Just how good are the Edmonton Oilers?

It’s a Friday in December and the Edmonton Oilers are not
only in playoff position, but they’re there on merit. The really scary thing?
If the underlying numbers are to be believed, the rest of the season may go
even better.

53.5% Score-Adjusted
Fenwick

NHLPuck

Advertisement - Continue Reading Below

As of today, the Oilers’ score-adjusted Fenwick rating is
53.5%
. Analytics-inclined readers don’t need a lot of explanation as to the
meaning of that number, while those on the opposite end of the spectrum aredoubtless
suppressing a comment about spreadsheet hockey in this very moment. For anyone
in-between those two extremes, though, this deserves a bit of time.

First, the metric itself. Score-adjusted Fenwick, to my way
of thinking, is the Cadillac of shot metrics. Fenwick is a giant plus/minus of
all shots and missed shots—it’s like Corsi, but doesn’t penalize teams for
blocking shots. The negative of that is the sample size is a little smaller. The
positive is that it tends to have a closer relationship to goals and scoring
chances.

It’s a leading indicator. Goals and even scoring chances bounce
around a lot, but given time tend to converge at Fenwick.

Adjusting that number for score gives a clearer picture. Teams
that fall behind early in games can look better by the shot metrics than they
actually are, because trailing teams tend to be more aggressive. Likewise,
teams that take a lead early tend to be more conservative. This, incidentally,
is where most of the ‘they keep losing when they out-shoot the other team!’
noise comes from – go down 2-0 early and you’ll probably out-shoot the other
team, but you’ll also probably still lose.

Advertisement - Continue Reading Below

What’s important, though, is that 53.5 number. It’s fifth in
the NHL right now, on-par with what San Jose and Pittsburgh managed
last year
. It’s also basically equivalent with what Chicago and Tampa Bay did
in 2014-15
. In other words, the Oilers are controlling play at five-on-five.

This is the first time that’s happened in the analytics era.
Here’s what the team looks like year-by-year (I’ve included the head coach in brackets):

  • 2007-08 (MacTavish): 46.2%
  • 2008-09 (MacTavish): 46.8%
  • 2009-10 (Quinn): 44.8%
  • 2010-11 (Renney): 44.6%
  • 2011-12 (Renney): 47.3%
  • 2012-13 (Krueger): 44.9%
  • 2013-14 (Eakins): 43.6%
  • 2014-15 (Eakins/Nelson): 47.0%
  • 2015-16 (McLellan): 47.7%

Edmonton’s been a bad team for a long time, and it shows
here. The increase this season is stunning, so stunning that I’m not sure it
can last. Even if it doesn’t, the Oilers are primed for their best possession
season in forever.

What about goals?

Oilers fans

Edmonton is not just out-shooting, but it’s also out-scoring
its opponents. The Oilers have scored 53.1
percent of the goals
at 5-on-5 this season, a number almost exactly in-line
with their Fenwick rating.

Advertisement - Continue Reading Below

Again, this compares reasonably well to good teams.
Pittsburgh scored 55% of all 5-on-5 goals in its games last year, while San
Jose managed 53%. Going back a year further, Chicago was at 54% and Tampa Bay
at 57%.

Put in more common terms, the Oilers today have a plus-10
goal differential, the fifth-best total in the league. If it continues—and based
on how they’re scoring their goals, it will—they should finish around the
plus-33 mark on the year, again basically on par with the Pens and Sharks a
year ago.

Keep doing that and the wins will come.

The standings don’t perfectly reflect goal differential
(there’s a lot of noise in NHL standings, with much of it due to the league’s overtime/shootout
policy) but they generally come close. Over the last 10 seasons, just eight
teams have missed the playoffs while posting a goal differential of +10 or
better; only one has missed the playoffs with a goal differential better than
+20 (Colorado, +21, 2006-07).

If Edmonton keeps scoring 53 percent of the goals and taking
53 percent of the shots, we won’t find ourselves wondering whether or not they’re
a playoff team. Instead, we’ll be wondering just how far in the postseason they
can go. 

Advertisement - Continue Reading Below

  • I am Batman

    I’m really looking forward to what Chia is gonna do as a buyer on the trade deadline.

    Go all out and win now or be more conservative? Being a seller is easier because nobody expects you to succeed immediately.

    Again, I think we will see how good is Chia this trade deadline, I happen to think this team as is doesn’t go far, but with a few key pieces, watch out

  • Jay (not J)

    Pretty [email protected]%&n’ good this morning. We’ll see how good they look Monday morning. Too early in the season to abandon hope.

    But we’ll see Monday morning.

