWS, goalie who pitches a shutout in seven games could be ranked lower than a goalie who plays adequately for 30. Total GWS rewards the latter, while GWS/GP gives credit to the former. For schools that play fewer games like the Ivys, or goalies who split time in a tandem, GWS/GP better reflects individual success. It also shows which goalies performed well in small stints, especially younger players. This is helpful data for evaluating the long term success of a school's goalie room, as well as any potential impact transfers once we reach the end of the season.

The Limitations of Goalie Win Shares

No formula is without its limitations, and goalie win shares has a pretty glaring one: it doesn’t account for shot danger. To this model, a point blast from me is the same as a slot wrister from Macklin Celebrini. While I appreciate the confidence in my abilities, this is a pretty glaring issue, right?

The same issue exists for the qualitative data points — in GWS, a Hasek-esque sprawling save is equivalent to an easy kick save that a beer leaguer could make. That’s not an issue exclusive to this model — this type of data is still largely collected by the eye test. However, it is a flaw I want to note.

The overarching issue here is the resources we have available. College hockey offers some of the worst public-facing statistics of any hockey league. No one even publishes time on ice, let alone the more advanced stats you’d see in NHL comparisons. Stats like expected goals (xG), zone exits, shot assists — you name it — these are only tracked individually. But this cumbersome undertaking has not been done publicly for every team.

So, out of all the data we have available to compare goaltenders, the GWS formula is the most complete — not necessarily the most thorough. That’s OK, because GWS isn’t the be-all-end-all to goalie performance; it’s just one more lens we can use to determine how players are performing this season.

Required Reading