Login

grok · 03-30-2019, 03:47 PM

I get lonely sometimes. Being a goalie is sad work, because everyone who does all these funtimes exciting advanced stats work only cares about forwards and defensemen. So I got inspired when Baelor did skater point shares, reached out for a blessing, and am proud to present...

S46 Goaltender Point Shares

Ok, you might say...what does this mean? Well, goaltender point shares are calculated similarly to skater point shares, but we calculate them based on Goals Against Average per Minute Played, compared to the league average. When calculated, a Goalie Point tells us roughly what amount of the team's league points can be attributed to a goaltender's performance in a season. Let's break down how we got there, using Hockey Reference's calculations for goaltender point shares.

Step 1: Stats we need

In order to start, we need to pull some stats from the sim database. We will need the following goaltender stats:

Shots Against
Minutes Played
Goals Against

We will also need some league stats:

Total Goals, all teams
Total League Points, all teams

From there, we will need to do some simple sum stats:

Total Shots Against, all goaltenders
Total Goals Against, all goaltenders
Total Minutes Played, all goaltenders

Step 2: Creating the baseline

Next, we're going to need to calculate some baseline stats. The league stats help us create something to compare against, and see if a goalie outperformed the field or underperformed. First, we need our League Shots Against per Minute (SA/m) and League Goals Against per Minute (GA/m). They are calculated very similarly:

SA/m = (Total Shots Against / Total Minutes Played)

GA/m = (Total Goals Against / Total Minutes Played)

We also need to know what goes into a win. How many goals are required to earn a league point? We find that, our Marginal Goals per Point (MG/P) with the following equation:

MG/P = (League Goals / League Points)

To follow along, you can see these stats on the horribly formatted second page of my spreadsheet. Or you could pretend it doesn't exist.

Step 3: Finding Goaltender Stats

Now, we need to start comparing a goaltender's stats to each of the league stats we derived earlier. First, we have to ensure judging is fair by seeing if each goalie faced more or fewer shots than a league average goalie. This derived stat is called the Shots Against Adjustment (SA Adj), and is most commonly somewhere between .800 and 1.200. Think of it as a percentage--a goalie with a 1.12 faces about 112% of the shots an average goalie faces. We find the SA Adj with the following formula:

SA Adj = (Goalie Shots Against / Goalie Minutes Played) / (League SA/m)

Next, we're going to find our Marginal Goals Against (MGA) for each goaltender. If you read Baelor's write-up, you probably remember the 7/12 appearing, and it's back here. We use a 7/12 coefficient in this calculation to account for each player's contribution to each side of the game. Hockey is a two-way sport, and each skater is responsible for two parts of the game--offense and defense. A goalie plays the same 60 minutes, but most goalies are only contributing to defense. So for each skater, we have 1 point of offense and 1 point of defense. The goalie breaks this, contributing 2 points to defense. So we use a coefficient of 7/12 to identify the 7 "points" contributed to the team's defense--each skater gives 1, and the goalie gives 2.

We're going to take a coefficient (1+(7/12) and combine it with our SA Adj, MP, League GA/m, and our goalies' goals against to find each goalie's MGA. Ultimately, this tells us how many total goals against we could expect each goalie to allow if they were playing behind a league average team:

MGA = (1 + (7 / 12)) x (SA Adj) x (Goalie MP) x (League GA/m) - (Goalie GA)

Step 4: Point Shares

We made it. One last thing though--we need another coefficient. Each goalie contributes two points to defense, so we can only credit our goaltender with 2/7 of the total defensive point shares. Goalie Point Shares can be calculated with the following formula:

Goalie Point Share = (2/7) x (MGA / MG/P)

---

And there we have it! Our point shares have been calculated. Since I'm in the mid-800 words, I'm going to go ahead and give my take as well, with three really interesting observations from the spreadsheet.

1. Benjamin Blue had a higher point share than any skater in the league this season. If you do a shitty proportion up to an 82 game season to hack an idea together like I did (50x = 10.41*82), you get an estimated 82-game point share of 17.07. That puts Blue's season up there with Cam Ward and Tim Thomas in 2007, a point shy of Dominik Hasek's greatest seasons in the NHL. Contextualizing to SHL, only one other player came within a full point of Blue--Kata Vlide's 9.57. 3rd place (Zhumbayev) was a full 1.3 point share below Blue. Cain was almost 2 points below the leader. Blue had a dominant season and the point shares tell us he should be in the MVP conversation come awards time.

2. Kata Vilde's season was way better than his SV% shows. In our model, Vilde is the 2nd most important goaltender to their team, and a big part of that is his minutes played. Vilde saw a ton of TOI and a ton of shots, and still cobbled together a season saving more shots against than the average. Vilde did really well and hopefully earned some conversation for the McBride.

3. Prepare for the attack of the S43'ers. Three S43 draft picks placed in the top 6 of all goalie point shares in S46--Alesteir Cain (4th - 8.54), Johannes Leitner (5th - 8.17), and Cedric Robinson (6th - 7.91). Team success definitely has something to do with it, but these three goaltenders are clearly making an immediate impact on the league as young players.

Pingy Pingu · 03-30-2019, 07:54 PM

HAI GROKKALO OOGLEY BOOGLEY

***luke*** · 03-30-2019, 07:59 PM

I like it

*JSS* · 03-31-2019, 07:28 PM

Blue for MVP Barracuda

***HoltzFan10*** · 04-01-2019, 10:14 AM

@grok this is absolutely amazing, props to you. This article is going in my “self loathing” folder

Navigation

Extra Menu

About us