Wednesday, September 21, 2011

A Thought Experiment about Goaltending, Fenwick, and Lineup Construction

Before I go any further, I must note that this piece is an introduction, and to be honest, I don't quite know exactly where it will go.  I do know that I'd like you, the readers, to give any input in the comments, as this topic will certainly be revisited in the future.

One of my favorite elements of NHL analysis is looking at line-up construction in the context of the salary cap.  My favorite team, the Blackhawks, are like a number of other teams in that they will almost always spend to the externally imposed cap, but frankly, this analysis can extend to every team in the league, as they will almost always be working with internally imposed constraints.  

The question is simple: What is the best way to put together a line-up that can consistently compete to win a championship?

Here is where Fenwick and Goaltending come into play, to illustrate my point, I ask these questions below:

What is the lower bound of goalie performance for a team with legitimate Stanley Cup aspirations?  Is it saving 87% of shots? 88%? 89%?

What about Fenwick (puck possession)? Is it 44%? 46%? 48%?

Spending big money on a goaltender likely comes at the expense of signing better skaters.  What is the optimal solution? Is it the team that has a goalie saving 90.5% of Even Strength shots but has a Fenwick of 55% more likely to win a Cup than the team that has a goalie saving 93% of Even Strength shots but gets crushed territorially, only managing to Fenwick 46%?

While we can safely assume that there is a correlation between salary and ability in goaltenders, we cannot escape the reality that dollars are fixed, and money spent on goaltenders cannot be spent on skaters.  This makes it difficult for teams that spend large chunks of salary on goalies to build teams that dominate at even strength. There are exceptions to this to the first rule, both in terms of mediocre goalies making big money (Cam Ward), elite goalies making back-up money (Tomas Vokoun, though this is only one year), and in terms of teams who have managed to build a team with excellent goaltending and excellent puck possession (Montreal and Vancouver come to mind), but by and large, teams that spend big money on their goaltenders put themselves at a competitive disadvantage when it comes to competing for a Stanley Cup.

My position is that goaltending, like rebounding in basketball, doesn't matter until it matters.  That is to say it is not as important as say, winning at even strength.  I recognize that this is a controversial position, and I hope that this opens up some discussion.  I will come back to this with more data throughout the season, but right now I want to limit this to a philosophical discussion.

Boston showed us last year that a team can be mediocre 5v5 and still win a cup, so long as the goaltending is on point, but there are few who peg them as favorites to repeat, which is pretty telling when we consider that most of their team will return.  People generally recognize, whether it is conscious or not, that relying so heavily on your netminder is not a recipe for success over multi-year timeframes.

Winning at even strength is important, but are we in the statistics world overrating its importance?  Is it as important as having a world-beater between the pipes?  How does a rising cap change this picture?

Let us know your thoughts, we believe this is a pretty interesting topic.


  1. Interesting stuff, Matt.

    I've always thought the best way to go is to spend very little on a goalie, $1-2M, so more money is available for skaters. If you look at Detroit since the lockout they've been the most consistently good team and most of that was with Osgood.

    A big issue with goalies is that they are tough to evaluate. The difference between the best and worst in the league at ES save% is about 3 percentage points. Especially with potential team effects and most giving the even smaller sample playoffs more weight, it takes a while to figure out if a guy is elite and the market doesn't give you that kind of time. If you want to have a top goalie long term you are either going to have to gamble or pay a bigger premium, possibly for a vet whose skills might diminish. Sometimes those gambles pay off, but a lot of the time they don't.

  2. Well, the flip side is that Detroit played Edmonton in 2006 with a vastly superior squad in most ways and their 6-figure goaltending (Manny Legace at the time) cost them that series. The Detroit Model is all well and good, but it sure in hell doesn't always work.

    Further with Edmonton, in the '80s the Oilers won with ~league average save percentage and ~league average shots for/against (no Fenwick available), but an out-of-this-world team shooting percentage. So they didn't exactly fit either model, all the while dominating the league. But percentages do exist at both ends of the rink, even as they are generally more attributable to a single player at the one end and are more of a team thing at the other. So from a salary cap perspective those are very different questions.

    Interesting ideas here, Kent. Look forward to seeing this develop.

  3. I think you need to find your Tomas Vokouns and become an exception to the first rule. Teams that handicap their cap situation like Carolina or the New York Rangers become hard pressed to add the right mix of talent and depth to compete. Even though Buffalo opened their wallets this offseason, the season will still hinge on the performance of Ryan Miller.

    The Panthers might be the test-tube baby in this process. They spent a lot of money on a lot of second-tier forwards, but they should probably be a better possession team than last season. Of course, they sacrificed having one of the best goalies in the league to one of the worst.

    Very cool thoughts. Good read.

  4. An issue similar to what Cam says about Vokoun is having guys on entry-level deals or their first contract after joining the league. Tampa Bay took a cap hit of $3.75M for Stamkos last year.

    The Canucks give us a good example of this as well. 2 or 3 seasons ago they were goalie-salary heavy team. 08/09 their cap hits for the Sedins were 3.5M, Kesler 1.75, Raymond 883K and Burrows 483K. Those guys have improved since then but that's basically their current top two lines for 11 million or so. Luongo's hit was 6.75M. Now they have a high skater payroll and probably close to average for goalies because those guys have signed pre-UFA deals and Luongo has the frontloaded extension. It's pretty much the same core though.

    Looking up those salaries bummed me out since both Pavol Demitra and Rick Rypien were on that team. What a horrible offseason.

  5. I love this idea, but I think it will be hard to find what percentages are best in the real world, simply because the percentages fluctuate wildly.

    I look forward to seeing where you take this though.

  6. What you're saying here is one of the reasons why I'm worried about Carolina this season. They were not a good possession team last year and Ward was the main reason they came so close to earning a playoff spot. If an injury happens to him or his play drops off, things could get very ugly.

  7. Geoff-

    Thanks, we have a few ideas that we're in the process of going through. There'll be a follow up soon.


    Yup, plus their situation is made worse because they are not a cap team.

  8. Nice post and I love the idea. I tried to look at something similar using GVT and GVS last year for Hockey Prospectus

    The main conclusion was that spending at any given position didn't guarantee success, but that a team was more likely to get good value from goaltenders if they paid league average or lower.

    I think the key, as Bruce pointed out, is not that there's any one way to construct a team, but that the GM gets the best value from the dollars spent. Look at the transition in baseball from OBP/OPS players being undervalued to now there are stellar defensive players being undervalue.