Photo Credit: Sergei Belski-USA TODAY Sports

Good Corsi, B.A.A.D Team

Oilers fans are rightfully upset at their team’s performance thus far. Expectations were sky high coming into the season, and they haven’t resembled anything close to the Stanley Cup Contender many thought they were. Fans don’t want to hear about Corsi and how this team is better than it’s shown. They’ve seen how poorly Edmonton’s played, and the hole they’ve put themselves in. But they’re also sitting at the top of the league in terms of shot attempts, so how could they been this bad while also out attempting teams regularly?

Edmonton has some interesting company. 15 out of 19 teams generating over 53 percent of shot attempts over the past five seasons made the playoffs. 10 of those teams made it past the first round. Four of those teams made the Stanley Cup Finals, with three of them (Chicago x2, Los Angeles) winning the Cup. Outshooting and out-attempting the other team is generally a good thing.

Advertisement - Continue Reading Below

Carolina, Montreal, and San Jose are all over 53 CF% this season, too. The Canadiens are letting in even more goals 5-on-5 than Edmonton, a result of Carey Price being injured, and struggling because of it. The Sharks and Hurricanes, however, are winning and in range of the playoffs.

The Oilers don’t look like any of the teams that made the playoffs, but their shooting percentage and save percentage amongst the lowest here. That’ll change but by how much? Their 6.63 shooting percentage isn’t too far off of a few teams, so there’s some room for improvement there. Cam Talbot hasn’t been good, but he’s saved .916 of the shots faced since coming to Edmonton and has a career .920 save percentage. He’s currently at .901 on the season. Talbot’s going to play better, but I’m less sure their shooting is due for a big outburst.

NHL History: Petr Klima scores fastest hat trick in Edmonton Oilers playoff history in 1991

I’m using 5-on-5 time on ice because hockey games are mostly played 5-on-5. Oscar Klefbom and Darnell Nurse are two of Edmonton’s most common shooters, which includes four defencemen in the top ten. The McDavid line is mixed in between Klefbom/Nurse and Adam Larsson/Matt Benning. Last year, Klefbom was the only defenceman in the top five. Nurse, Larsson, and Sekera rounded out the nine, ten, and eleven spots then.

Advertisement - Continue Reading Below

It’s good to have your defence active and shooting, but having them so involved in your attack could explain the lack of scoring despite strong shot totals. Defenceman aren’t as good finishers as forwards, as they’re in less dangerous shooting areas. The loss of Jordan Eberle hurts. He was one of their leading shooters and better finishers, and they never replaced him.

Micah Blake McCurdy (@IneffectiveMath) has a wonderful site called HockeyViz.com that has heat maps for the most common areas teams take and allow shots. 

The Oilers are generating a lot of chances in close, but a large percentage of their shots are coming from the points. They might be trying to hammer it in front and score via rebounds and deflections, in fact the players have said as much, but they’re doing it more than most of the league. They’re one of the lowest scoring teams, so it might not be a strategy that’s working.

It’s still early, but 22 games means a quarter of the season has already passed. They’re sitting second in Corsi For %, but they were 56% percent in their first 11 games, and that went down to 51.88% in their next ten games. Instead of their shooting percentage getting much better, it’s possible they just lack the scoring talent outside of the McDavid line and Ryan Nugent-Hopkins and remain a lower finishing team.

Two Months Later...

Edmonton might be more 2014-15 and 2016-17 Los Angeles Kings than the dominant Corsi teams that made noise in the playoffs. The Kings were a big, heavy team that had impressive Corsi numbers, but were too slow and didn’t score enough following their Cup wins. The Kings acknowledged this and changed coaches, as well as personnel, in an attempt to get quicker and score more this season. The Oilers are trending this way, with a notable lack of speed outside of Connor McDavid, when the NHL is getting faster.

Stats from Corsica.hockey.

Advertisement - Continue Reading Below

  • Slipknot 8

    There’s a couple of problem with volume shooting, one is actually hitting the net, the Oilers are terrible at that, the other is the Oilers are slow, so a blocked shot or a missed net is almost always ending up on an odd man rush…..this is why the Oilers have so many odd man rushes against them.
    The other huge concern is shooting percentage from the defensmen, sporting a 2.5 SH% isn’t going to get you a lot of goals so the absolute key must be hitting the net so the forwards can pick up the puck. this isn’t going as planned.

