In our “Stat of the Week” series, Jon examines non-traditional statistics. Emphasis will be placed on how the Oilers are performing according to each statistic.
In this installment of Stat of the Week, we’re looking at a fairly new statistic called Dangerous Fenwick. In case you don’t know, a team or player’s Fenwick For is the number of unblocked shot attempts taken by that team or player. But what is Dangerous Fenwick For? This statistic has come to us from the mind of G Money, and we thank him for being kind enough to answer a few of our questions.
JON: Dangerous Fenwick. Sounds like a nerd armed with a sharp protractor. What exactly is Dangerous Fenwick?
G MONEY: Ha-ha! Well, Woodguy likes to joke that Dangerous Fenwick is Matt Fenwick‘s drunk cousin with a gun! [Fenwick is named after Matt Fenwick, a Flames (*spit) fan and blogger who first came up with the stat.] Dangerous Fenwick is actually just a catchy short form of Danger Adjusted Fenwick. The idea is to take unblocked shot attempts (aka Fenwick) and weight them by their danger. That’s been one of the intuitive concerns that people have had with shot metrics–that they just count a shot as a shot, regardless of whether it’s a 50-foot backhand (which has really almost no chance of scoring) versus a 20-foot slap or snap shot (which is often deadly).
I calculated the danger weighting of NHL even strength shots by looking at five years of data on shot distances and shot types. Using these weightings [available on G Money’s blog with a full explanation of Dangerous Fenwick For (DFF) here] I adjust the shot such that the more dangerous a shot is, the more I count it for (or against). I use a very different methodology Manny Elk‘s xGF metric [xGF = Expected Goals For], but I end up with very similar results to him. So I think we’re capturing something real. The end result tends to match the ‘eye test’ better than does raw Corsi or Fenwick, both for games and for players.
JON: Most analytics geeks are disciples of Corsi. [A team or player’s Corsi in the number of shot attempts by that team or player.] Do you think Dangerous Fenwick is a better stat?
G MONEY: Well, that’s a really good question. As I noted before, DFF (and its ilk) tend to match the eye test quite well. What I take away from that is that DFF is a really good descriptive stat, and in that sense “better” than Corsi. It gives you a reliably objective view of what actually happened. And that has value.
But over a large number of games at the team level, it’s actually mostly on par with Corsi. Corsi actually has slightly higher predictability for the future. What that tells you is that over the long term, highly skilled teams tend to generate shot volume in tandem with shot quality–even though in the short term, shot quality likely dominates in terms of determining what actually happened.
That can be a tough concept to get your head around! The danger weighted metrics are just so much more intuitive. But lots of results in statistics (in all fields) are often counterintuitive that way. I’d say that’s why Corsi is still so well embedded after so many years of people trying to find something better. It’s not great, but it’s still the best we’ve got, especially when it is score and venue adjusted.
Now notice that I said teams are better evaluated using Corsi. I believe over the long run, this is mostly true for players too. But my working theory right now is that there are a small fraction of players who have a meaningful ability to either generate or suppress higher danger shots. It’s a small fraction, which is why I think most studies looking at this effect find the difference washes away. But I do think the effect exists at the extremes.
I’m pretty certain Connor McDavid will prove to be one of those freaks of nature at the top end. I also think Adam Larsson may prove to be similar but opposite–a consistent danger suppressor, as he was in New Jersey, and as he has been so far this season. We’ll see!
JON: Of course it’s early, but how have the Oilers done so far according to DFF? [These questions were asked prior to the game versus Ottawa.]
So far the Oilers have done a terrific job at generating shot danger–top 12 at even strength and top 6 on the powerplay. The shot volume balance on the other hand is really not very good, coming in at 24th overall. This is mostly as a result of giving up too many shot attempts, both at even strength and on the penalty kill.
That said, there is a bit of a caveat in that the Oilers have had several games where they’ve taken an early lead, and this situation (called score effects) can play havoc with shot metrics as teams with a big lead often go into a defensive shell. So the Oilers are doing great at DFF, but not good at CF. What do we take away from that? Mostly that we hope the two converge in the middle.
The other cheery thing is that the Oilers are doing better on score adjusted CF than the raw CF, and I think Score Adjusted CF is a key metric to watch over the course of the season. (I publish CF, SACF, and DFF in my post game analysis posts.)
JON: How likely is it that the Oilers continue to be bottom-12 in CF and top-12 in DFF?
Most of the Oilers indicators are showing a mid pack team, which is pretty much what I’d expect out of the Oilers this year (anything less would be a major disappointment, anything more just doesn’t seem likely given the remaining gaps and weaknesses in the roster).
As I mentioned, I’d like to see the shot metrics converge so that the Oilers as a team end up mid-tier in both CF and DFF. And that should happen over time, given the nature of the divergence. All of the different metrics provide important context about what the team is doing and why. But I believe Score Adjusted Corsi is the key stat to track for the team over the course of the season. Maybe your next “stat of the week”!
SACF and DFF both tend to stabilize between 10 and 20 games (i.e. at that point what you see is almost always going to be close to what you get for the season). I don’t think this early season run will continue, as it’s been driven by McDavid and Talbot more than anything else. When those two guys struggled, the Oilers were beaten badly by a poor and hurting Sabres team. That’s a little scary!
But there are also encouraging signs too. Remember that even if the Oilers are on an unsustainable hot streak, those points are banked. They can’t take them away! So even playing mid-tier hockey the rest of the way might be enough to secure a playoff berth.
A hot streak (or cold streak) has made the difference in making or missing the playoffs for several teams over the years. As Oiler fans, we’ve already been on the cold streak side of the equation lots of times, so riding a DFF-fueled early hot streak to the playoffs … why not us this time? Dare to dream!