    On another note, does the fact that players and coaches know that Corsi /Fenwick are being tracked compromise them as indicators?

    • That’s a question I get all the time. I haven’t really seen signs of it happening, and I also tend to think that the answer is ‘no’ for a few different reasons.

      1. Players like to win. Deliberately taking a bunch of shots from long range – something which is easily verifiable, btw. – would be a strategy that leads to losing, which goes against what they’ve been trying to do their whole careers.

      2. Coaches like to win. Perhaps more importantly, their jobs depend on it – post a losing record long enough, and your team’s Corsi won’t save you.

      3. Doing things to try and inflate Corsi may actually damage it. There’s an opportunity cost to every shot. If we take the typically imagined scenario – a guy shooting a lot from bad places – he’ll get one shot, but then the other team will get and keep the puck, leading to shots against. Shooting from good locations generates rebounds and second & third opportunities, while also preventing the opposition from getting shots itself.

      I’ve heard people talk about players or teams ‘gaming’ Corsi, but I’ve never seen actual evidence and I don’t think trying to do it would work in the real world.

      • ROil Family

        This discussion reminds me of Jordan Eberle’s comments about the Eakins PP in the spring of 2015:

        To quote from Matt Henderson’s ON article “Jordan Eberle made an interesting comment today as to why the PP is doing better under Nelson. As per Derek Van Diest, Eberle had this to say: “I think maybe on the start of the year we were really focused on shooting the puck and Corsi numbers. When that starts creeping in your head, you’re just shooting the puck just to shoot it and you’re not trying to create the best opportunity you can.”

        http://oilersnation.com/2015/3/16/gaming-corsi-on-the-power-play

        I think it’s dangerous if the players start trying to boost their corsi. Positive shot differentials are an outcome of strong play and possession, not the other way around.

        • Gravis82

          That is dangerous, agreed. See my post above. Corsi only works as a model of scoring. If you try to increase your corsi by fundamentally changing the approach to playing hockey by just shooting from places where you are not likely to score from and never would have considered shooting from before, you immediately invalidate the metric. In that case, corsi becomes a predictor of corsi, which in itself is not valuable. It needs to be used to measure and compare teams who are not trying to game it. This risk is low though, because teams will find out that if they try to game corsi, results will be bad (eakins again).

          Every model has assumptions. The assumption of corsi is that teams use it to evaluate how well they are doing at trying to score. Once you start using corsi as the outcome of interest (i.e. instead of goals), the model fails.

          So what does this mean? That there is something we are not measuring.

          Corsi —-> something else ——> Goals.

          Corsi/fenwick is predictive of something, which is the true predictor of goals

          That something else is likely of shots from high danger areas, stratified by player skill. Theoretically you can see that teams with more shots overall likely also have more shots from high danger areas, and a higher number of shots from those high danger areas will be goals with players of more skill. But as soon as you try and increase your corsi by shooting from low dangers areas, you lose this association. Just figure out where these high danger areas are and we are good. But the thing is that the high danger area will be different depending on what the skills are of each player (accurate wrist shot vs. hard slap shot) and formation on the ice (maybe if the slot is plugged up then high danger area is off to the side, not the middle).

          To really model what is likely to occur you need to account for each player on the ice, and where each player is most dangerous + the most likely defensive formation of the opposing team. Combining those two things together you can then find where the soft spots are and also which of your players will have the most success in those soft spots.

          Now that would be powerful, and I wouldnt be suprised if teams are doing this. With a bit of work, this would be possible now.

          Doing this in game is more difficult. You would need to know at each moment where the soft spots are, and which player would be best to shoot from where depending on both formations. This would require a supercomputer, but we have one!!!! His name is Conner McDavid!

          • Gravis82

            Eberle fails to realize when he has entered into a high risk shooting zone. He is not bad at getting there despite his size, he is bad a recognizing when he is there. Because his definition is likely based on a different assessment, ego (which is not a bad thing, everyone has one, and successful people usually have larger ones). But I think its hurting him. He wants to see the shot and beat the goalie, man on man.

            Seeing a shot with a bit of traffic or it looks like the goalie could stop it, present some uncertainty. Some sense that if it goes in its not me, its just luck. Which takes some of the satisfaction away from scoring, which is why players like this pass up those chances. And they probably don’t even know thats why.

            Send him to a sports shrink asap!

          • Natejax97

            Eberle just needs to hit the net. Missing the net so much chases us out of the zone. At least if he hit the net there might be a rebound.

            He is trying so hard for that perfect shot…lol…sports shrink maybe you are onto something there.