  • ed from edmonton

    So what you are saying is that Corsi provide limited information and provides only a moderate correlation with wins/losses. I think most hockey observers have already come to this conclusion and track coring chances and grade A scoring chances as a better leading indicator of success.

  • ubermiguel

    Odd that 4 of the top 9 teams on that list missed the playoffs; if Corsi measures “good hockey” then I wouldn’t expect non-playoff teams to cluster near the top.

  • JimmyV1965

    Goaltending is killing us. Our defence has been awful, but Talbot had let in too many soft goals early in the game and late in periods. Even last night both goals were weak. Talbot would have had a shutout out last year. He’s a good goalie but he’s in a horrible funk right now.

      • OilersGM

        Talbot’s G.A.A of .919 last year and .900 this year resemble Dubnyk’s numbers almost exactly from 12-13 .920 G.A.A and 13-14 .894 of course traded later in the 2014 season. I think Dubnyk is a good comparison for Talbot. Now only if Talbot can string together 3 shutouts like Dubnyk recently did with Minny it would be a good sight for sore eyes.

  • TKB2677

    Here is a prime example of how corsi while a tool in a tool box shouldn’t be taken as being an overly good stat. The Oilers beat the hell out of the Wings last night. They won 6-2. While Talbot was good, it wasn’t like he has having to stand on his head and stole the game. They won 6-2 where all 6 goals were scored 5 on 5. The Wings score 2 goals, one of which was on the PP.

    The definition of corsi is as follows.

    Corsi is an advanced statistic used in the National Hockey League to measure shot attempt differential while at even strength play. This includes shots on goal, missed shots on goal, and blocked shot attempts towards the opposition’s net minus the same shot attempts directed at your own team’s net.

    Now hardcore corsi believer and blogger here – Matt Henderson tweeted the Oilers defense cosi last night.
    Auvitu – 66.7
    Benning – 64.5
    Russell – 48.3
    Klefbom – 48.3
    Larsson – 35.7
    Nurse – 35.7

    So the Oilers completely outplayed the Wings. They outshot the Wings. They whipped them by 4 goals. They destroyed them 5 on 5 scoring 6-1 in goals. According to Staples, the scoring chances were 13-4 in favor of the Oilers yet the majority of the Oilers defense was below 50% with one pairing being at a brutal 35.7.

    If the game was the opposite and the Wings beat the hell out of the Oilers, then I coule believe the corsi Henderson posted. But when the Oilers beat the hell out of the team and do it ALL on 5 on 5 and corsi is only a 5 on 5 stat, one of 2 things has to of happened. Either Henderson doesn’t know how to calculate the stat properly OR the stat is borderline BS. I am assuming that Henderson knows how to calculate the stat properly so I have to believe that corsi is a borderline useless stat. If you listen to the more mainstream guys like a Woodguy, he doesn’t talk about corsi much anymore. If you look at the definition of corsi, it’s a measure of how many shots are directed at your net. So if you as a dman do your job almost perfectly and not allow the opposition to get into any sort of actual scoring position, you can even barely allow that opposing forward over your blueline yet if the opposing forward manages to float a 60ft wrist shot towards your goal, that counts as a negative to your corsi. There is almost ZERO chance it ever goes in but it was a shot attempt directed towards your goal so you get dinged for it. So I kind of almost think its a joke that people still report it and use it as a way to bash players.

    • TKB2677

      Here is a great example I forgot to mention. In the Rangers – Canes game. A Rangers player had the puck, got barely over the center line and floated the puck in towards the goal. Now Raanta reached up to catch it, somehow screwed it up and it went in off him. BUT since that was a “shot attempt at your goal”, the Canes dmen would have got a negative corsi value on that play. The Ranger was probably firing it on for a line change but it was directed at the goal. So I am curious to know what should the Canes defender had of done differently to not get a negative corsi on that play considering the shot came from center ice.

      • serlio

        Well im willing to bet that that player also had his corsi boosted when their line dumped the puck in on net then changed. Mostly random events like that tend to even out over larger sample sizes it really shouldnt affect corsi over a sample size thats actually relevant.

    • Agreed. As a defender, making your opponent shoot a bad, weak shot is almost as good of a defensive play as a turnover. And if your the opponent who’s team takes Corsi seriously, you’ll “shoot for the sake of shooting” as Eberle put it after Eakins got canned.

      Nurse and Larsson did a good job last night against the Wings top line as while they maybe didn’t suppress shots, they made them take weak ones. And that is also when a goalie’s save percentage climbs. Not only that, but when the goalie saves it and the play is called dead, Oilers can change, and be more rested.

    • Gravis82

      I don’t think there’s ever been any person who has ever said that the only thing we should look at Is corsi. This is the of statement you’re making though, and then arguing points everyone on both sides agrees on to tear down a position that Does not exist .

      No traditionalist would ever say we should just look a Corsi. No statistician would ever say we should only look at one metric either, because if there’s anything you learn in statistics is that every measurement is imperfect, literally every single one. So I’m not sure what you’re trying to say ?

  • BringitbacklikeSlats

    Couldn’t agree more. Corsi as an individual players performance indicator is the biggest fallacy perpetuated by the whole of the advanced statistics community.

    I think it has value as an overall team stat though and as was seen earlier in the season when the Oil played well but still lost.
    Guys like Henderson that live and die by the false idol of Corsi and use it to determine who’s a “good” player and who isn’t are more irrelevant than anyone that writes about the game. They’re fervent conviction that it’s the most accurate evaluator of a players worth is ultimately leading people to believe that advanced analytics in its entirety holds no value. It’s too bad, because there are a lot of great numbers to be utilized.

    • Gravis82

      Corsi is better than rolling a dice, but is not right all the time. If you are trying to make decision…and you are unsure…and have looked at all other angles, scouting reports, videos…etc etc and are still unsure, what should you do? Roll the dice and pick randomly? Or go with corsi which is at least a bit better than random. Option 2. But there is a decision tree, on when it is appropriate to do that based on what information you’ve already considered that has not lead you to a decision

  • Jordan88

    Good teams have good corsi, having good corsi does not make you a good team.
    Good corsi is just part of being a good team. But bad teams can have good corsi as well. Its all part of a larger picture, this is why I am not a fan of advanced stats.

    • TKB2677

      I agree to a point with what you are saying. Good teams usually have decent corsi numbers but your corsi can also be affected by how you play as a team, especially defensively. If you as a team allow the opposition to take long range shots and shots from the side boards that have little to no chance of going in but don’t allow any good scoring chances, you as the defending team will probably win the game because your goalie has an easy night but your corsi will probably be lousy. So does it matter that much? Same goes the other way. You can fire a ton of pucks towards the opposition goal, not give up a lot of shots against 5 on 5 but if you only give up 5 legit scoring chances but they are all insanely good scoring chances that all go in and you lose 5-3, who cares if you win the corsi battle.

  • KenBone18

    Here’s a question most of you are probably not bright enough to ask – what is the historical correlation coefficient between a team’s Corsi (over an entire season) and where they finish in the standings?

  • puckle-head

    I think this is article is a great example of how advanced stats can supplement our understanding of the game. Corsi and fenwick are important stats because they are excellent at tracking who is controlling the majority of the play, but often they fail to breakdown why the shots are or aren’t going in, and instead assume that the percentages will average out over the long term. I find it interesting how reliant the Oilers have been on point shots, and that may suggest that their possession statistics are “gamed” for lack of a better word. It reminds me of a quote after Eatkins was fired when Jordan Eberle was asked about their improved powerplay and he said it was because they were no longer shooting for the sake of shooting.

    This team is clearly better than their record suggests, but having watched them struggle with scoring so much this year I can see them being one of those good possession teams that is golfing in June.

  • Anton CP

    Interesting heat map. Apparently that the Oilers love to shoot the glove side a lot.

    Anyway, Corsi is a poison stats. Anyone likes to apply Corsi to how team performed have some serious misconception of how it works. First of all, it doesn’t. Second, a good team normally will have some good Corsi stats because that they are dominating the opponents so they have a better Corsi as result. Analytics are mistaken as the other way around and believed better Corsi means better team. One of the example as I remembered about how ridiculous analytics effects coach’s decision was when Yakupov scored on his first ever shootout chance that he was never used due to 0 success rate…when he was never having one!

    Most of the analytics don’t even know how to work with numbers even though their damn job is all about numbers.

  • ScottV

    I would like to see possession measured with a stop watch, not by shots or shot attempts.
    Have someone pushing buttons when one team possesses vs the other in all three zones.
    Take actual possession and then run other numbers off of it.