It is all-star weekend and this past week Sportsnet.ca made their picks for “Analytics All-Star Team”. Their eastern conference picks are here and you can find their Western Conference picks here. Their criteria for picking the rosters was simple. They wanted to pick the most fun team to watch and that meant all-out offense.
With that in mind, we thought we’d try selecting teams using a statistical basis, shooting for the most possible fun.
The rules are pretty simple. In order to get the best quality game, you want players who generate tons of offence and play a fast transition game, so we will ignore defensive play entirely.
Despite defense not being a factor in team selection, I wanted to take a look at how these offensive oriented players performed defensively. In particular, I wanted to see how these all-fun all-star teams have impacted their teams save percentages. So, here are each players 5v5 Sv% Rel over the past 2 1/2 seasons (2014-15, 2015-16 and the first half of this season). Here is what we find.
(Note: Werenski, Matthews and Radulov have limited sample size with only half a seasons worth of data. Radulov had a -1.33 Sv% Rel in his earlier stint in the NHL)
It seems clear, a good offense is not the best defense. Just 6 of the 36 players (16.7%) had a positive impact on their goalies on-ice save percentage while 30 of these players had negative impacts on their goalies save percentage. On average the group had a negative impact of 0.54 percentage points.
It is important to remember that these are not players pre-selected or hand-picked by me to make a point. These are players independently selected by someone else on the basis of what analytics says about their offensive ability. Can we really claim that these observations are due to randomness alone?
Now I am not saying that these are bad players. They are not. Generally their offensive ability far out weighs their negative impact on save percentage. However, when comparing a Pavelski (+1.36 Sv%Rel) to a Kane (-0.97 Sv% Rel) the better defensive play of Pavelski may offset some or all of the better offensive play of Kane. What I am saying is that players can, and do, impact their goalies save percentage and for a complete and true evaluation of a player one must take this into consideration.