          • bazmagoo

            I think Ebs is just in a slump and is over thinking his chances. Some times putting your head down and shooting blindly is just as effective as a dipsy doodle with a perfect snap shot at the end. Mix it up!

        • acesaaron

          Eberle’s comments about trying to shoot to boost Corsi are certainly dangerous for a player to think about but, until yesterday, I thought the pendulum on the PP had swung too far the other way. It seemed like the powerplay was trying so hard to get the greatest chance possible and not simply direct pucks to the net. Balance is what is needed and that should hopefully come over the course of the season.

          It is up to the coaching staff to keep the powerplay from getting stale and predictable by changing looks, attacking schemes, players on different units etc. One change to the first unit PP and look what happens. It may not be all on Letestu but I’m sure the Jets penalty killers were not looking at a whole lot of Letestu PP video to prepare. Keeping the opposition on their toes is how coaches earn their keep. Let’s hope this continues.

      • S cottV

        Whether or not its to pad the corsi, I think the Oilers tend to take too many premature low probability shots at net.

        As you say – bad shots lead to turnovers and the puck going the other way for shots at the wrong net.

        Looking at NHL stats 5vs5 shot attempts – the leaders are

        Boston 1129 948 +181

        LAKings 1120 952 +158

        The differential would indicate more quality offensive zone time with not as much fluff from too many low quality shot attempts, comparing with the Oilers numbers.

        Oilers 1130 1043 +86

        To me – the top end teams tend to spend more time with possession in the o zone and get a higher differential, as the by product of this possession. They enter the zone more in the mind set of lets get it in the zone and establish stable possession and then work from there, to work the puck toward a move to the net.

        The Oilers mindset – is lets gain the blue line and then take it to the net.

        A better balance between the two above mind sets, is what still sets the improving Oilers from the top end of the league.

        Sure – the Kings are behind us in the standings, but when they get healthy again, with their above numbers – I doubt we stay ahead of them, unless we make changes to the way we play.

    • Gravis82

      It would only compromise them as indicators if the only way to get paid was based on those metrics (e.g. deliberate attempt to inflate these metrics at the expense of goals/assists/wins).

      But currently that is not the case. Getting paid is based on wins and goals and points. Players and coaches have every reason to focus on that reward system. What reason would they have to make a conscience decision to improve Fenwick while sacrificing goals/assists/wins? None.

      But saying that, if a player or team did try to ‘game the system’ by increasing Fenwick, and were successful at it, production would likely increase anyway, as would success. But of course, that is an experiment we cannot do. It is likely that it could fail as well though if you change your approach too much (aka, Eakins and corsi). So it is important to focus on scoring, and then fenwick is ok. If you focus on fenwick, then fenwick measures fenwick and not scoring

      These metrics are a proxy for goals/assists/wins. So, why use them at all? Well because there are simply way more shots in a game than goals. When you learn about statistics and its application to real world decision making and forecasting (used by every company imaginable to predict how we will buy their things next christmas), you realize that the higher the sample size you have the more precise picture of a future state you are able to construct.

      Over many seasons it would be possible to do this type of analysis using goals/wins, but players/teams change so much over multiple seasons that this would be a bit too inaccurate for decision making (although still likely interesting). To get an accurate forecasting dataset within 1 or 2 seasons, you need to increase your sample size…thus the utility of fenwick.

      While this is not important to most fans, it is extremely important to those in the business of hockey (or should be). Given proper training and knowledge on how to interpret results, adjust for limitations and apply findings, teams that take advantage of this approach will have a leg up on the competition.

      But its not just haphazard application here and there that will lead to success, you can only use this information to make better decisions if you truly understand the underlying theory of the metric, the theory behind why these numbers are likely to model more conventionally measured real world phenomenon, and how that happens.

      Eakins is the perfect example of this. Lets get more shots from everywhere to increase corsi!! Sure, that must work. No. The theory of corsi is that it measures teams that are actually trying to score. What Eakins did was change his team into one that was trying to create more shots at the expense of trying to actually score. So in that scenario, corsi was not predictive of anything and the model breaks down.

  • Hazlehurst11

    And with this Oilersnation is going crazy talking 2nd and 3rd round of playoffs this year. God I hope they make it there! Play like last night and they will

    • Gravis82

      I think the 3rd and 4th lines are fine. And we have lander who is great in faceoffs on the farm if need be. We also have Davidson, who hopefully returns this season…which will no doubt improve the D corps.

      1st line is awesome with Drai obviously, Lucic is fine there for now.

      Success of Pouliot Nuge Ebs for the rest of the year is the biggest unknown. How they preform will define the season.

      Also I JP should go back to finland. He cant speak a word of english, so I dont think going to bakersfeild will be useful. Send him back to finland and force him into english lessons.

  • whateverhappenedtoearledwards

    “It’s a leading indicator. Goals and even scoring chances bounce around a lot, but given time tend to converge at Fenwick”

    Some sources have been stating that scoring chances are the best metric, and so far this season this may be the case for the Oil. In a number of their high shot total losses scoring chances have been close or even favoring the other team. I’ll grabt that a scoring chance is more subjective than if a shot is taken or not.

    • I’m less worried about subjectivity than I am about predictive value.

      Here’s how I see it: Scoring chances are a better reflection of the balance of play in an individual game than shots are. But if I have 20 games of shot/scoring chance data and I’m trying to predict chances for the next 60 games, the shot data is going to do a better job of prediction.

  • freelancer

    Hey Willis, how do those numbers change with McDavid on the ice vs without? I’m guessing it’s safe to assume they are stronger with him but I would like to know how much worse the team is without him.

  • bazmagoo

    Too inconsistent to make the playoffs. I see us finishing 9th or 10th in the conference, exactly the same place as I thought at the beginning of the year. We don’t play well enough against the Ducks, Sharks, Yotes, and Kings. Hate to be a party pooper, but we rely way too much on McDavid.

  • GCW

    I prefer to look at the individual game states, but we do see improvement there as well.

    Tied: 52%, 8th
    Down one: 59.4%, 3rd
    Up one: 50.9%, 8th

    In past years they were bad in these metrics.

  • It’s great to have a winning team where the underlying numbers support the result. This is not the Flames or Colorado of last year and two years ago.

    What’s crazy is as they get a better back up, as Nurse, Klefbom, Larsson, Draisaitl, McDavid, and Puljujarvi progress, as Chi gets that right shot D who can quaterback a PP, and as potential core players like Cagguila, Shlepyshev, Pitlick, Kaharia, and Benning all progress, this team is only going to get better and better.

    Settle in Oiler fans, cause it feels like the days of the infinibuild are finally over.

  • Natejax97

    Hahahahahaha Fenwick, Corsi, Vodgeman Sledgehammer…whatever.

    Last night – Grit won. Skill is always there for Oilers, they will always get lots of shots and always have the puck. The team is built to be a possession team. When they play with grit, they can beat any team in the league (on display last night). When they play with only skill, it’s a recipe for disaster (see leafs game November 29th).

    I liked the coaches comment after Toronto game – he said that the leafs made checking a priority, and that the oilers didn’t. Oilers outshot the leafs, had better chances, but summed up they did not put emphasis on checking. I understand MacL’s words to mean body and stick, and ensuring that the other team has a miserable night in all 3 zones.

    Looked to me last night that the oilers put an emphasis on checking. Scary track meet at times…but much more concerted effort to make Winnipeg’s life miserable.

    We do this for the next 55 games, we will be okay.

    • Gravis82

      when you play with grit, you will yourself into high danger areas….and score more. Which is what they did last night. You also will yourself into those areas in the D zone, and attenuate the chances for the other team.

      Grit and advanced stats are not in opposition. They are related. You need grit to get to where you are most likely to score.

      Advanced stats just measures who and what teams do this on a more regular basis. IN this sense, advanced stats are a measure of grit as much as they are a measure of shots…because both are required to get into a high danger zone and score or generate an assist

      And, the best players are better at finding these areas and taking advantage. Some use more grit that shot, some use more shot than grit. But its unlikely that you can rely on just one or the other. Different ratios work for different players under different deployments

  • giddy

    The Oilers goal differential is interesting. When we win, we seem to love to win by big margins. 7 of the 13 Oilers wins were by 3+ goal leads. Yet the Oil were only beat by 3+ goal leads twice, which means we’re blowing teams out but not getting blown out ourselves.

    If we cut down on all the stupid goals given up by the Oil because of just idiocy in the dzone, this team would have won a lot more hockey games.

    Imagine if the NHL decides to crack down on clutch and grab hockey again. Oilers will be one hard team to stop.

  • Hemmercules

    This website has the most hilariously bi polar comment section on the internet. After 3 loses to the worst teams in the league it was blow up time, fire everyone, everyone sucks but Mcdavid. One win later and a positive article and its deep playoff runs on the horizon.

    I will be overjoyed if they make it but not surprised what-so-ever if they miss by a few points. Still too inconsistant.

    • What can you expect? After ten years of losing the team finally fires the management and replaces it with someone with experience, then does the same for the coach. Around the same time McDavid lands in the lap. You gave an eternally hopeful yet rightfully soured fan base the perfect storm. And then Connor got injured, Hall was traded, Lucic arrived, Puljujarvi got picked, it was a weird summer for Oiler fans who are used to finding ways to believe in players like Arcabello, Belov, Nikitin, Fayne, Ference, Scrivens, Petry, Schultz, all while pinning the entire year on high end draft picks like Hall, Nuge, Yak, Nurse, Draisaitl, and McDavid.

      Top all that off with tempered expectations at the beginning of the year where everyone was saying they will be better, but still not a playoff team; only for those expectations to be obliterated after a crazy hot start. Now that ten years of waiting has rushed forth like a burst damn, and very little is going to stop the tidal wave of expectations.

      Oh ya and a new arena. No wonder it’s high highs and low lows. Or should I say Lowe lows?

      • Hemmercules

        Not denying the passion. Its great. Winning is great too and Im enjoying the season so far. After the two Arizona games people were calling for a house cleaning on the bench bosses. Its crazy and funny.

        The Oilers are in a playoff spot but are they a playoff team?? I’m still not sure about that. Losing multiple games to bottom feeders would suggest they aren’t yet.

        I guess I just expect the worst and hope for best when it come to the Oilers. I get a bit worried on 5 game losing streaks, 3 not so much unless they got totally dominated.

        • Well, plenty of playoff teams lost to the Oilers over the last ten years. Sometimes, good teams lose to bad teams.

          Also Tippet has ran Edmonton’s show, for ever now. The man knows how to beat an inexperienced team that relies on skill. He bores them to do death until they make mistakes.

          As for the loss to Toronto, I mean, they also seem to have Edmonton’s number. I got lower bowl to a game about two years back only to watch a 5-0 shut out for Toronto. At least the last loss was an exciting game.

          Also people don’t seem to realize Toronto was like 8 and 2 in their last ten games and had just beaten Washington 4 -2 a few nights before. Is Edmonton better than Washington? Maybe not at the moment, but again, good teams lose to bad teams sometimes.

  • Glass

    If I had to play a GM 2017:

    Deadline:
    Trade Pouliot + Musil for a 3rd line RHC and a depth pick. Preferably a veteran, then move Caggulia to the left wing.

    Shop one of Oesterle/Simpson/Laleggia + Eetu Laurikainnen for a draft pick or RHD prospect. (Room needs to be made for Paigan, Jones, and Bear.)(We also have Broissoit/Ellis/Wells/Svobotka goalie depth.)

    Shop Hendricks, Gustavsson, and Gryba/Fayne for picks/prospects.

    (Call-up Broissoit or Ellis, again we have a couple more goalies in the pipeline, and if need be in the off-season you can sign an AHL back-up goaltender as a stop gap).

    (Fayne makes Gryba expendable, and 7th D are easily replaceable.)

    Off-season:
    Trade Eberle for a top 4 RHD who can run the PP.
    Sign veteran middle-6 RW. (Insurance in case Puljuarvi is not yet ready for top 6 by then.)

    2017-18:
    Lucic/McDavid/Draisaitl
    Slepyshev/Nuge/Puljujarvi
    Caggulia/RHC/Veteran RW
    Maroon/Letestu/Kassian/Pitlick

    Klef/Larsson
    Sekera/RHD PP QB
    Nurse/Benning
    7th D

    Talbot/Broissoit.

    • Gravis82

      so trade all of our mid range prospects and mid range players for picks, and then our top 2/3 scorer for a #4 dman?

      There goes depth and offence. I hope you like pheonix Coyotes hockey, because that’s where your proposal is taking us, in my opinion. Not trying to troll, but I don’t like this.

        • Gravis82

          we have depth on our NHL team. We need depth in the minors. People will be moving on (Hendricks), people get injured. Trading away what you propose compromises our ability deal with short term injuries in the next 1-2 years, for not too much more gain.

          I think you can find a 3rd line checking C in free agency.

          And I don’t think you can get a top 4 PP qb by trading who you are proposing.

  • Leaking5w-30

    Wills. The fenwick numbers for the past few years mean nothing to me in raw form. Could you post where the oil ranked those years? Im assuming pretty low?

      • FISTO Siltanen

        People who run this site better be careful. The longer this stews the more “Truthers” will come out of the woodwork with theories and ideas on what is really happening. By the time this explodes to that level, even telling the truth with video evidence and insider confessions will be too late Those points will be dismissed to further twist the conspiracy on longer.

        Just saying.

  • My concern is how much McDavid is a part of those positive numbers.

    This roster isn’t made up of a bunch of rec league players, so I know sooner or later they’ll play a more prominent role in future success, but I’d like to see that sooner than later.