Goals-Against Average Should Be Abolished from the NHL's Statistical Lexicon

Jonathan Willis@jonathanwillisNHL National ColumnistJuly 16, 2014

Boston Bruins' Ken Hodge flips puck past New York Rangers' goalie Terry Sawchuk in third period of their NHL playoff game at Boston Garden, April 10, 1970.  Boston won 5-3 to lead the Stanley Cup East Division semifinal playoffs 2-0. (AP Photo/A.E. Maloof)
A.E. Maloof/Associated Press

For hockey fans of a certain era, it was the single most important goaltending statistic.

Goals-against average, generally reduced to the acronym, GAA, is simple to calculate. It’s merely the number of goals allowed by a goaltender over 60 minutes of play; i.e. the total number of goals allowed per full-length game.

It’s an old idea, going all the way back to the birth of the NHL. When Toronto (just Toronto; the team had no official name in 1917-18 because it was a stand-in for the Eddie Livingstone-owned team that was barred from the league) won the Stanley Cup in 1918, they used just 12 skaters all season and starter Hap Holmes posted a 4.73 GAA.

That’s ancient history, of course. In a roundabout way, that’s the point. GAA is a relic of the past, a statistic with no value outside the historical for the entirety of my lifetime—a dated reference that should have been dropped around the same time Return of the Jedi opened in theatres.

George Widman/Associated Press

In 1983-84, the NHL started tracking the save percentage of goaltenders league-wide. The new statistic was not only manifestly superior to GAA, but rendered its forebear completely redundant.

This is obvious if we think about what GAA really is: save percentage plus noise.

Let’s consider a hypothetical example: A goaltender who plays a full 60-minute hockey game and makes 27 saves on 30 shots. Obviously, his GAA his 3.00 (3 goals/1 60-minute game) and his save percentage is 0.900 (27 saves/30 shots).

But we can also calculate his GAA another way. A goalie’s save percentage is always going to have a fixed relationship to his opposition’s shooting percentage; in this case, the 0.900 save percentage means that the opposing team scored on 10.0 percent of their shots. If we multiply that number by the total number of shots faced per game, we also end up with our goalie’s GAA (0.1 * 30 shots per game = 3.00).

In other words, GAA is just a goalie’s save percentage combined with the number of shots the team in front of him allowed.

UNIONDALE, NY - FEBRUARY 09:  Pekka Rinne #35 of the Nashville Predators makes a save as John Tavares #91 of the New York Islanders looks for a rebound on February 9, 2010 at Nassau Coliseum in Uniondale, New York.  (Photo by Jim McIsaac/Getty Images)
Jim McIsaac/Getty Images

The latter number can best be described as “noise.”

A goaltender has no influence on the number of initial shots he faces. Some goalies allow more rebounds than others, but at the NHL level the gap is so small as to have no practical effect; as Rob Pettapiece of NHL Numbers notes, the regression is so severe that an analyst is better off predicting an NHL-average control rate rather than drawing any inference from an entire season of play.

Naturally, if GAA is just save percentage plus a bunch of factors outside of a goalie’s control, it follows that save percentage is a manifestly superior statistic for measuring goalie performance. GAA adds nothing that save percentage doesn’t already cover, and the addition of noise only serves to obscure a goalie’s true talent.

We shouldn't use GAA at all. It should be dropped from all discussions of goaltender achievement, it shouldn't show up in press releases from teams or reports in newspapers or even online comment sections. Stats lines shouldn’t even include it.

For 30 years now, the NHL and the people who cover it have been referencing a statistic that adds nothing to the conversation, one that actually serves to confuse rather than clarify. It’s time that stopped.

Jonathan Willis covers the NHL for Bleacher Report. Follow him on Twitter for more of his work.

Statistics courtesy of Hockey-Reference.com.