Applying Sabermetrics to Hockey II - Pitching Runs (Goalie Goals)

Once again, diving into baseball Sabermetrics has proven useful for some hockey analysis; proving exceptionally true when looking into goalie analytics.

Goalies get the short end of the stick when it comes to advanced hockey metrics. War-On-Ice is doing some great work with shot quality by counting not just scoring changes, but high, medium, and low danger shot opportunities. They have also developed an adjusted save percentage that has proven quite useful. However, goalies just do not have the depth that skaters do when it comes to analysis.

We are constantly reminded of a few points when looking into goalies. Most specifically, that save percentage is probably your best bet when analyzing goalies. Skaters have an extremely limited effect on goalie save percentage. On top of that, goals against average can be viewed as more of a team statistic, than an individual goalie statistic. Just like ERA in baseball. 

Pitching Runs was introduced in 1993 by Thorn and Palmer as a way to get a look at how many runs a pitcher saved or lost against the league average. The formula takes the pitcher's innings pitched, multiplies it by the league ERA divided by 9, and finally subtracts the pitcher's earned runs from that value. 

This measure benefits pitchers for efficiency and durability as it rewards them for pitching more innings. A reliever and a starting pitcher who have the same ERA will not have the same Pitching Runs, because the starting pitcher is more likely to have more IP than the reliever.

Let's get off baseball.

The thought today was to apply Pitching Runs to goalies in a statistic we will now be referring to as Goalie Goals. To get this metric, we will substitute the formula above as follows:

Innings pitched = Goalie TOI
League ERA = League GAA
9 = 60
ER = GA

Our data set will be from 07-08 through 14-15, 5v5, goalies who appeared in 25 or more games in a season (15 or more games in the lockout shortened 12-13). 

We also obtained two metrics. One for Goalie Goals as a reflection of each season, and one for Goalie Goals as a reflection of the full data set. Our numbers saw a minor change, as the league average goals against average has not fluctuated all that much since 07-08.

Don't love this graph because it makes a small range seem like a large one. But, as you can see, the range does go from 2.206 in 07-08 as a min to 2.319 in 09-10. 

Top ten Goalie Goals with a seasonal baseline:

Yellow highlighting denotes that season's Vezina Trophy Winner

Yellow highlighting denotes that season's Vezina Trophy Winner

And top 10 Goalie Goals using a league average GAA for the full data set:

Yellow highlighting denotes that season's Vezina Trophy winner

Yellow highlighting denotes that season's Vezina Trophy winner

On a season by season basis, the goalie who won the Vezina Trophy with the lowest Goalie Goals statistic was Ryan Miller in 09-10 with a Goalie Goals of 6.77. That season, there were 15 goalies with a better Goalie Goals metric than Ryan Miller. Martin Brodeur led the league that season, and finished 3rd in Vezina voting. Ilya Bryzgalov finished 2nd in the Vezina voting, and 4th in Goalie Goals that season. 

Further analysis:

The sum of our total Goalie Goals metric comes back to near 0 in both metrics, showing that each goal 5v5 is accounted for.

Goalie Goals correlates strongly to save percentage.

All Vezina winners: