Friday, April 27, 2007

Playoff Report: Goalies A Non-Factor

The excellent Hockey Numbers blog is putting up some very interesting statistical analysis during these playoffs, including expected goals for and against for each team in each game, as well as how each goalie played after adjusting for shot quality. This allows us to more accurately measure how much goaltending is helping the winning teams.

The answer so far? Not very much at all. Based on expected goals, 40 out of 46 games so far have been won by the team that was expected to win. Of the other six, five of them had expected goal numbers that were tied or within 0.1, essentially dead heats that could have gone to either team.

In the entire first round of the NHL playoffs, therefore, there was only one game where a team was clearly expected to win based on scoring chances but did not. That was game 2 of Ottawa against Pittsburgh, where Ottawa was expected to win 3.9 goals to 2.0 and ended up losing 4-3. Fleury was pretty good in the game, but the main reason for the loss was Emery's bad play.

So despite all the broadcaster and fan talk about the importance of goalies, there arguably wasn't a single goalie in the entire first round who managed to steal a game for their team, i.e. won a game against a superior opponent because of their outstanding play, and there was only one goalie who managed to blow a game by playing badly. Otherwise, it was the 18 guys up front that decided everything.

In the second round, things are going exactly the same, as all four teams with the higher expected goals won. Note that San Jose, despite getting outshot 34-17 in game 1, actually had more expected goals than the Red Wings, who tend to have a shoot from everywhere style with a low shot quality. This matches my anecdotal viewpoint as well, as Nabokov's saves certainly looked very routine.

Based on shot-quality neutral save percentage numbers for the round, the losing goalies were actually better than the winning ones. Luongo and Lundqvist were the only two winning goalies to clearly outplay their counterparts once shot quality was factored in, and Lundqvist was actually outplayed in the games where Hedberg was in the other net. So the only series where goaltending was a decisive factor was the Vancouver - Dallas series, where Turco was very good but Luongo was better. In round 2, however, it looks like the Canucks are overmatched by the Ducks, despite Luongo.

So if announcers or analysts talk about a goalie stealing a series, realize that this is an extremely unlikely scenario, since in these playoffs goalies aren't even stealing games. It's a convenient and lazy way to make a prediction, but simply comparing the goaltenders is a poor method because it is the rest of the team that really decides who wins or loses.

No comments: