I was having a conversation on twitter with Jesse Marshall (a must follow) about Brandon Sutter, and we were discussing whether Sutter is legitimately great at defense due to his high year-to-year on-ice save percentage. At even strength over the last three years, Brandon Sutter put up an on-ice save percentage of 94.4%, 93.7%, and 93.1%. Those are all really high, so one could plausibly see this and think that Brandon Sutter possesses elite defensive skill.
But if that were so, why is he paid so little? Why are his possession numbers so bad? Why does the eye test make him look so bad? More importantly, why isn't every team lining up to snap him up from us? In addition to these questions, we should always be skeptical of goal-based stats (even over big samples) because it takes such a long time for the signal to separate from the noise.
I got to thinking that part of this might be driven by ice time and competition. Let's first talk about shooting percentage. The forwards who play a lot of minutes are generally really good, and we would expect them to have a higher on-ice shooting percentage because of that. Conversely, we'd expect forwards who don't see a lot of ice time to have a lower on-ice shooting percentage because they're not as talented with the puck. To figure out if this is true, I pulled all the forwards from extra skater who played 62 or more games this year and popped their ice time and on-ice shooting percentage into a scatterplot (please click all images to enlarge them).
There's definitely something there. An R-squared of 0.2 or more is a moderately positive relationship, but I'm inclined to give it more effect here because we can't control for as many things in hockey. So my takeaway is that there's substantial evidence for the claim that players with more ice time generally have higher on-ice shooting percentages.
Now onto save percentage. If top six players only played against other top six players, we'd expect all of them to have a (relatively) low on-ice save percentage. And if bottom six players only faced bottom six guys, they'd all have high on-ice save percentages. The graph below contains the data.
Nothing there. The R-squared is practically zero, which suggests that ice time is not correlated at all with on-ice save percentage. This also makes intuitive sense, since coaches do not exclusively use top six players against top six players and bottom six players against bottom six players. The upshot is that we don't see a league-wide relationship between time on ice and save percentage.
Nevertheless, I wanted to break things down on an individual level. Over the last three years, Brandon Sutter averaged a time on ice QoC% of 28.2% and an on-ice save percentage of 93.7%. That QoC is really low; if you break down the QoC among forwards this year into quintiles, Sutter's QoC puts him in the bottom 20%. Guys who see competition that easy are, by definition, skating against guys who don't see much ice time. We know from above that guys who don't play a lot have a low on-ice shooting percentage, so perhaps at the extreme, guys with very easy competition have high on-ice save percentages.
To figure this out, I went through all of the forwards who played at least half the games in each of the last three seasons and recorded each one who (a) had a time on ice QoC% of 28.2% or below each year (Brandon Sutter's three-year average), and (b) had an on-ice save percentage of 93.7% or higher each year (again, Sutter's three-year average). The list of guys who did this all three years and two of the three years is below.
|Two Years||Three Years|
You'll notice how aggressively unimpressive this group is--there are some real duds there in Ryan Jones, Mike Rupp, and Aaron Volpatti. I also don't even know who Brian Flynn is. And keep in mind, there were equally low-quality hockey players who didn't make the cut because their on-ice save percentage was a tad off. The list would thus be bigger and even more bland if we broadened our parameters just a bit.
But look at those three names who did it all three years. John Mitchell is a decent guy who has bounced around from the Leafs to the Rangers and now to the Avalanche, all on short deals that paid little money. Trevor Lewis has been with the Kings' organization his whole career but has been relegated to short contracts in six figures. And Shawn Thornton is an awful hockey player who assaults people on the ice (note that while we're on the topic of useless enforcers, Sutter's "defensive prowess" is shared by Colton Orr and Cody McLeod, two face-punchers who should not be in this league).
The point? Sutter isn't a unique defensive asset. When you actually look at players who routinely see the light competition he's faced the last three years, a chunk of them have high on-ice save percentages over one, two, or three years. GMs don't invest much term or money into these guys, and both stat heads and casual fans recognize that these players are not very good. If we thought their high on-ice save percentage was a product of innate talent, then GMs would be sitting on a pool of seemingly elite shutdown players who are underpaid and vastly undervalued.
I don't think that's true. When you get into the bottom 20% of players in terms of quality of competition, things change. Their opponents get little ice time because they're offensively challenged, which we see at the left end of the first graph above. The fact that their goals against numbers are good isn't because of their talent but rather their deployment. So if most players can provide "good defense" against weak competition, the key is to get guys who are cheap possession-drivers. Sutter is neither, which means Jim Rutherford needs to find a way to trade Sutter before October 9th.